Elucidation of Side Reactions in Lithium-ion Batteries with Electrolyte Decomposition Products via Overdischarge for Li[Li_1/3Ti_5/3]O₄/Li[Li_0.1Al_0.1Mn_1.8]O₄ Cells with an Imbalanced State-of-Charge

Abstract

A major factor in the capacity fading of lithium-ion batteries is the imbalance in the state-of-charge (SOC) between the positive and negative electrodes, which is caused by the difference in the side-reaction current between the two electrodes. Therefore, it is crucial to understand these side reactions to extend battery life. There are two types of side-reaction current, I_SR: “intrinsic”-I_SR results from electrolyte decomposition, while “additional”-I_SR results from the reaction of decomposition products. The “additional”-I_SR is closely related to crosstalk reactions. In this study, we conducted overdischarge tests on Li[Li_1/3Ti_5/3]O₄ (LTO)/Li[Li_0.1Al_0.1Mn_1.8]O₄ (LAMO) cells with the SOC of the positive and negative electrodes intentionally imbalanced to understand the nature of the “additional”-I_SR. While the cell capacity initially increased after the overdischarge, the recovered capacity decreased in subsequent cycles. This phenomenon can be explained well by a side-reaction model that considers the oxidant generated in the cell due to overdischarge, which indicates that the concentrations of oxidants and reductants in the cell are important for the I_SR.

1. Introduction

Lithium-ion batteries (LIBs) are widely used in portable electronics, electric vehicles, and grid storage systems, among other applications. Their long lifetimes is a key advantage, among those of other types of rechargeable batteries.^1–3 This is because the lithium insertion materials that are used as positive and negative electrodes are highly resistant to degradation because they maintain their structural framework during lithium insertion reactions.^4–9 Consequently, LIBs can be cycled hundreds or even thousands of times, without significant deterioration. Nonetheless, LIBs still exhibit capacity fading, as differences in the side-reaction currents of the electrodes cause an imbalanced state-of-charge (SOC) between the positive and negative electrodes, which leads to reduced battery capacity.^10–14

Many investigations have been conducted on the side reactions that cause a SOC imbalance between electrodes,^15–20 with a particular focus on crosstalk reactions in a cell. In this paper, “side reactions” includes “parasitic reactions”, because parasitic reactions are essentially local cell reactions constituting insertion reactions and side reactions. We previously reported that the side-reaction current measured in a symmetrical cell differed from that measured in an actual full-cell configuration,^19,20 indicating the presence of an additional side reaction at the electrodes. According to a study using a four-electrode cell,²¹ this additional side reaction involves the decomposition products generated at one electrode reacting with the other electrode. Kinetic analysis of the additional side-reaction current showed that the amount of side-reaction decomposition products significantly affected the side-reaction currents at both electrodes.

In this study, we examined the effect of the amount of decomposition products on the side-reaction current in LIBs using overdischarge tests. We found that when an Li[Li_1/3Ti_5/3]O₄ (LTO)/Li[Li_0.1Al_0.1Mn_1.8]O₄ (LAMO) cell with intentionally misadjusted SOC for the LAMO positive and LTO negative electrodes was overdischarged, the cell capacity temporarily increased, but the recovered capacity decreased in subsequent cycles. We analyzed the decrease in the recovered capacity in relation to the change in the side-reaction currents caused by the generation of decomposition products during overdischarge.

2. Experimental

2.1 Fabrication of electrodes and electrochemical cells

Li[Li_1/3Ti_5/3]O₄ (LTO; Ishihara Sangyo Co., Ltd.) and Li[Li_0.1Al_0.1Mn_1.8]O₄ (LAMO; Tosoh Corp.) were used as active materials. The electrodes consisted of 88 wt% active material, 6 wt% acetylene black, and 6 wt% polyvinylidene fluoride (PVDF). A slurry prepared by adding a mixture of the active materials, acetylene black, and PVDF into an N-methyl-2-pyrrolidone solution was cast onto an Al foil as a current collector, which was then dried under vacuum at 150 °C overnight. After drying, a disk (16 mm diameter, 2 cm²) was punched out and used as the electrode. The electrode weight and thickness were 30–35 mg and 118–135 µm, respectively, for the LTO electrodes and 27–39 mg and 100–121 µm, respectively, for the LAMO electrodes. Two sheets of Celgard 2320 were used as separators. A 0.2 mL solution of 1 M LiPF₆ dissolved in ethylene carbonate/dimethyl carbonate (3/7 by volume) was used as the electrolyte. The cell structure and fabrication procedure have been described previously.¹⁹

