1. Introduction
Lithium-ion (Li-ion) cells are part of the modern technical and industrial world. They are widely used to store electrical energy for various mobile and stationary applications. The increase in demand has led to a rise in the production of Li-ion cells, their second use, and recycling. This recycling step after the second life helps to obtain a sustainable battery life cycle. Over-discharging a Lithium-ion battery (LIB) is usually necessary before initiating the actual recycling process. It can be carried out in different ways, such as the electrical discharging of the battery with a resistor or insertion into a conductive liquid [
1].
This work focuses on the deactivation of LIB in different liquids (aqueous calcium chloride (CaCl
2 solution), Tap Water (TW), and Demineralized water (DW)). This will help in the disposal of cells, modules, or the whole vehicle traction system, including deactivation in an extinguishing container where the cells are in critical condition [
2]. Some of the major objectives of this research include looking into the deactivation process, i.e., the release of internal residual energy of a Li-ion cell, based on a variety of deactivation fluids, the deactivation process of CaCl
2 solution in relation to TW and DW by considering the comparison of multiple different parameters such as conductivity, lithium concentration, pH value, fluoride concentration of the liquid, and cell Open Circuit Voltage (OCV).
From the literature, authors have identified that there are several different procedures for cell deactivation based on a variety of material compositions and concentrations of liquids [
3,
4,
5], which makes it quite difficult to compare and contrast these procedures. An example of such research is the deactivation of a single cell by Lu et al. [
3] and Li et al. [
4] in the years 2013 and 2016, respectively. Both the researchers used aqueous sodium chloride (NaCl) solutions in varying concentrations; however, the outcome obtained by Lu et al. [
3] was a rapid breakdown of the OCV of the cell within seven minutes after the cell was placed into a 10% NaCl solution, which indicates a fast discharge, while Li et al. [
4], had a comparatively slower discharge, including the corrosion of cells phenomena. In the same research, further analysis of the metal ion concentrations of the liquid was also undertaken. Due to high concentration levels of several metals such as Sodium (Na
+), Aluminium (Al
3+), and Iron (Fe
2+), it was concluded that all of these metal ions can be traced back to the casing of the 18,650 cylindrical cells, which means that the cell casing reacts with the liquid. Additional observation depicts a high content of phosphorous in the fluid, which, according to the research, results from the conductive salt Lithium Hexafluorophosphate (LiPF
6) and indicates electrolyte leakage in the cell [
4].
Shaw-Stewart et al. [
5] inserted 18,650 batteries (fully charged) in 26 different liquid solutions for 24 h. The cell voltage, pH value, specific gravity, conductivity, and Electrochemical Impedance Spectroscopy (EIS) behaviour were tracked. It was found that the discharging process depends on the conductivity of the solution, the cell chemistry, and the competing reactions of the liquids. Furthermore, an electrolytic reaction due to Hydrogen gas was noticed at the cell’s negative terminal. This reaction was dependent on the pH level of the liquid and the corresponding reaction at the opposite electrode. On the positive cell terminal, water reduction took place, which resulted in the generation of gaseous oxygen. Shaw-Stewart et al. [
5] also considered the research undertaken by Lu et al. [
3] and interpreted that the OCV decrease was due to contact loss from the outside and found it improbable that the voltage loss represents actual chemical energy loss inside the cell.
Ojanen et al. [
6] investigated the usage of varying aqueous solutions of salts like NaCl, NaSO
4, FeSO
4, and ZnSO
4. In the experiment, the cylindrical cells were not inserted into the solutions, but only a set of platinum wires, which were fixed on both terminals on the cell, were inserted to measure the OCV. In this ex situ experiment, a 20% NaCl solution turned out to be an effective medium for the full and fast discharging of Li-ion cells within 4.4 h. Furthermore, it was also found that stirring the liquid can accelerate the discharging process, e.g., cells in the NaSO
4 solution reached a deactivation time of 3.1 h. Moreover, it was also found that metallic particles are an effective way to reduce deactivation time and can prevent corrosion of the connectors. It is recommended to add these particles in a 20 wt% NaCl solution. The commonality between research [
5,
6] shows that the positive terminal of a LIB is mainly affected by corrosion, and consequently uncontactable terminal leads to difficulties during discharging, leading to an incomplete discharge.
