Battery safety has very important application value in transportation and modern travel, especially in energy security, and it is also a point of global concern.

Safety is a key concern of electric vehicles now, and there are various reasons for safety accidents. Once thermal runaway is induced in one battery, it spreads through the entire battery system, resulting in an accident.

The battery safety laboratory has a series of test methods, among which the more distinctive one is to use ARC to conduct adiabatic thermal runaway experiments on thermal runaway.

After a lot of experimental research, engineers have summed up three characteristic temperatures of battery thermal runaway, self-generated heat starting temperature T1, thermal runaway induced temperature T2, thermal runaway maximum temperature T3, we have also done many types of power lithium battery tests, conform to this rule. Among them, T2 is the most critical. Everyone knows what reaction T1 is. Generally, it starts from the SEI film. T3 depends on the entire reaction enthalpy. T2 is not very clear, but it is also the most critical. Heat will suddenly cause a sharp rise in heat, and the heating rate can reach 1000 degrees per second or more, which is the key to causing thermal runaway. Therefore, through the exploration of T2, we found that there are three important reasons. The first one is relatively clear to everyone is the internal short circuit, which is ultimately related to the diaphragm, which is the internal short circuit of the positive and negative electrodes. There are also our newly explained positive electrode material oxygen release and negative electrode lithium evolution, which can be summed up as positive oxygen release, negative electrode lithium evolution, and diaphragm collapse. These three reasons are the main causes of thermal runaway and the formation of T2.

Below, I will introduce to you our progress in mechanism and thermal runaway control for the three mechanisms mentioned above, including, first, internal short circuit and our method of controlling internal short circuit, which is BMS. Second, the thermal runaway induced by the oxygen release of the cathode and the thermal design of the battery. Third, the thermal runaway caused by the violent reaction between the lithium deposition of the negative electrode and the electrolyte and our charging control. If these three mechanisms and three technologies cannot solve the problem of thermal runaway, we still have the last resort, which is to suppress the heat spread. We must understand the law of heat spread, and at the same time restrain the heat spread, and ultimately prevent the occurrence of safety accidents.

Below I will introduce these four aspects to you:

First, internal short circuit and BMS. It is clear that mechanical reasons, such as collision, mechanical methods, and finally tearing of the diaphragm, or electrical reasons, overcharging, lithium dendrites, dendrites piercing the diaphragm, or overheating, of course, will eventually arrive. Overheating and overheating will lead to the collapse of the diaphragm. All the reasons are related to the internal short circuit, but the degree of the internal short circuit is not the same and the evolution process is not the same, but in the end, the diaphragm will collapse and the diaphragm will melt. Therefore, we combined the heating calorimeter and DSC. One is to explain its mechanism from the heat release of the material, and the other is to conduct a thermal runaway experiment on the entire single cell from the heating calorimeter, and put the thermal runaway experiment with the material. The thermal characteristics are analyzed together, which is the mechanism of our conventional thermal runaway after overheating. We can see that the melting of the diaphragm will lead to an internal short circuit. When the temperature rises, T2 will be formed when the diaphragm collapses, which directly causes thermal runaway. This is a relatively common reason. We also use many other auxiliary methods, including various material analysis methods, as well as thermogravimetric and mass spectrometry methods, to analyze various substances.

This is our set of basic analysis methods, which can analyze various batteries and various mechanisms.

This is the first and most familiar method of thermal runaway. In any case, we can certainly do a lot of work from the design point of view. For example, the diaphragm should not be too thin, the strength should be enough, etc., but the middle There is a common problem is internal short circuit, so we must prevent internal short circuit, we have to study internal short circuit, the experiment of internal short circuit is relatively complicated, there is no mature normative method, so we invented a new method , is to implant the battery with memory alloy and heat it to a certain temperature, so that the sharp corners of the memory alloy are lifted, triggering thermal runaway.

From the literature and our own research, we found that there are four types of important internal shorts, some can cause thermal runaway immediately, but some are slowly evolving, some may not be dangerous, but some are evolving It will be very dangerous afterwards. There are also some internal short circuits that are always gradual, and some internal short circuits that change from gradual to sudden change, and there are various types. To this end, we have also carried out some simulation analysis, which I will not introduce in detail here.