2.2 Overdischarge tests for the LTO/LAMO cells

Figure 1 shows the scheme of the procedure for the overdischarge tests for the LTO/LAMO cell. Both the LTO and LAMO electrodes were preliminarily examined in a Li half-cell to confirm their voltage profiles and reversible capacity (Fig. 1a). Before fabricating the LTO/LAMO cells, the LTO electrode was oxidized fully at 3 V (Fig. 1b), while the LAMO electrode was charged to 4.5 V and then discharged (reduced) by 1 mAh (Fig. 1c). The fabricated LTO/LAMO cells were pre-cycled at 0.5 mA cm⁻² and 25 °C for 10 cycles (Fig. 1e). A schematic illustration of the overdischarge tests for the LTO/LAMO cell is shown in Fig. 2. As the cell capacity was limited by the LAMO positive and LTO negative electrodes during charging and discharging, respectively, the cell had a capacity of only 1 mAh (Fig. 2a). This created a state in which the cell capacity faded owing to the imbalanced SOC between the positive and negative electrodes.¹⁴ In the overdischarge tests, the cell was discharged once to 1.5 V at a constant current density of 0.5 mA cm⁻², and then overdischarged with various capacities (namely overdischarge capacity, Q_OD = 0.25, 0.75, 1.25 mAh) (Fig. 1f). During overdischarge, as the LAMO positive electrode had residual discharge capacity, lithium insertion (reduction) occurred. In contrast, lithium insertion could not occur at the LTO negative electrode because it was fully oxidized; thus oxidative decomposition of the electrolyte occurred at the LTO electrode. Therefore, by overdischarging the LTO/LAMO cell, the SOC of the LAMO electrode changed, whereas that of the LTO electrode remained unchanged, as shown in Fig. 2b. Consequently, the overdischarge capacity (Q_OD) was added to the cell capacity after overdischarge. After the overdischarge, a cycle test was performed for 150 cycles under the same conditions as the pre-cycle (Fig. 1g). Electrochemical tests were performed using a charge-discharge tester (Battery Laboratory System, Keisokuki Center Co. Ltd. Japan) was used.

Figure 1.

Scheme of the procedure for the overdischarge tests for the LTO/LAMO cell.

Figure 2.

Schematic diagrams of the overdischarge tests of the LTO/LAMO cell. (a) The LTO/LAMO cell was fabricated by combining the fully oxidized LTO and partially charged LAMO electrodes. (b) During overdischarge of the cell, side reactions occur at the LTO negative electrode and Li insertion occurs at the LAMO positive electrode, resulting in increased cell capacity by overdischarge capacity (Q_OD); Q after overdischarge test = 1 mAh + Q_OD.

2.3 Calculation of side-reaction current using cumulative capacity plots

The side-reaction currents of the LTO negative and LAMO positive electrodes were calculated as follows. First, the charge-end and discharge-end capacities (CQ_Cha and CQ_Dis) were estimated from continuous charge-discharge curves, where the discharge and charge curves were drawn starting from the end of the previous charge and discharge curves, respectively. When the cell capacity for charge and discharge was limited by the capacities of the LAMO positive and LTO negative electrodes, respectively, CQ_Cha and CQ_Dis were determined by the following equations:^19,22   

\begin{equation} CQ_{\text{Cha}} (t) = Q_{\text{Cell}} + \int I_{\text{SR,LAMO}}\text{d}t, \end{equation} (1)

and   

\begin{equation} CQ_{\text{Dis}} (t) = \int I_{\text{SR,LTO}} \text{d}t \end{equation} (2)

where Q_Cell is the initial cell capacity, and t is the cycle time. By modifying the above equations, the following equations were derived to determine the side-reaction currents of LTO and LMAO, I_SR,LTO and I_SR,LAMO, respectively.   

\begin{equation} I_{\text{SR,LAMO}} = \text{d}CQ_{\text{Cha}} (t)/\text{d}t, \end{equation} (3)

and   

\begin{equation} I_{\text{SR,LTO}} = \text{d}CQ_{\text{Dis}} (t)/\text{d}t \end{equation} (4)

Therefore, when CQ_Cha and CQ_Dis were plotted against cycling time, namely cumulative capacity plots,^19,22 I_SR,LAMO and I_SR,LTO corresponded to the slopes of the lines of CQ_Cha and CQ_Dis, respectively.

