Battery Knowledge Base

Item:OSWc83ef5798d8840e8863bc0bb67cd8d2c: Difference between revisions

Item Discussion

Item:OSWc83ef5798d8840e8863bc0bb67cd8d2c (view source)

Revision as of 17:23, 21 January 2025

36,290 bytes added ,  Yesterday at 17:23
add transcript and video
VisualWikitext
No edit summary
Tag: Metadata slot edit
 
(add transcript and video)
Tag: 2017 source edit
Line 1: Line 1:


== Recording ==
{{#ev:youtube|tRuSOmzkdTg|||Battery 2030+ Excellence Seminar, Exploiting Defects in the Design of Battery Materials, Montse}}
== Transcript ==
Thank you. Thank you so much, Robert, for this very kind introduction. And thanks also to Kristina and the Battery 2030+ community for inviting me today to share some of my research work.
So I'm just starting to share my screen. I think it's okay now.
=== Defects and Disorder in Battery Materials ===
I'm really happy today to have the opportunity to speak about the topic of defects and disorder in battery materials.
So this is a research topic that has been a long-term research effort in my group and throughout my career and which has been motivated by two main misconceptions or oversimplifications. The first one is that we believe or we always think of crystals as perfectly ordered arrangements of atoms, like the one that is shown here in this picture. And this is in fact how we are taught crystallography and the tools that we use in crystallography very often implicitly describe crystals as being perfect.
However, the reality is often very far from this. Crystals are never perfect, as they can exhibit some of the different types of defects that are indicated in this picture. They can exhibit different types of defects and even simultaneously different types of defects.
The second one, the second oversimplification is that defects are always detrimental. So, in some research fields, it's already clear that defects can bring value, for example, in the field of plastic semiconductors. In the field of batteries, I have the feeling that we always think of defects as detrimental and we should get rid of them.
But this is not necessarily the case and we will see today that sometimes they are detrimental, sometimes they are not. And either way, understanding defects matters. So, the realization of today's batteries has been sustained so far on structural design and on our ability to identify host structures for the intercalation of ions.
And now if we want to go further, we really need to complete the picture and really include defect structures in our understanding of thermodynamic and kinetic phenomena. So this is not an easy endeavour, but it's really an opportunity in our field and this has been understood for a very long time. We can see here an example of a review paper dealing with nickel hydroxides, which are the positive electromaterial used in nickel-metal hydride batteries, where it is very clearly said that the lack of perfection in crystallization can be taken as a sine-qua-non reason for electroactivity.
So empirically, it has been understood for now, more than 40 years, that disorder and defects are relevant. And there have been very good pioneering works from Claude Delmas or Yves Chabre at those batteries where they have put the focus into that. However, from understanding that this matters, to really have a full picture of which type of defects and how many and up to which point they are involved in the mechanism, there is a long way, and it is in fact challenging.
And understanding defect structures in battery materials is, as I was saying, not an easy endeavor, because we are dealing with many different materials, all of them with very diverse crystal structures and constituent elements. There are a different number of defects that we can encounter depending on the material. Very often they appear in low concentrations, or as I was saying before, they can even appear overlapping.
I mean, different types of defects in the same material. And the number of characterization methods has been traditionally limited, as is also their application. So we can see here a summary of different types of defects we can find in crystalline solids.
The two main categories that have been investigated in battery materials are highlighted in yellow, and they are mainly point defects or planar defects. And in bold are some of the defects we have been studying in my group. As you can see, a lot remains to be done.
There are defects for which we have not even examples of how they might impact, which doesn't mean that they are not present in our battery materials. Fortunately, there has been progress in recent years, and now there's a number of recent models and tools that are applied to scattering and spectroscopic techniques which allow studying defect structures. And these include, typically diffraction techniques, but also electron microscopy, solid-state NMR or Raman spectroscopy, to name a few.
You can see here a few examples related to the LNMO high-voltage spinel material, which I will describe in more detail in the following slides, where you can see that if we take two structures with different degree of ordering, we can indeed observe very different signals when we observe them using these techniques. There are still significant limitations. None of these techniques are perfect.
Sometimes it's because the information is indirect, as is the case of diffraction, because the structure and microstructure appear convoluted in the pattern. Others are too local and the length scale is too limited. Sometimes we don't have enough spatial resolution.
Some methods are destructive or light elements do not offer enough contrast. But I hope to convince you today that there's already a lot that we can do and learn about these materials.