In short, we finally found that the evolution law of the evolutionary internal short circuit is the voltage drop, and the first process is the voltage drop. It is not until the second part that there will be a temperature rise, which will eventually lead to thermal runaway. So about this kind of gradual change, we should detect it for fault diagnosis in its first process, that is, the voltage drop stage, and pick it out to prevent its further deterioration. This is our internal short circuit detection. This is an algorithm for series-connected battery packs, including the first analysis from the consistency of the voltage. The voltage of a certain battery drops, indicating that this battery may have an internal short circuit. But if we can't confirm it, we will add the temperature. If there is a sudden change after the evolution, we will add the sensor of the combustible gas, so that we can have a solution for the slow change and the sudden change.

For example, for example, the consistency identification of the voltage of the battery pack in series, I will not introduce the specific algorithm, you can clearly see that the cells that drop under the voltage can be clearly seen.

Of course, we need to carry out a series of engineering methods. It is not enough to have a simple algorithm. We also need to add a lot of engineering related experience to make judgments. This requires a database. In short, we can warn from this aspect that for mutants, such as a micro-short circuit, due to fast charging, because the battery will be deformed and strained during the charging and discharging process, it will lead to the sudden deterioration of the micro-short circuit, such as human blood vessels. The plaque inside will suddenly thrombus when it is pressed. If we only use voltage and temperature, it is not enough, because it is too slow, it cannot be seen, and by the time you can see it, the heat is out of control. what to do? We are going to use this gas sensor, which can provide thermal runaway warning at least 3 minutes in advance. In conclusion, we develop a new generation of battery management systems with safety as the core based on these algorithms.

The second part is the second mechanism we just mentioned. Is it only an internal short circuit that can cause thermal runaway? Is there no thermal runaway without an internal short circuit? In fact, there is no thermal runaway without an internal short circuit. With the continuous enhancement of the separator and the continuous increase of the nickel content of the positive ternary material, its oxygen release temperature continues to drop, that is, the thermal stability of the positive electrode material is getting worse and worse, but our separator will become better and better, so weak Instead, the link will slowly become the positive electrode material.

This is the experiment we did. There is still thermal runaway without the internal short circuit. We remove the electrolyte and still have thermal runaway, and it can be seen from the middle that there is an exothermic peak. This is the combination of the positive and negative electrodes. After charging Putting the positive and negative powders together, there will be a violent exothermic peak, which is the reason for him. Specifically, where does the exothermic peak come from? Phase change of cathode material, oxygen release. Look at the peak of oxygen release. When the positive electrode and the negative electrode are combined, the negative electrode is oxidized. If there is a peak when they are not combined together, they disappear when they are combined together, which proves that the heat generation comes from the violent heat release of the reaction between the positive electrode and the negative electrode. So what is this mechanism? It is the exchange of materials between the positive and negative electrodes, that is, the oxygen released from the positive electrode runs to the negative electrode, forming a violent reaction, which causes thermal runaway. Regarding thermal runaway without internal short circuit, we can build a model based on all the side reactions just now. Through DSC multi-rate scanning, the reaction constants of all the side reactions just now can be calculated by this method. Of course, through a certain method, finally Combining the conservation of energy and conservation of mass, the complete process of thermal runaway just now can be calculated, and it can be in good agreement with the experiment. In this way, we can develop from the relevant experience trial and error to the model-based design. Of course, we must have many databases. It is impossible to have no databases. This is the relationship between the reaction enthalpy of various materials and the exothermic power of the reaction.

Based on the database, we certainly have to improve the materials. I think the key improvements are two, one is the improvement of the cathode material, and the other is the electrolyte. First of all, we can increase the temperature of oxygen release by 100 degrees from polycrystalline to single crystal, and it can be seen that the characteristics of thermal runaway have also changed. For example, we use high-concentration electrolytes, which is also a method. Of course, we are talking more about solid-state electrolytes. Solid-state electrolytes are very complicated. We think that concentrated electrolytes have good characteristics themselves. For example, its thermal weight has dropped, and its exothermic power has dropped. From this, we can clearly see that the positive electrode does not react with the electrolyte, because our new electrolyte uses DMC, and DMC is at 100 degrees. It has been evaporated. This is what we think the next step for electrolytes is not only solid electrolytes, but also additives from electrolytes, high-concentration electrolytes, and new electrolytes.