3. Results and Discussion

3.1 Procedure for capacity recovery by overdischarge tests

Figure 3 shows the continuous charge-discharge curves of the LTO/LAMO cells during the overdischarge tests. During overdischarge, the cell voltage dropped sharply to approximately −1.0 V, and then gradually decreased to −1.8 V. If lithium insertion occurred in the LAMO positive electrode during overdischarge, the electrode potential of the LAMO was maintained at approximately 4.1 V vs. Li. Therefore, when the cell voltage dropped to −1.5 V, the electrode potential of the LTO negative electrode rose to 5.7 V. Therefore, the voltage plateau at −1.5 V during overdischarge indicates that oxidative decomposition of the electrolyte occurred on the LTO negative electrode.^23–27 Immediately after the overdischarge, the cell voltage increased sharply and became constant at 2.5 V. This indicates that lithium insertion occurred in both the LTO and LAMO electrodes,^28–30 because the electrode potential was 1.5 V vs. Li for the LTO^31–34 and 4.1 V vs. Li for the LAMO.^35,36 The cell capacity clearly increased immediately after overdischarge, and the increase was proportional to the overdischarge capacity Q_OD. However, the cell capacity decreased in the subsequent cycles. It is noticed that the amount of electrolyte decomposed due to overdischarge is less than 0.05 mmol, which is smaller than the amount of electrolyte in the cell (0.2 mmol). Therefore, there is no significant change in the concentration of electrolyte solution in the cell.

Figure 3.

Charge-discharge curves of LTO/LAMO cells for overdischarge tests. The cells were overdischarged for (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh. The green lines indicate the curves during overdischarge. The charge and discharge curves were continuously depicted from the end of the previous discharge and charge curves, respectively (called continuous charge-discharge curves). The shifts in the end of the charge and discharge capacities were caused by side reactions on the LAMO positive and LTO negative electrodes, respectively.

Figure 4 shows the cell capacity during the over-discharge tests. The initial cell capacity was 1.01 to 1.07 mAh for all cells; after 10 cycles it was essentially unchanged, 0.96 to 1.03 mAh. The charge capacities of the cells immediately after overdischarge were 1.27, 1.70, and 2.28 mAh for Q_OD = 0.25, 0.75, and 1.25 mAh, respectively, indicating that the cell capacity increased by Q_OD. The subsequent discharge capacity was 1.11, 1.24, and 1.50 mAh for Q_OD = 0.25, 0.75, and 1.25 mAh, respectively. These values were significantly lower than the charge capacity but were clearly higher than the capacity before overdischarge. The increased cell capacity after overdischarge means that the capacity loss caused by the SOC imbalance of the positive and negative electrodes could be recovered by overdischarging the cell. During overdischarge, oxidative decomposition of the electrolyte occurred at the LTO negative electrode, whereas the discharge reaction occurred at the LAMO positive electrode. Consequently, overdischarge changed the SOC of the LAMO electrode, despite the uncharged SOC of the LTO electrode.

Figure 4.

Cell capacity of the LTO/LAMO cells for overdischarge of (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh. Before the overdischarge tests, the cells were cycled 10 times.

Unfortunately, the cell capacity gradually decreased in the subsequent cycles. The capacity fading was faster for a larger Q_OD. The cell capacity at the 150th cycle after overdischarge was 0.89, 0.90, and 0.95 mAh for Q_OD = 0.25, 0.75, and 1.25 mAh, respectively. These values were similar to the capacity of the cell without overdischarge (0.83 mAh). The capacity recovered by overdischarge diminished in subsequent cycles owing to more severe capacity fading in the cells with a larger Q_OD. This suggests that the side-reaction current in the positive and negative electrodes was greatly altered after overdischarge; thereby, the SOC balance between the LAMO positive and LTO negative electrodes returned to its original state.

3.2 Calculation of side-reaction current in overdischarged LTO/LAMO cells

To clarify the cause of the reduced recovered capacity after overdischarge of the LTO/LAMO cells, the side-reaction currents of the LAMO positive and LTO negative electrodes (I_SR,LTO and I_SR,LAMO) were calculated. The side-reaction currents were calculated from cumulative capacity plots, where the charge-end and discharge-end capacities (CQ_Cha and CQ_Dis) were plotted against cycling time. According to Eqs. 3 and 4, the cumulative capacity plots shown in Fig. 5, the slopes of CQ_Dis and CQ_Cha correspond to I_SR,LTO, I_SR,LAMO, respectively.

Figure 5.

Cumulative capacity plots of the LTO/LAMO cells for overdischarge of (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh. Closed and open symbols indicate CQ_Cha and CQ_Dis, respectively. Side-reaction current of LAMO positive and LTO negative electrodes were calculated from the slope of the charged-end and discharge-end capacities, respectively; I_SR,LAMO = dCQ_Cha/dt and I_SR,LTO = dCQ_Dis/dt.