=== Techniques for Characterizing Defects ===
So powder diffraction, in my opinion, remains one of the most relevant techniques to characterize defects in crystalline materials.
Because any disruption of the periodicity in a crystal structure will have an impact in the diffraction pattern in the form of peak broadening and its extent and its angular dependence will allow us to quantify the amount of disorder, but also to distinguish among different types of disorder. And this is typically done through the Rietveld refinement method, where we assume that we have an average 3D structure that is repeated infinitely throughout space, but where we can introduce in the structural model some types of defects like point defects, strains, anti-phase boundaries. And also I include here size, because in terms of diffraction, size or grain boundaries are also considered to be a disruption of periodicity and as such, also impact in broadening the diffraction peaks.
Okay, so perhaps the most well-known example in the battery community when it comes to materials and defects is LFP. LFP has been reported to exhibit anti-site defects depending on the synthesis conditions, because lithium and iron atoms exhibit a similar size, iron II. And therefore it is easy to obtain materials where we have a certain number of exchange positions between lithium and iron.
And these types of defects have been shown to have a two-fold impact. First of all, a negative impact, because the size of the channels will be blocked for lithium diffusion. And therefore, if we have, for example, as is shown here, two defects within the same channel, we will have a number of lithium ions that will be trapped.
And second, while this will block these channels, it will allow the crossover between channels, which will partially mitigate the detrimental impact of having these blocked channels. But because lithium crossover is lower than lithium diffusion in this plane, overall, still it will lead to a lower rate capability. And it has been shown by theoretical calculations, but also by experimental evidence that this can be circumvented if we are able to prepare a smaller particle, so that the impact of having these blocked channels is significantly decreased.
A long time ago, when I was at LRCS, as Robert was mentioning before, we also showed another impact of defects. In this case, we were working with LFP materials that were nanocrystallized and that exhibited a very large number of vacancies and also iron in lithium sites with a formula that is like the one shown here. And what we found was that by these changes in the microstructure, we were able to modulate the reaction mechanism.
And instead of having the typical two-phase reaction mechanism that is often obtained with LFP materials, and that leads to this very stable plateau at 3.5 volts versus lithium. With this material, the reaction mechanism changed and it operated through a full-solid solution mechanism. So this means that we no longer have two phases, but a single phase of average composition that is evolving homogeneously throughout the reaction.
And this is interesting because first, it may lead to improved rate capabilities but also it would allow to monitor the state of charge of the battery through the measure of the voltage. Another example is the lithium nickel manganese oxide I was mentioning before. So this material crystallizes in spinel structure and, while it's analogous to LMO, which is a commercial spinel as positive electromaterial, but in this case, we are replacing part of the manganese atoms by nickel atoms, which can lead to two different structural polytypes.
In one case we will have a structure where nickel and manganese atoms will be perfectly ordered and following a very specific pattern, while in the other case they will be totally disordered and randomly located throughout the lattice. This has important implications when it comes to the properties of the material, which are still a matter of debate. And this is because the crystal chemistry of these systems is a little bit complex.
So what has been shown throughout the years is that this ordered polymorph is obtained at around 700°. This is the temperature at which nickel and manganese atoms become ordered. And if we increase the synthesis temperature up to higher temperatures, the material will crystallize in the disordered structure.
But this will be accompanied by a manganese enrichment which is driven by the crystallization of the rock-salt phase that contains both nickel and manganese, but in a different proportion than in the initial spinel material. And this is detrimental because this rock-salt phase is insulating. Now, it is typically considered that the best performances are obtained with disordered samples.
And an example can be shown here. So this sample would be prepared at 500, 800 and 1000. At too high temperatures, the amount of rock-salt is too important, so the material is practically insulating.
At too low temperatures, the surface area-- So the materials are still disordered, but the surface area of the material is too high, and because it operates at a voltage that is beyond electrolyte decomposition, then we have a huge reversibility, while at moderate temperatures we have, as I was saying, the best performance. Now, why the disordered performs better than the ordered still remains controversial. And one of the questions that first needs to be answered is what does order and disorder exactly mean in this system?
So, in order to understand this, a while ago, in a collaboration with Jordi Cabana, we decided to look at a series of samples that had been prepared at temperatures close to the ordering temperature, so between 670 to 730°, but at different atmospheres, air or oxygen. And then we looked at them by means of neutron diffraction. In this case, neutron diffraction is much more adequate than X-ray diffraction, because the scattering length of nickel and manganese are much more different, so we can really distinguish them in our structural model.