Based on the above literature [
3,
4,
5,
6], it can be seen that the majority of the research focuses on the insertion of LIBs in different liquid media, conductivity of the liquids, comparison of the specific weight, and comparison of the impedance spectra before and after insertion. However, none of those authors have performed an experimental investigation on commercial 18,650 cylindrical cells, which are inserted into DW mixed with CaCl
2. To the authors’ knowledge, until now, research related to deactivating the cells in mixtures with CaCl
2 has not been considered, and hence this concept makes this experimental investigation as novel research. With all the aforementioned concepts, CaCl
2 was chosen because of its availability, its nontoxicity, and its reactivity with the unwanted upcoming of the molecule HF.
Lithium Tetra Fluoroborate (LiBF
4) and LiPF
6 are the two commonly used salts in the electrolyte of LIB. Both salts possess the danger of Hydrofluoric acid (HF) formation when reacting with water or humid air. The chemical reaction sequence for the case of LiPF
6 is demonstrated in Equations (1)–(3) [
7],
However, CaCl
2 with HF forms a poorly soluble CaF
2 molecule along with HCL, as shown in Equation (4),
In this research, the behaviour of CaCl2 solution is studied by inserting the cells into the water with soluble CaCl2 for different lengths of time. These results will also explain the cell deactivation behaviour and the state of safety over time. The additional goal of this research is to assess the time and respective critical points at which the cell deactivates, and the results will indicate the overall time period for which a cell needs to be inserted in a liquid media to be considered safely deactivated.
To research the efficiency of the deactivation of Li-ion cells with a CaCl
2 solution, various comparative tests such as analysis of weight loss and cell insertion times into different media are carried out. DW and TW are used as reference deactivation mediums for the CaCl
2 solution. Additionally, the comparison between CaCl
2 and TW is made to recreate a more realistic review about deactivating the accident-damaged battery of electric vehicles in extinguishing containers [
2] because TW is easily accessible and is widely used for fire/smoke extinguishing purposes.
Overall, this work aims to give a wide-ranging analysis of the effects of deactivation via different liquid insertions. To investigate the occurring damage to the cells after insertion, different measurements were carried out. For these experiments, each cell was given a specific identification number to make sure that they do not mix up during the destructive test, and these numbers are only used for reference during the destructive abuse test/analysis, i.e., serve the purpose of a tag (TW-24h-1, i.e., TW = Tap water; 24 h = Insertion duration of 24 h; 1 = Cell number 1).
4. Discussion
Based on the distinct insertion media (CaCl2 solution, TW, and DW) used in this investigation for cell insertion, this section provides a discussion about the results obtained from the experiments.
4.1. Insertion in CaCl2 Solution
Considering the OCV curve of the cells, after the deactivation of cells in CaCl
2 solution, a voltage drops to 0.27 V after 5h can be observed as an over-discharge phase. After 2 h of insertion, a voltage drop close to the Umin (0% SoC) of the cell was determined. Theoretically, the voltage loss can derive from two causes: (a) the electrical energy of the cell could be discharged until it reaches the referenced voltage level, or (b) the terminals lost contact with the electrodes over time due to the corrosive reactions with the fluid. However, the authors suspect this voltage decrease can be attributed to the actual loss of electric energy of the cell, as this aligns with the results of the residual energy determination of the destructive test. The biggest voltage change (from more than 4 V to 0.25 V) occurred between 0.5 h and 5 h, and before this interval, the positive terminal reacts with the fluid and it dissolves (see
Figure 5). This is also supported by the measurement of the cell mass, as after 0.5 h, a slight decrease in cell mass (0.6%) was noticed, which can be attributed to the reaction of the terminals with the fluid (dissolution). At the 5 h insertion period, an increase of 2% in weight was noted, while the maximum weight (approx. 43.5 g) was noticed during the 2nd week (refer to
Figure 6). The authors attribute this to the opening of the cell and the reaction of the fluid with the inner materials, where incrustations may have formed over time and have contributed to the weight gain. This also aligns with the results from the CT scans in
Figure 20 at 48 h and the 1st week.
No drastic temperature rises or fires occurred during the destructive test conducted after 0.5 h (refer to
Figure 7). Even though the voltage of the 2 h deactivation cell was still above its minimal voltage according to the datasheet, no thermal reaction of the cell took place during the abusive test, which indicates only a limited amount of residual energy. This supports the deduction from the voltage analysis.
For the flame photometer, the initial lithium values were in the same range as subsequent values, leaving no significant change. From the measurement of the ion-selective electrode, no significant readings for any fluoride release were detected. All recorded F- values were below or nearby the minimum detection limit of 0.1 ppm, and thus cell opening time was unable to be analysed from the experiment.