Figure 6 shows I_SR,LTO calculated from the slope of the CQ_Dis in the cumulative capacity plots in Fig. 5. I_SR,LTO was 5 µA before overdischarge (t < 0 h), but just after overdischarge (t ∼ 0 h), it increased significantly depending on Q_OD; 8.4, 16.9, and 19.4 µA for Q_OD = 0.25, 0.75, and 1.25 mAh, respectively. This indicates that oxidative decomposition products generated at the LTO electrode during overdischarge were reduced at the LTO electrode. During the cycles after overdischarge (t > 0 h), I_SR,LTO largely decreased with increasing Q_OD, and after 50 h, it asymptotically reached a constant value. After the final cycle (t = 150 h), I_SR,LTO was 1.9, 3.0, and 3.5 µA for Q_OD = 0.25, 0.75, and 1.25 mAh, respectively. These values were slightly larger than the value without overdischarge (1.6 µA).

Figure 6.

Side-reaction current of the LTO negative electrode (I_SR,LTO) calculated from the cumulative capacity plots shown in Fig. 5. The LTO/LAMO cells were overdischarged for (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh.

I_SR,LAMO was also calculated from the slope of CQ_Cha (Fig. 7). Before overdischarge (t < 0 h), I_SR,LAMO was ∼2.5 µA, which was smaller than I_SR,LTO. This means that electrolyte decomposition at the LAMO electrode was slower than at the LTO electrode, in agreement with previous reports.¹⁹ Immediately after overdischarge (t ∼ 0 h), I_SR,LAMO increased with increasing Q_OD: 3.6, 9.1 and 13.2 µA for Q_OD = 0.25, 0.75 and 1.25 mAh, respectively. The increase in I_SR,LAMO might have been caused by the reductive decomposition products generated at the LTO negative electrode during the overdischarge being oxidized at the LAMO positive electrode. After overdischarge (t > 0 h), I_SR,LAMO showed the same trend as I_SR,LTO; it rapidly decreased and then asymptotically approached a constant value. After the final cycle (t = 150), I_SR,LAMO was 1.3, 2.2, and 1.9 µA for Q_OD = 0.25, 0.75, and 1.25 mA, respectively. These values were slightly higher than for the cell without overdischarge (1.1 µA).

Figure 7.

Side-reaction current of the LAMO positive electrode (I_SR,LAMO) calculated from the cumulative capacity plots shown in Fig. 5. The LTO/LAMO cells are overdischarged for (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh.

Based on the cell mass balance,¹⁴ the capacity fading of the cell is proportional to the difference in the side-reaction currents, Q_fade = |I_SR,LTO − I_SR,LAMO| × t. Before overdischarge, the difference was very low (<2 µA), resulting in very slow capacity fading. However, the difference dramatically increased due to overdischarge because I_SR,LTO increased significantly more than I_SR,LAMO. After the overdischarge, the difference in the side-reaction currents reached ∼5 µA, which was more than twice that before the overdischarge. Consequently, the recovered capacity was reduced owing to the change in side-reaction currents.

To explain the loss of recovered capacity, the following reaction mechanism is proposed by considering the side reactions at the LTO electrode. Before overdischarge (Fig. 8a), a side reaction due to reductive decomposition of the electrolyte occurred at the LTO negative electrode. The side reaction was slow because of the relatively high LTO electrode potential (1.55 V vs. Li), indicating that the side-reaction current I_SR,LTO was small. This explanation agrees well with the results shown in Fig. 6, where I_SR,LTO was small for t < 0 h. When the LTO/LAMO cell was overdischarged, oxidative decomposition of the electrolyte occurred at the LTO electrode, which generated a high concentration of oxidants, C_Ox, in the electrolyte (Fig. 8b). During overdischarge, Q_OD affected the SOC of only the LAMO positive electrode, thereby increasing the cell capacity. Because C_Ox was high immediately after overdischarge, a side reaction at the LTO electrode occurred due to the reduction of its oxidant (Fig. 8c). Therefore, I_SR,LTO was increased significantly by overdischarge, which is consistent with the significantly increased I_SR,LTO at t ∼ 0 h. In the cycle after overdischarge, the oxidant was no longer generated; thus, C_Ox decreased because of the reduction of the oxidant at the LTO electrode. Because the rate of the reduction of the oxidants is proportional to C_Ox,²¹ I_SR,LTO decreased with the cycling time. This assumption can explain why I_SR,LTO gradually decreased with cycling time.

Figure 8.

Schematic illustration of side reactions occurring at the LTO negative electrode (a) before, (b) during, and (c) after overdischarge. (a) Before overdischarge, only the “intrinsic” side reaction, where electrolytes are reduced by the LTO electrode, occurred. (b) During overdischarge, oxidative decomposition of the electrolyte occurred on the LTO electrode, generating an oxidant. (c) After overdischarge, the oxidative decomposition product was reduced at the LTO electrode; therefore, LTO is oxidized at the rate of I_SR.