And the type of pattern we obtained is something like this. So we have a lot of reflections. This indicates that there is this superstructure in the material.
But we can see that we have a number of very sharp reflections which are coincident with those of the spinel sub-cell and are the reflections we would obtain with a fully disordered material. Plus the superstructure reflections are broadened, and the broadening was varying among samples. And we found that in order to explain this observation, we needed to describe the material as a fully ordered material, but with antiphase domains.
This means that the material is ordered throughout the whole particle, but the order is disrupted at antiphase boundaries. So we will have a particle where the oxygen skeleton is fully coherent, fully ordered, but the nickel manganese order will be disrupted in the form of these antiphase domains, as shown here, whose size can be extracted from the broadening of these superstructure peaks. And so this is the typical sizes we obtained for the range of samples that we characterized.
And when we tried to correlate this with the electrochemical properties, first we saw that this size roughly correlates with the size of the voltage step of this material. And second, we also saw that there was a very clear correlation with the rate capability of the material. And in fact, the samples with antiphase boundaries exhibited poorer electrochemical performance.
And the smaller the domain, so the higher the surface at the boundary, the poorer the electrochemical properties, except for one sample, which is the green sample here, which exhibits a rate capability that approaches that of the disordered material. And we believe that this is because in this sample we found another type of defect, which are nickel-manganese antisites. So the material overall is ordered with these antiphase domains, but has more disorder than the other samples.
It can be described with a formula like this one here and this site mixing disorder really enhances the electrochemical performance, we believe, by improving lithium-ion pathways. So two types of disorder in this system, with opposite impact. And this means that while this material overall appears fully ordered in terms of diffraction data, a little bit of disorder seems to be enough to have a performance that really approaches the fully disordered one, which I think is really interesting.
Okay, so the two previous examples correspond to situations where we can describe defects with average structures. But there are many other defects, such as the stacking faults shown here, where this is not possible. In particular, when we have materials where the periodicity in one of the directions is not found.
So this is very common in layered materials such as NMCs or sodium layered oxides. These type of defects typically appear during crystal growth. It can also appear by deformation, but it's typically not the case in battery materials.
And this is important because it can influence phase transformations occurring through layer gliding, which is the reaction mechanism of many of these materials. And well, there are different types. And this includes also twinning, which is having a mirror plane somewhere in the structure, and then the growth continues following a different pattern that mirrors the one below.
So in order to deal with this type of cases, we developed a while ago the FAULTS refinement software in collaboration with Juan Rodriguez Carvajal from ILL, which allows to, instead of describing the structure as an average unit cell and space group, we can now slice down the structure in different atomic layers, which we can describe as it is more convenient for us, and then when we can stack one over the other, following different types of stacking vectors to which we assign different stacking probabilities. And this really gives a lot of versatility to our structural models, because we now do not impose having a periodical structure along the stacking direction. And you can see here the difference when we deal with diffraction data exhibiting this type of defect.
I will come back to this material later but you can see, these are some superstructure reflections these material exhibits, but you can see there's also a diffuse intensity which Rietveld refinement cannot capture. While with FAULTS refinement we can not only capture, but we can exploit to quantify the defects that are causing this diffuse intensity. So let me show you some examples of this.
The first one refers to these lithium- or sodium-rich layered oxides. So this would be the typical structure of a classical layered oxide such as NMC. We have alternating layers of transition metal atoms and lithiums or sodiums.
But in this lithium- or sodium-rich materials, this extra lithium will be located in the transition metal planes. And because of the difference of size between the transition metal and the lithiums or the sodiums, there will be an in-plane order of the two. So this will follow a very specific pattern, but can still be described in terms of a layered structure.
And the fact that we have extra lithium or sodium unlocks anionic redox activity, which leads to a much higher capacity in these systems. Now, all these type of materials typically exhibit very large amounts of stacking faults. So instead of having a perfect stacking of these planes, what we find in reality is that there is in-plane ordering, but along the stacking direction, these planes do not follow a specific pattern.
And we can find structures with really very large amount of defects, which leads to this type of patterns I was showing here. So this peaks here represent the superstructure peaks of this in-plane ordering, but then the diffuse intensity refers to this lack of periodicity. So a while ago we were well aware that these materials exhibited this type of defects, but it was not very clear to us whether this was important or not.
And how could this be controlled from the synthesis parameters. So we decided to undergo a systematic study by taking Li2MNO3 model system that we prepared using three different precursors for manganese, two types of manganese oxide and a carbonate. And then we also annealed the materials at different temperatures.