For the conductivity reading of the CaCl2 fluid, an increment of 8.1% (up to 60 mS/cm) was noticed between 10 h and 2 weeks. It was expected that the release of additional ions from the inner parts of the cell would not lead to a drastic increase in conductivity, as the electrical resistance of the liquid was mainly dominated by the dissolved salt CaCl2 (2 mol/dm3 concentration). Although, such a high increase (8.1%) could be because of the cell being ruptured between 10 h and 2 weeks.
When considering the pH value of the CaCl
2 solution, all values were within the acidic category. The analysis of the pH value over time showed a declining trend from 0.5 h to 2 weeks. Though, the maximum fall in value was noticed between 48 h and 2 weeks, i.e., from pH 6 to pH 5.16 (refer to
Figure 19).
4.2. Insertion in TW
Looking at the OCV curve for cells that were deactivated in TW for 1 week, it can be said that the cells had a medium range voltage with 3.56 V. Furthermore, the value dropped at 0.86 V and 0.11 V for 2 weeks and 10.5 weeks, respectively, for which it can be considered as an over-discharged cell (based on OCV curve). For the CaCl2 solution, the voltage loss can be assigned to the two aforementioned phenomena: loss of contact with the electrodes or electrical discharge. From visual observations of the cell by the authors, both terminals—especially compared to the positive one in CaCl2—were still in a close to pristine condition. This voltage loss is attributed to the electrical discharge over the ions of the TW. This hypothesis can be supported by the conductivity readings, which were not as high as the CaCl2 solutions, but were still not as low as the value from DW. Additionally, the voltage loss between 48 h and the 1st week indicates a cell opening scenario. The temperatures of both test cells during the 48 h nail penetration for TW reach a maximum at 373.4 °C and 476.5 °C, respectively, with a steep incline in the beginning period, and thereafter, at 1 week insertion, the temperature curve for one cell was completely flat at 37.3 °C and the second cell showed minimal exothermal behaviour up to 91.8 °C. Both cells did not show any signs of fire, and thus, the residual stored energy must be very low. In particular, one cell after 10 h insertion also showed negligible exothermal behaviour, even though the longer insertion intervals (24 h and 48 h) do react in an exothermal manner. The authors credit this phenomenon to a probable cause in which the nail would be because of nail bending or the nail not hitting the active material of the cell correctly during the test.
During the period from 48 h to the 1st week, the most remarkable change in weight occurs, i.e., from 43.5 g to 44.1 g (1.4% increment; refer to
Figure 6). This can be attributed to a cell opening and internal reactions in the cell as well as to some residual remains from the liquid inside of the cell because the outer parts have been dried with a paper cloth and additionally air-dried for 15 min. It is also noteworthy that no initial weight decrease was observed as the terminals did not dissolve in reaction with the fluid compared to the CaCl
2 solution situation.
Regarding the readings of the flame photometer, the initial Li-ion values from 0 h to 48 h were in the range of 0.1 ppm. Thereafter, the Lithium concentration raised slightly up to 0.15 ppm between the 1st and 2nd week and reached a peak of 0.45 ppm at 10.5 weeks, which indicates a cell opening between 48 h and by the end of the 1st week.
The recorded F- concentration of TW showed reduced significance up to 48 h. For example, in the 1st week, one of the cells had a concentration of 0.046 ppm, and the second one had 0.163 ppm. After 2 weeks, this value reduced to 0.026 ppm and 0.089 ppm, and thereafter, it increased to 13 ppm and 9.67 ppm at 10.5 weeks. Between 48 h and the end of the 1st week, the fluoride content for one cell was above the detection range, which could be due to a cell opening for at least one of the cells and a reduction in residual energy. This matches with the behaviour of both cells during the nail penetration test in the 1st week, where one cell showed exothermal behaviour. More specifically, cell 81, which has a 0.046 ppm F-concentration and thus was less likely to have opened in the 1st week, showed higher exothermal reactivity during the abusive test (see
Figure 14), whereas cell 119, which has a 0.163 ppm F-concentration, implying to a cell rupture, shows the flat thermal behaviour at the nail test. For the 2 weeks experiments, no exothermal behaviour was detected, so there were no F-concentration readings. The authors suspect the reason for this contradictive behaviour to be F-concentration readings close to the detectable range of the device. This, in addition to the remaining unchanged Li-ion concentration from the flame photometry analysis for 1 and 2 weeks, suggests that the cells at 2 weeks should already have opened and decreased in residual energy. After 10.5 weeks, the cells were ruptured and reacted with the solution.