3.3 Analysis of crosstalk reactions by fitting side reaction currents

Figure 9 shows a schematic of the side reactions at the LTO negative electrode. At the LTO electrode, the electrolyte continuously decomposes to generate a reductant (Red′), which corresponds to an “intrinsic” side-reaction current (I_SR,Int). Because there is sufficient electrolyte in the cell, I_SR,Int can be considered constant. In addition, the oxidant (Ox) generated at the LAMO positive electrode is transported to the LTO negative electrode and reacts to produce the reductant (Red), which corresponds to an “additional” side-reaction current (I_SR,Add) at the LTO negative electrode. Therefore, the actual side-reaction current of the LTO negative electrode is the sum of the two side-reaction currents I_SR,Int and I_SR,Add.

Figure 9.

Side reactions at the LTO negative electrode after overdischarge. Current for the “intrinsic” side reaction (I_SR,Int) was caused by electrolyte decomposition, and that for the “additional” side reaction (I_SR,Add) by the reduction reaction of oxidants generated at the LAMO positive electrode.

Based on the schematic diagram, we derived a kinetic equation for the side-reaction current at the LTO electrode to fit the observed-I_SR,LTO curves shown in Fig. 6. The time variation of C_Ox [mol cm⁻³] near the LTO electrode (dC_Ox(t)/dt) was determined by the supply rate v_Ox′ [mol s⁻¹] of the oxidant generated at the LAMO positive electrode and the consumption rate v_Ox [mol s⁻¹] of the oxidant at the LTO negative electrode.   

\begin{equation} \frac{dC_{\text{Ox}}(t)}{dt} = \frac{1}{V} (v_{\text{Ox}}{}^{\prime} - v_{\text{Ox}}) \end{equation} (5)

where V [cm³] is the amount of the electrolyte in the cell. According to our previous study,²¹ I_SR,Add is correlated with the concentration of the oxidant, C_Ox, and follows first-order kinetics. Thus, v_Ox can be expressed using Eq. 6.   

\begin{equation} v_{\text{Ox}}\ [\text{mol/sec}] = k\ [\text{cm$^{3}$/sec}]\times C_{\text{Ox}} (t)\ [\text{mol/cm$^{3}$}] \end{equation} (6)

where k is the rate constant for reduction of the oxidant. Substituting Eq. 6 into Eq. 5 gives Eq. 7:   

\begin{equation} \frac{dC_{\text{Ox}} (t)}{dt} = \frac{1}{V}(v'_{\text{Ox}} - kC_{\text{Ox}} (t)) \end{equation} (7)

Solving this differential equation for C_Ox(t) yields Eq. 8, assuming that v_Ox′ is a constant independent of t, and that this assumption is valid for t > 50 h from Fig. 7 because of the almost constant I_SR,LAMO in that time region.   

\begin{equation} v_{\text{Ox}} = v'_{\text{Ox}} + \{ k \cdot C_{\text{Ox}} (50) - v'_{\text{Ox}}\}\exp \left(-\frac{k}{V}t \right) \end{equation} (8)

C_Ox(50) is the concentration of the oxidant at t = 50 h. As shown in Eq. 9, v_Ox is correlated to the additional side-reaction current I_SR,Add.   

\begin{equation} v_{\text{Ox}}\ [\text{mol/sec}] = \frac{I_{\text{SR,Add}}\ [\text{A}]}{n_{\text{Ox}} \times F\ [\text{C/mol}]} \end{equation} (9)

Here, n_Ox is the number of electrons participating in the reduction reaction and F is the Faraday constant. Substituting Eq. 9 into Eq. 8 yields Eq. 10, expressing the time variation of I_SR,Add.   

\begin{align} &I_{\text{SR,Add}} (t) \\ &\quad= n_{\text{Ox}} \cdot F \cdot v'_{\text{Ox}} + n_{\text{Ox}} \cdot F \cdot \{ k \cdot C_{\text{Ox}} (50) - v'_{\text{Ox}} \}\exp \left( -\frac{k}{V}t \right) \end{align} (10)

From I_SR,LTO = I_SR,Int + I_SR,Add,   

\begin{equation} I_{\text{SR,LTO}} (t) = I_{\text{SR,Int}} + \alpha + \{ \beta C_{\text{Ox}} (50) - \alpha \} \exp (\gamma t) \end{equation} (11)

  

\begin{equation} \alpha = n_{\text{Ox}}Fv'_{\text{Ox}},\quad \beta = n_{\text{Ox}} Fk,\quad \gamma = -\frac{k}{V} \end{equation} (12)

In Eq. 11, the side-reaction current of the LTO negative electrode can be expressed by the concentration of the oxidant at t = 50 h and the four parameters α, β, γ and I_SR,Int.