So these are the X-ray diffraction patterns that we obtained. You can see, if we look at this area here, where superstructure reflections appear, you can see the samples look very different and there is more or less diffuse intensity, depending mostly on the precursor that we are using. So this is already telling us that we have obtained samples with a great variability of microstructures that we can now characterize with the FAULTS program.
So here is an example of refinement. In this case, this is one of the most defective materials. We found that the material contains about 40% of the stacking disorder.
And we did the same for all the samples. And here you can see some correlations. If we look at the size of the crystallites that we extracted from diffraction data versus the synthesis temperature, we can see that indeed temperature does have a strong impact in crystallite size.
And as expected, the higher the temperature, the larger the size of the crystallites. Although the sample prepared from manganese MNO had a much slower growth compared to the other two precursors. On the other hand, we found that temperature does not have a major impact in stacking faults, in the amount of stacking faults.
This was somewhat unexpected, but the precursor does. So the carbonate leads to the highest amounts of stacking faults, while MNO to the lowest. And this is the only sample where we found that increasing temperatures allowed decreasing stacking.
And then the morphologies that we obtained were also very different. But we believe that all these differences when it comes to precursors is due to different decomposition by temperature and mechanism of these materials. But interestingly, we have now a set of samples that exhibit the same crystallite size and very different amounts of stacking faults.
Those prepared at 800º or the other way around. If we look at samples prepared with the same precursor, the amount of stacking faults is pretty constant, while the size, when we increase the temperature, changes quite a lot. So it was a nice set of samples to extract correlations.
And here are the results when it comes to electrochemical properties. This slide is a little bit packed, but I walk you through step by step. So when it comes to size, we compare here samples that are prepared with the same precursor at different temperatures, so they exhibit different size, as indicated here.
And we found that very large particles lead to a very slow activation and therefore the capacity slowly increases, cycle after cycle. Too small leads to a very fast activation, but then capacity fades quickly as well. While intermediate sizes are those with the best compromise, the activation is not as slow and then the capacity retention is much better.
So size has a strong impact in capacity. Next, we fix the temperature. We look at different precursors, so we are fixing the size, but we are looking at samples with more or less defects and different morphology.
And in this case, what we find is that morphology is the main feature that has an impact on capacity. And best performance is obtained with isotropic defective samples, while those that have more platelet size, irrespective of the amount of defects, exhibit much lower capacity. And finally, if we look at the evolution of the derivative curves with the number of cycles here, for Cycles 10, 50 and 100, we will be able to see that stacking faults also have an impact, and specifically, when we look at the voltage fading of these systems.
So these systems are known to be prone to voltage fading because of the structural transformations occurring throughout site cycling. These are very visible in this part of the dQ/dV. And we can see that when the amount of stacking faults is small, after 100 cycles, we can already have a significant voltage fading of about 0.21 volts, while as we increase the amount of stacking faults, the voltage is much more stable.
And we believe this is because the local environments are stabilized by stacking faults. And this prevents, or not prevents, but delays the spinel formation in these systems. And since we had a nice set of samples that had been characterized with the FAULTS program, and for which we had quantified stacking faults, we decided to partner here with Dany Carlier from ICMCB in Bordeaux to try to explore whether stacking faults can also be quantified with solid-state NMR.
So we characterized the same samples with lithium solid-state NMR. Then we used the FP calculations to assign each of the peaks we found in the solid-state NMR spectra to specific environments in these defective materials. And then we quantified the number of environments from the solid-state NMR that corresponded to what we call modified lithium environment.
This means that are environments that appear because of stacking faults. Now, then we compared the amount we obtained from solid-state NMR and the amount we obtained from XRD. And as we expected, always what we obtain from NMR is higher because in fact, every time we have a stacking fault, it will affect also the layer above and the layer below, the lithiums of the adjacent layers.
And in principle, if the stacking fault is isolated, the number of modified lithium environment should be three times the amount of stacking faults we measured with XRD. But as you can see, this is not really the case in any of the samples that we measured. And this is because this will also be dependent on the fact, I mean on whether these stacking faults are isolated or they are closer or even consecutive.
And for example, for two consecutive stacking faults we wouldn't expect three times, but only two times the number of modified environments. And therefore this is very interesting because if we now compare the quantitative information we obtain from diffraction with that that we obtained from solid-state NMR, we will be able to have an idea of how distributed these stacking faults are. So we cannot directly quantify with solid-state NMR, but the ratio can indeed provide very complementary information.