The conductivity readings of the TW displayed a slow monotonic decrease from 0.65 mS/cm at 0.5 h to 0.46 mS/cm at 48 h. From there on until the end of the 1st week, the reading dropped to 0.3 mS/cm. This 34% decline can be attributed to changes within the cell, i.e., it is suspected that ions within the TW may have reacted with the inner parts of the cell, and thus, they are unavailable for charge transport, which results in reduced conductivity. A similar pattern was noticed for the pH analysis as well. The initial readings up to 48 h were in the alkaline category, i.e., close to pH 8. At 48 h it reached pH 9 and dropped to the neutral area at pH 7.1 by end of the 1st week, moving further into the acidic domain. Fitting to the scheme, the most apparent change happens between 48 h and the end of the 1st week, i.e., change occurs due to the reaction of H2O and LiPF6 creating HF.
4.3. Insertion in DW
The OCV curve readings after the deactivation in DW showed no significant variation for up to 2 weeks. The last recorded value at 10.5 weeks shows the average voltage drop to 2.35 V and only one cell being at 4.08 V. This behaviour can be explained by the low conductivity of DW, as can be seen from the conductivity measurements. The low conductivity leads to the slow self-discharge of the cell in DW. From 0 h to 0.5 h the reading was very low, and, according to the authors, this could be due to some small particles of dust or salt possibly depositing on cells while handling them for the experiment. Overall, the conductivity measurement shows a 48% increase over time, from 0.5 h with 13.5 µS/cm to 20 µS/cm at 10.5 weeks.
The cell mass was nearly constant (at approx. 43.2 g with ±0.2 g) until 48 h of insertion, but then increased by 2.2% at the 1st week and stayed at that level until 10.5 weeks (refer to
Figure 6). This shift can be interpreted due to potential cell openings like TW and CaCl
2 solution. Considering the temperatures during the destructive nail test for the 2 weeks insertion, a noticeable observation for TW cells can be made. One cell (DW 76) reacted in a similar exothermal manner as the cells with lower insertion times, with a peak temperature of 322.4 °C, whilst the second cell (DW 84) showed a flattened curve, only reaching 79.6 °C without fire and ignition. From this, it can be said that, for deactivation in DW, the reactions that lower the residual energy content of the cell take place between the 1st and 2nd week, even though no distinct voltage loss is visible from the outside. A possible explanation of such could be due to the high impedance behaviour of the cell that did not react exothermally. From the destructive test for the DW cell at 2 weeks, it can be said that at least some energy was lost before, with one cell staying below 80 °C and without fire. This means that even though a high voltage is still present at 2 weeks, a reduced thermal reaction is observed, leading to the outcome that the voltage may not be a reliable indicator for all situations regarding deactivation.
When considering the readings of the Lithium concentration over time, no Lithium was found for the period of 0.5 h to 48 h. From the 1st week onwards, a Lithium concentration of 0.05 ppm started emerging and reached up to 0.1 ppm till 10.5 weeks. As DW is considered pure water (compared to 99.9% of CaCl2), no impurities from the solution are to be expected passed on. However, if the impurities are coming from the equipment, it is improbable that three distinct measurements contained the same error. Thus, the detected Li-ions are suspected to originate from the interior of the cell.
The readings of the Fluoride concentration of the DW samples showed quite low-value readings, i.e., below the tolerance of the equipment, and therefore, no significant additional information was gained. Similarly, the pH progression over time was comparatively stagnant between pH 8 and 9 for the interval of 0.5 h to 48 h. Thereafter, a negligible trend was identified towards neutral, and therefore, it can be said that the overall trend for the complete data set is alkaline.
4.4. Predictive Capabilities of the Measured Parameters
Throughout the experiment, various parameters were recorded, but not all readings were suitable for the determination of an accurate state of a cell during deactivation in a liquid medium. Therefore, this topic is addressed in this section, starting with the parameter voltage. The behaviour of the DW cells after 2 weeks of insertion was so noticeable that, even though a nominal voltage was close to full SoC before the destructive test, both cells reacted with reduced exothermal behaviour during the nail penetration test. The other reason towards to voltage parameter not being a completely reliable parameter is because of the dynamic reaction of the outer cell terminals with the fluid, especially in the case of the CaCl2 solution. The risk that emerges here is the faulty voltage readings, i.e., a false result due to the loose contact between the terminal and the electrode and due to incrustations at the terminals.