To verify the validity of the model, the calculated-I_SR,LTO derived from formula 11 was used to fit the observed-I_SR,LTO shown in Fig. 6. Since our earlier work showed that I_SR,Int at the LTO electrode was negligible (∼0.2 µA),²² we ignored I_SR,Int for fitting. Figure 10 shows the fitting results of I_SR,LTO for each Q_OD with β × C_Ox(50), γ, and I_SR,Int as variable parameters. For any Q_OD, the calculated-I_SR,LTO reproduced the observed values well in the region after 50 h. The average values of errors in observed- and calculated-I_SR,LTO were at most 30 nA; 15 nA for Q_OD = 0 mAh, 15 nA for Q_OD = 0.25 mAh, 20 nA for Q_OD = 0.75 mAh, and 32 nA for Q_OD = 1.25 mAh. The errors were less than 1 %, confirming that the calculated values agreed well with the observed values. Therefore, after ∼50 h the side reaction at the LTO negative electrode progressed according to the model shown in Fig. 9.

Figure 10.

Fitting results of I_{SR, LTO} for LTO/LAMO cells over-discharged for (a) 0, (b) 0.25, (c) 0.75, and (d) 1.25 mAh. Open symbols and solid lines indicate observed and calculated-I_{SR, LTO}, respectively.

The value of each parameter obtained by fitting was plotted against Q_OD (Fig. 11). The value of α increased with increasing Q_OD: 1.21, 1.31, 1.50, and 1.81 µA for Q_OD = 0, 0.25, 0.75, and 1.25 mAh, respectively. Because I_SR,Int should be proportional to the supply rate of the oxidant (ν_Ox′) carried by crosstalk from the LAMO positive electrode, the increase in I_SR,Int proportional to Q_OD was due to the increase in I_SR,LAMO, as shown in Fig. 7. The value of γ was almost unchanged with the magnitude of Q_OD: −9.0 × 10⁻³, −11.2 × 10⁻³, −8.4 × 10⁻³, and −8.8 × 10⁻³ s⁻¹ for Q_OD = 0, 0.25, 0.75, and 1.25 mAh, respectively. This means that the rate constant for decomposition of the oxidant at the LTO negative electrode (k) was constant, regardless of Q_OD. The value of β × C_Ox(50) increased proportionally to Q_OD: 1.66, 2.42, 4.63, 5.34 mA for Q_OD = 0, 0.25, 0.75, and 1.25 mAh, respectively. Because the decomposition of the oxidant at the LTO negative electrode did not change, β should remain unchanged regardless of Q_OD because of the constant values of n_Ox and k. Thus, the increase in β × C_Ox(50) is due to the change in C_Ox(50), which is consistent with the assumption that the oxidant concentration near the LTO negative electrode increased during the overdischarge. The change in the parameters as a function of Q_OD was verified by the side-reaction model shown in Fig. 9 after overdischarge.

Figure 11.

Plots of the fitting parameters as a function of Q_OD: (a) βC_OX(50), (b) α, and (c) γ.

3.4 Side reactions in the LIBs leading to the SOC imbalance of the electrodes

Based on the above results, we have summarized the side reactions occurring inside the LTO/LAMO cell (Fig. 12). Because of electrolyte decomposition, an “intrinsic”-I_SR always flows at both electrodes, and the reductant (Red)/oxidant (Ox) is constantly generated by reductive/oxidative decomposition at the negative/positive electrode, respectively. A part of these products is transported to the other electrode and Red/Ox is oxidized/reduced at the positive/negative electrode by the crosstalk reaction, causing an “additional”-I_SR. Therefore, the “additional”-I_SR depends on the concentration of the oxidant (C_Ox) and reductant (C_Red) near the electrodes.

Figure 12.

Schematic illustration of side reactions occurring in lithium-ion batteries.

To suppress the capacity fading of LIBs caused by the SOC imbalance between the two electrodes, it is important to match the “actual”-I_SR of both electrodes, which is the sum of “intrinsic”- and “additional”-I_SR. If the crosstalk reaction can be diminished, the “additional”-I_SR becomes zero. In this case, the actual-I_SR at both electrodes should be equal to the “intrinsic” I_SR and independent of C_Ox and C_Red. In other words, if there is no crosstalk reaction, the magnitude of I_SR does not change, even if C_Ox and C_Red in the cell change. Therefore, if the crosstalk reaction can be prevented, the battery capacity can be permanently recovered without losing the recovered capacity during subsequent cycles. One approach to prevent the crosstalk reaction is to use a solid electrolyte, which stops the migration of the oxidant and reductant to the other electrode.^37–40

If the crosstalk reaction cannot be eliminated, battery life could be extended by controlling C_Ox and C_Red to match the I_SR of the positive and negative electrodes. A significant change in C_Ox/C_Red caused by overdischarge will increase the difference in I_SR and cause drastic capacity fading. Subsequently, as the side reaction proceeds, excess oxidant will be consumed, C_Ox/C_Red will approach a steady value, and capacity fading will slow. This is because the capacity-fading rate slows as the battery is cycled, and C_Ox/C_Red becomes steady as the side reactions progress. One hypothesis to explain why additives extend battery life is that they alter C_Ox/C_Red, resulting in a more balanced I_SR.^40–44 As mentioned above, in extending battery life, C_Ox and C_Red must be considered because I_SR is a function of C_Ox and C_Red.