This is another example of a sodium-rich material with the same type of structure as the previous example. It's sodium-rich ruthenium oxide. This sample was prepared by the group of Atsuo Yamada in Japan.
And they found that this material also crystallized with very large amounts of stacking faults, and therefore we collaborated in the characterization of this system. Again, we can identify them by this diffusing density which is also observed with the electron diffraction pattern of certain planes. And faults, again, we found that the amount of disorder is quite high, about 40%.
This is almost random stacking. But interestingly, when we look at the evolution of the XRD patterns of this system and we refine them also using the FAULTS program, we see a very interesting evolution. First we see that there are-- the material undergoes two successive two-phase transformations, because there are two intermediate phases that are formed as we remove sodium from the structure.
And this is a result of the arrangement of sodium-ion vacancies plus the layer gliding which is driven by columbic forces. This is specific of this system, but it happens also in other systems. But what is really particular in this system is that we found that when we start with a material with about 40% of stacking faults and we start removing sodium from the structure, the amount of stacking faults progressively diminishes until we end up with almost no stacking faults at the end of the charge.
And then the process is fully reverted as we discharge the material back. So this means that this material self-reorganizes upon charge and discharge. And this really allows reaching a much higher capacity than if this didn't happen.
And to understand why this is occurring, we need to look at the local environments in this system. So this would be our starting material. Here would be the ideal environments.
Here would be a typical stacking fault. And in this case, the local sodium environments in the sodiums in the interlayer spacing are very similar. The two systems would be energetically very, very similar, which is why these materials crystallize with these very large amounts of stacking folds.
However, as we start removing sodium from the structure and layers start gliding following these two vectors indicated here, we will end up with a structure that has less sodium in both the mixed layers but also in the pure sodium layers, with this ordering that results from these arrangements. But in this case, if we compare ideal and faulted environments, we will see that in some faulted environments, there will be a much higher repulsion between layers than in the ideal one. And this is even more accused when we look at fully desodiated material.
It's not fully, but at the highest desodiation degree, where we can see that all ideal environments will be very stable. So sodiums will not face any transition metal in the layers above and below, contrary to faulted one. And this is what drives this self-ordering of stacking folds.
But having the possibility of reordering is also what allows this material to reach this very high level of desodiation and also very reversible. So this is a very interesting positive feature of defects in this system. Another type of materials where we have been looking at defects are these intermetallic compounds that are used as negative electrodes in nickel metal hydride batteries.
They are also used for hydrogen storage. And these systems are typically described with two types of fundamental layers, AB5 and AB2B4, whose structure is shown here. So A2B4 has also a specular version, while AB5 only exists in one form.
And then by combining them, we can have very different types of-- well, very similar types of arrangement, yet different, which will correspond to different compositions. And in fact very small changes in stoichiometry. For example, from AB3 to A2B7, the B content changes only from 75 atomic percent to almost 78, leads to very different structures and each of them can exist in two different polytypes.
And as in all materials, the type of layer will play an important role in the hydrogenation and electrochemical properties. So we can see here an example of how the local structure of one of these materials would look like. It nicely corresponds to this A2B7, 3R type.
But in fact, because these materials are so similar, it is very easy to find intergrowths, defects and all types of configurations in these systems. And we enlarge our view in these different systems, we will see here we have intergrowths of the two polytypes. Here we still have intergrowths plus additional types of defects.
Also for this type of materials, you see that the regions, ideal regions are really very, very small. And of course this is also observed in the diffraction pattern, as we can see the peaks that correspond to each of the polytypes. But also we can see that some of the peaks are really very broadened because of this lack of periodicity along one of the directions.
So we characterized all these samples to quantify the amount of defects. We found where we were able to extract the amount of each polymer and also by using specific arrangements, we were able to determine how extended these ideal structures were. And also we could include this type of defects.
And we found that in fact, the thing that really correlated the most with the hydrogen storage ability was the different volume of the difference of units. So this really allowed us to find a design parameter to use in order to make more performing materials. And I think this will be my last example.
So this is another type of material that is used in primary batteries, in alkaline primary batteries. It is the electrochemical manganese dioxide, which also typically exhibits a diffraction pattern that exhibits a lot of broadening of certain diffraction peaks. And it had already been proposed in the 90s by Yves Chabre, and also by previous researchers before that this could be the result of the intergrowth of two types of structures, ramsdellite and pyrolusite, which differ in the size of these channels here.
And therefore, by means of the FAULTS program, we can describe an intergrowth structure between these two polymers. We can quantify how much of these ramsdellite and pyrolusite motifs we have in the material, and it nicely fits our experimental data. And these were just a few examples.