Another important parameter is cell mass. This parameter can be used for evaluating processes such as the dissolution of the positive terminal or formation of incrustations. However, in real-world applications, this is not pragmatic when considering larger battery packs. The additional way that can be used for identifying the activation/deactivation status of a cell is to undertake the mechanical destructive test (abusive test). The issue with this approach is that this method can be used for analytical investigations but not for analysing the large battery packs from the wrecked EV. The reason for this is that, in many cases, it is impossible to remove the cell from the pack.
One of the other approaches is to look for the lithium concentration in the fluid and identify whether the cell casing has opened and if there are any major impurities in the solution. However, during the experiments, it was found that the value of impurities was below the detection tolerance of the measurement setup, and therefore, no readings were identified, except for the last insertion period for DW, where nominal readings were recorded. Therefore, with the help of this approach, it can be said that, if there is a deflection in reading, then the cell is deactivated.
Moreover, the conductivity parameter could also be used for evaluation purposes, but the issue with such a parameter is that the influence of other ions present in the solution, along with the potential dirt and dust, can falsify the results. Similarly, the pH value as a parameter is also not well-suited as a small change in the concentration of H30+ or OH- ions may lead to a high change in the pH value, i.e., demonstrating drastically wrong results.
5. Conclusions
In this research, various aspects (mass loss, voltage loss, pH, conductivity, Lithium and fluoride concentration, and abusive behaviour) of three different media (TW, DW, and CaCl
2 solution) were analysed and compared against each other to see if they demonstrated a safe and reliable deactivation process of Li-ion cylindrical cells. An exemplary general overview of the relative changes in the various parameters is summarized in
Figure 21.
From
Figure 21, it can be seen that the addition of the salt CaCl
2, in combination with water, can accelerate and improve the deactivation procedure. The reason for such is that the supplement of CaCl
2 leads to a faster voltage decrease in comparison with the TW and DW, respectively. After 5 h of insertion time, the CaCl
2 solution led to a voltage close to 0 V. Insertion in TW resulted in the most noteworthy change between 48 h and the 1st week, where the destructive nail test shows a reduction in temperature, and thus reduced residual energy content, meaning a safe deactivation is possible between 48 h and the 1st week.
Looking at the third deactivation fluid (DW), the inserted cells retained their voltage (close to Umax) for up to 2 weeks. It was only for the last reading at 10.5 weeks where the voltage dropped to 0.62 V for one of the two samples. From the destructive test of these two samples, it was concluded that at least some energy was lost before (up to the 2 weeks interval), with one cell staying below 80 °C and without fire. This means that even though a high voltage was present at 2 weeks, a reduced thermal reaction was observed, leading to the suspicion that the voltage may not be a completely reliable indicator for all situations concerning deactivation.
Accordingly, another reason why the voltage may not be the only parameter is the dynamic reaction of the outer cell terminals with the fluid, especially for CaCl2. Here, the risk emerges that a false deduction can be made, as the measurement may deliver a faulty voltage due to contact loss between the terminal and the electrode, and due to incrustations at the terminals. The dissolution of the positive terminal is confirmed by the weight measurements for CaCl2. This risk is mainly expressed for CaCl2 solution deactivated cells as the other two references (TW and DW cells) did not display such anomalies. Regarding the discharge process, CaCl2 solution deactivated cells were initially expected to electrically discharge much faster because of the high volume of the ions in the solution, whereas for TW, this process takes much longer, and for DW, almost no electrical discharge was noticed. Deriving from the results, it seems that for the first two instances, this effect occurs, whereas for DW, not much discharge is expected, and the authors expect the cell to have opened in the period between 48 h and 2 weeks.
When considering the emergence of HF, it is difficult to give a clear statement, as the only fluoride reading above the detection range was measured for TW after 10.5 weeks with 11.3 ppm. For the CaCl2 solution deactivated cell, no fluoride was measured because it was captured completely by the calcium ions, which confirms the positive effect of CaCl2.
Altogether, the parameters within the circumstances from this investigation can be placed in the following order from most suitable to least suitable: Abuse testing > Voltage > Lithium concentration > Mass > Conductivity > Fluoride concentration > pH. Out of all three solutions, the CaCl2 solution yielded the best results in terms of required time for residual energy loss (with 2 h) and cell opening. Additionally, as no F- ions were detected, the authors suppose that if any HF was released, it was likely captured by the calcium ions.