The third part is about lithium precipitation and charge control. As everyone knows, I talked about lithium-ion batteries earlier. The batteries will decay after a period of use. What will the safety of the whole life cycle look like? We found that the most important factor affecting the safety of the whole life cycle is lithium precipitation. If there is no lithium precipitation attenuation, the safety of the battery will not deteriorate. The only reason for its deterioration is lithium precipitation. We can find a series of evidences, such as low-temperature fast charging. After low-temperature fast charging, the temperature of T2 gradually decreases, and thermal runaway occurs earlier. This is the attenuation of battery capacity, from 100% to 80%. Clearly corresponds to the formation of lithium deposition from low-temperature charging of new to old batteries. The other is fast charging. After fast charging, it can be seen that the temperature of T2 drops, and T2 drops to 100 degrees. From 200 degrees to more than 100 degrees for the new battery at the beginning, thermal runaway occurs earlier and faster. What is the reason for this? The same is the lithium precipitation, we can see that the lithium precipitation is more and the lithium precipitation is less obviously different. Lithium precipitation has a large heat release, so it is still lithium precipitation. Lithium precipitation will directly react violently with the electrolyte, causing a large temperature rise, which can directly induce thermal runaway. So we must study lithium precipitation, just like we study internal short circuit, how to study lithium precipitation? First of all, we can see the process of this lithium precipitation. This is charging. After charging and standing, it can be seen that the lithium precipitation has just started to come out, and a large part of it has returned back. This is the process of lithium precipitation. The experiment just now can be seen from the red line, which is active lithium, reversible lithium. Another part is dead lithium. Reversible lithium can be re-inserted, and the overpotential of the negative electrode changes. After the over-electricity rises to 0 in the stationary phase, re-insertion of reversible lithium is performed. Of course, dead lithium cannot be re-inserted. This gives us a hint whether we can detect the amount of lithium precipitation through the process of reversible lithium. For example, it goes back to the process. This process corresponds to a voltage platform. We simulated and found this platform. For example, when we charge at a very low rate, there is no such phenomenon. It is a normal voltage depolarization, and there is no such platform. Therefore, this platform is a good signal. We can determine the end point of the platform through differentiation. This is the end point of the end of the platform, which represents the amount of lithium precipitation and has a relationship with our total amount of lithium precipitation, which can be predicted by formula.

We also found from experiments that this is a process of charging and standing. We see again that the lithium precipitation can be seen from the middle, which is the result of the experiment. So in this way, we can find it after charging, but this is a result after charging, can we prevent it from precipitating lithium during the charging process? To be able to eliminate lithium precipitation as much as possible during charging, of course, this requires the help of our model.

This is a simplified p2D model we made. We can see the potential of the negative electrode. We just said that the negative electrode potential is related to lithium precipitation. As long as the overpotential of the negative electrode is controlled, we can ensure that no lithium is precipitated. Through this model, the charging curve without lithium precipitation can be deduced. We make the negative electrode potential not lower than zero all the time, and the optimal charging curve without lithium precipitation can be obtained. We can use three electrodes to calibrate this curve to do our charging algorithm. We have cooperated with the company. It can be clearly seen that using this algorithm can completely achieve no lithium precipitation, but this is a calibration process. The attenuation performance of the extended battery will change. What should we do? We need feedback, so we have made a feedback control algorithm without lithium precipitation, that is, we must have an observer to observe the overpotential of the negative electrode. This is the negative electrode observation. The overpotential is the observer, which is actually a mathematical model. This is very similar to our SOC estimation. We have an observer algorithm, and we have a terminal voltage feedback, so that we can perform real-time control of lithium-free charging. We also cooperate with the company in this regard.