We have been looking at other types of materials where these type of defects appear. We are now revisiting graphite because graphite is also a material that can crystallize in two different polytypes. It has already been proposed that apart from intergrowth having turbostratic domains, which means that it's not even a stacking fault, but somehow a random stacking of the layers has a negative impact.
We are now trying to quantify all these and extract quantitative correlations. We have also been looking at calcium insertion compounds with Rosa Palacín, potassium with Romain Berthelot. I just showed here the example of MNO2, whose mechanism is proton intercalation.
So any type of intercalation type of battery chemistry can find this type of defects and also has to be looked case by case in order to see whether there is an impact or not. And with this I reach almost the end of my talk. So I hope I've been able to show you that defects are universal in battery materials and the impact is still poorly understood, but there is an impact in most of the cases.
We have now more tools to characterize this, and in fact when we use them in complement, the type of information we can obtain is much more interesting and can allow to correlate very nicely with what we observe in the electrochemical properties. We still need better techniques and methods, as many of them still have important limitations, but there's still progress on that and doing systematic studies, I think in this case is key whenever possible, because it is, for me, the only way to really find one-to-one correlations with electrochemical properties of function in general. I've shown you an example of Operando studies which allows understanding dynamic evolution of defects.
These are still very rare, and this is mostly because it is already difficult to characterize one state of a sample. To look at the dynamic behavior takes much more effort. But I really think this is one of the paths forward within this topic.
And let me just say a couple of words regarding this. So indeed, characterizing Operando experiments, and in particular when it comes to diffraction, it takes time, it is a significant effort, because we are generating a very large amount of data and each pattern has to be characterized carefully. But in order to mitigate this, we have been developing what we call the FullProfApp, which is a refinement engine that still uses Fulpro, which is one of those openly available, but which we have modified so that it can deal with very large amounts of patterns simultaneously.
It really simplifies the process of characterizing Operando experiments. It can also be used, for example, in high throughput schemes. This has been done within the context of Battery 2030+ and thanks to the BIG-MAP project.
And it is available for anyone interested in the BIG-MAP app store or in the ILL or CIC website. I really invite you to try and test it if you are dealing with this type of experiments. The interface is really nice and it really can deliver a number of results in a very, very short time.
And second, there's a growing number of works that call for caution when running Operando experiments because of beam inhibition effect. We have observed in many of our experiments that sometimes we do not observe the behavior that we would be expecting. It has been also the case in other research groups.
And when discussing this with Rosa Palacín, we decided to join forces and look at this in a more systematic way, which is the results that have been recently published in this paper, where we took NMC and LFP, two very well-known materials for which we know what to expect in an Operando experiment. And we measured them using diffraction, but also X-ray absorption at the synchrotron and monitored the evolution of their structure. And what we found was that depending on the experimental conditions that we use, we obtain very different results, from total lack of activity in some cases to the expected behavior.
We used different loadings, we used inhibitors to control the dose, we used different protocols also to give time to relapse. So we have explored the impact of all of these things into the results. And what we conclude from this study is that indeed the beam inhibition effects are correlated to cumulative dose, so dose per cycle, and we have to go to very low doses to really have safe values.
It is material-dependent and also depends on the electrode and electrolyte properties, such as the loading, but also the voltage profile. So here we can see in the case of NMC, when we irradiate the material too much, we will only see a change at the end of charge, when there is an important change in the voltage value. It is reversible But it takes a while to go back to the expected composition and structure.
And we believe it's likely caused by photoionization and secondary electrons affecting the interface. Maybe it's the result of the catalytic degradation of the electrolyte. We still don't understand everything.
There are, I think, many groups looking at that and I think this is something very important we need to figure out. So Operando studies allow understanding the dynamic behaviour of defects and of structures, but careful with beam inhibition.
=== Conclusions ===
And finally, as a main takeoff message, defect engineering offers a unique design opportunity to make better battery materials.
There's a lot to be done and understood, but here are a few examples where this has been shown. And I would like to finish by thanking all collaborators throughout the years on this topic, especially also my colleagues Jon Serrano and Marine Reynaud, who have been working with faults closely with me, and of course all the funding agencies, and you for your attention. I'll be happy to take any questions you might have.
Thank you. Thank you a lot, Montse. Really great educational.
Simon Clark
895

edits

Retrieved from "https://battery.knowledge-graph.eu/wiki/Special:MobileDiff/3837"