4. Conclusions

We have explored the impact of decomposition products on the side-reaction currents at the electrodes through overdischarge tests of LTO/LAMO cells where the SOC of the LAMO positive and LTO negative electrodes was intentionally imbalanced. When overdischarge was conducted at various capacities, the cell capacity increased proportionally to the overdischarge capacity, indicating that the SOC of the LAMO electrode shifted, while the SOC of the LTO electrode remained constant. However, the recovered capacity decreased in the subsequent cycles. To understand this phenomenon, we calculated the side-reaction currents in the LAMO positive and LTO negative electrodes and found that both side-reaction currents significantly increased because of overdischarge. The more notable increase in the side-reaction current at the LTO negative electrode compared to the LAMO positive electrode led to severe capacity fading and the loss of recovered capacity. This indicates that the significant change in the side-reaction currents at both electrodes accelerated capacity fading.

To investigate the relationship between the side-reaction currents and decomposition products, a side-reaction model was developed by considering the change in the concentration of decomposition products due to their generation and consumption, and an equation for estimating the side-reaction current was derived. The calculated values of the side-reaction currents were in good agreement with the observed values, demonstrating that the oxidant concentration had a significant effect on the side-reaction currents at the LTO electrode. This means that it is necessary to adjust the concentrations of the oxidant and reductant produced and consumed by side reactions to suppress capacity fading, by matching the side-reaction currents of both electrodes. In other words, we determined that it is crucial to consider the mass balance in the cell, including decomposition products generated by side reactions, to understand battery deterioration due to SOC imbalance.

CRediT Authorship Contribution Statement

Taichi Nakamura: Data curation (Lead), Formal analysis (Equal), Investigation (Equal), Writing – original draft (Lead)

Kingo Ariyoshi: Conceptualization (Lead), Formal analysis (Equal), Writing – review & editing (Lead)

Conflict of Interest

The authors declare no conflict of interest in the manuscript.