In this process, we still have some regrets. Can we directly use the negative electrode overpotential sensor? So our further research is to develop this overpotential sensor. Everyone knows that the traditional three-electrode I mentioned earlier has a limited life and cannot be used as a sensor. Recently, our battery safety laboratory has cooperated with the Department of Chemical Engineering. Zhang Qiang's team from the Department of Chemical Engineering, because they are a team with very relevant experience in lithium anodes, have made breakthroughs in this area. Our test life can be longer than 5 months, and it should be said that it can be used if it is longer than 5 months, because we actually When using it, it is only used for access testing during fast charging, not all the time, 5 months is enough. The next step of our work is the feedback charging control based on the negative electrode overpotential sensor.

The fourth part, thermal runaway, if our previous three methods fail, it is the spread of thermal runaway and our suppression methods. As we all know, this kind of mechanical abuse directly pierces or squeezes the battery will immediately form a combustion explosion. This is the process of spreading. This is the spread test we conducted. The first is the test of the temperature field. This is the spreading process of our parallel battery pack. The mechanism of the spreading process is on this. Why does it come down section by section? It all goes this way, so it causes the voltage to drop, but once at the end it goes off, it goes back again, which is characteristic of parallel thermal runaway.

This is a series of battery packs, which are purely caused by a heat transfer process.

This is another situation. It spreads in an orderly manner at the beginning, and finally spreads violently. Of course, it is caused by combustion in the middle, not only heat transfer. This will immediately lead to explosion accidents, combustion accidents, and so on.

This is the process of the entire system and the entire pACK propagation. Its propagation is regular. From D2 to U2, D1 is almost at the same time, and then the others. Basically, there is no way to come here. Because of the heat insulation in the middle, this prompts Our battery pack design is still very important.

According to this, our purpose is of course the design based on model simulation, because the process is very complex, and if it is very difficult to rely on relevant experience alone, this is the simulation we do. You must know that how to select the parameters of the simulation is the most important. You can adjust the parameters, but it is meaningless to adjust the parameters, so we have done a detailed study on the parameters. How to select the parameters is a very skillful process. I It will not be described in detail here, but a series of methods are required.

With this parameter-calibrated model, we can design, which is the design of thermal insulation. It is not enough to heat the battery only, there is also a heat dissipation design. There are also some batteries that can be insulated and dissipated at the same time. This is the firewall technology developed by our student-founded company. Heat insulation and heat dissipation are one piece, which is blocked by heat insulation and heat transfer, and heat dissipation quickly takes away the energy. These two a match. This is a lot of experiments. This is the experiment of the whole battery pack in the field, a traditional battery pack, a battery pack with a firewall. The battery pack with the firewall had a lot of smoke at the beginning, but it gradually disappeared, and there was no burning and no heat spread. The traditional battery pack formed heat spread and burning at the end. We can pass this and really block it. This is related to this work, and we have also participated in the formulation of a series of international regulations.

Now, there is a process in the middle of what we are doing further, which is the eruption, which is more complicated. Now we have not added it to the simulation. Of course, the eruption model has, but it is not accurate. It can be seen from the experiment that there are three states of solid, liquid and gas. The intermediate gas is some combustible gas, which is the fuel, and the solid is some solid particles, which often form a flame. what to do? One is to collect particulate matter, just like in traditional cars, by passing the particulate matter through a filter to capture it. The other is dilution, getting the flammable gas out of its flammable range, and that's what we're doing right now.

Finally, let me make a summary.

There are three processes of thermal runaway, from induction, occurrence to spread. As far as induction is concerned, there are various reasons for induction. I have already talked about it a lot. Of course, there is also the mechanical part of our collision, which I did not say. Now we are talking about these things at the core, and these things are not yet available. There are no regulations to regulate, we think the latter is necessary. Second, thermal runaway occurs. We mentioned three temperatures, of which the T2 temperature is shown here for three reasons. There are eruptions and fires inside the battery, which are mainly determined by the state of the electrolyte and the boiling point of the electrolyte. There is an eruption, a second eruption, and finally a fire. If we want to prevent it, we must remove all these links. Here are some measures . Finally, there is the spread, there is the spread that we can expect, and there is the sudden spread, such as the spitting fire, which is the eruption to the spitting fire to the violent spitting, and finally to the violent burning, and all the problems we show here have solutions. of.