Footnotes

K. Ariyoshi: ECSJ Active Member

References

• 1)  J.-M. Tarascon and M. Armand, Nature, 414, 359 (2001).
• 2)  G. Zubi, R. Dufo-López, M. Carvalho, and G. Pasaoglu, Renewable Sustainable Energy Rev., 89, 292 (2018).
• 3)  M. Li, J. Lu, Z. Chen, and K. Amine, Adv. Mater., 30, 1800561 (2018).
• 4)  M. Winter, J. O. Besenhard, M. E. Spahr, and P. Novak, Adv. Mater., 10, 725 (1998).
• 5)  M. Broussely, P. Biensan, and B. Simon, Electrochim. Acta, 45, 3 (1999).
• 6)  M. S. Whittingham, Chem. Rev., 104, 4271 (2004).
• 7)  T. Ohzuku and R. J. Brodd, J. Power Sources, 174, 449 (2007).
• 8)  B. L. Ellis, K. T. Lee, and L. F. Nazar, Chem. Mater., 22, 691 (2010).
• 9)  A. Manthiram, ACS Cent. Sci., 3, 1063 (2017).
• 10)  P. Arora, R. E. White, and M. Doyle, J. Electrochem. Soc., 145, 3647 (1998).
• 11)  J. Christensen and J. Newman, J. Electrochem. Soc., 150, A1416 (2003).
• 12)  J. Christensen and J. Newman, J. Electrochem. Soc., 152, A818 (2005).
• 13)  S. R. Li, C. H. Chen, X. Xia, and J. R. Dahn, J. Electrochem. Soc., 160, A1524 (2013).
• 14)  K. Nakura, Y. Ohsugi, M. Imazaki, K. Ariyoshi, and T. Ohzuku, J. Electrochem. Soc., 158, A1243 (2011).
• 15)  R. D. Deshpande, P. Ridgway, Y. Fu, W. Zhang, J. Cai, and V. Battaglia, J. Electrochem. Soc., 162, A330 (2015).
• 16)  J. Xu, R. D. Deshpande, J. Pan, Y.-T. Cheng, and V. S. Battaglia, J. Electrochem. Soc., 162, A2026 (2015).
• 17)  S. E. Lee, O. C. Harris, T. Siboonruang, and M. Tang, Joule, 5, 551 (2021).
• 18)  T. Boulanger, A. Eldesoky, S. Buechele, T. Taskovic, S. Azam, C. Aiken, E. Logan, and M. Metzger, J. Electrochem. Soc., 169, 040518 (2022).
• 19)  K. Ariyoshi, Y. Fukunishi, and Y. Yamada, J. Electrochem. Soc., 166, A3314 (2019).
• 20)  K. Ariyoshi, T. Toda, and Y. Yamada, J. Electroanal. Chem., 864, 114110 (2020).
• 21)  K. Ariyoshi, A. Maekawa, and T. Nakamura, J. Electroanal. Chem., 916, 116383 (2022).
• 22)  K. Ariyoshi, Y. Fukunishi, and Y. Yamada, Electrochemistry, 87, 188 (2019).
• 23)  M. Ue, K. Ida, and S. Mori, J. Electrochem. Soc., 141, 2989 (1994).
• 24)  R. Imhof and P. Novák, J. Electrochem. Soc., 146, 1702 (1999).
• 25)  K. Xu, M. S. Ding, and T. R. Jow, J. Electrochem. Soc., 148, A267 (2001).
• 26)  M. Ue, A. Murakami, and S. Nakamura, J. Electrochem. Soc., 149, A1572 (2002).
• 27)  J. Kasnatscheew, B. Streipert, S. Roser, R. Wagner, I. Cekic Laskovic, and M. Winter, Phys. Chem. Chem. Phys., 19, 16078 (2017).
• 28)  T. Ohzuku, M. Imazaki, N. Tsukamoto, and K. Ariyoshi, Chem. Lett., 38, 1202 (2009).
• 29)  M. Imazaki, K. Ariyoshi, and T. Ohzuku, J. Electrochem. Soc., 156, A780 (2009).
• 30)  M. Imazaki, L. Wang, T. Kawai, K. Ariyoshi, and T. Ohzuku, Electrochim. Acta, 56, 4576 (2011).
• 31)  T. Ohzuku, A. Ueda, and N. Yamamoto, J. Electrochem. Soc., 142, 1431 (1995).
• 32)  K. Ariyoshi, R. Yamato, and T. Ohzuku, Electrochim. Acta, 51, 1125 (2005).
• 33)  T. Doi, Y. Iriyama, T. Abe, and Z. Ogumi, Chem. Mater., 17, 1580 (2005).
• 34)  M. Wagemaker, D. R. Simon, E. M. Kelder, J. Schoonman, C. Ringpfeil, U. Haake, D. Lützenkirchen-Hecht, R. Frahm, and F. M. Mulder, Adv. Mater., 18, 3169 (2006).
• 35)  K. Ariyoshi, E. Iwata, M. Kuniyoshi, H. Wakabayashi, and T. Ohzuku, Electrochem. Solid-State Lett., 9, A557 (2006).
• 36)  K. Ariyoshi, M. Eguchi, and M. Hirooka, J. Electrochem. Soc., 167, 060532 (2020).
• 37)  L. Medenbach, P. Hartmann, J. Janek, T. Stettner, A. Balducci, C. Dirksen, M. Schulz, M. Stelter, and P. Adelhelm, Energy Technol., 8, 1901200 (2020).
• 38)  Y. Han, S. H. Jung, H. Kwak, S. Jun, H. H. Kwak, J. H. Lee, S.-T. Hong, and Y. S. Jung, Adv. Energy Mater., 11, 2100126 (2021).
• 39)  R. Sahore, F. Dogan, and I. D. Bloom, Chem. Mater., 31, 2884 (2019).
• 40)  O. C. Harris, S. E. Lee, C. Lees, and M. Tang, J. Phys. Energy, 2, 032002 (2020).
• 41)  C. Shen, D. Xiong, L. D. Ellis, K. L. Gering, L. Huang, and J. R. Dahn, J. Electrochem. Soc., 164, A3349 (2017).
• 42)  H. Q. Pham, M. T. Nguyen, M. Tarik, M. E. Kazzi, and S. Trabesinger, ChemSusChem, 14, 2461 (2021).
• 43)  S. Klein, L. Haneke, P. Harte, L. Stolz, S. van Wickeren, K. Borzutzki, S. Nowak, M. Winter, T. Placke, and J. Kasnatscheew, ChemElectroChem, 9, e202200469 (2022).
• 44)  J. H. Felix, J.-H. Cheng, S. Hy, J. Rick, F.-M. Wang, and B.-J. Hwang, J. Phys. Chem. C, 117, 22619 (2013).