Silicon has received much attention as a next-generation anode material for lithium-ion battery (LIB) because it can give a theoretical capacity of up to 4000 mAh/g. Typically, silicon electrodes show poor cycle performance, which is attributed to large volume expansion during lithiation, pulverization of the active material, continuous solid electrolyte interphase (SEI) layer growth etc. Though, there are hardly any studies that can tell how much of the measured capacity comes from each process. The aim of this study is to establish a method to understand and quantify the contributions to the reversible and irreversible capacities of silicon anode in order to develop strategies to improve the stability of the electrodes. Here, we report the use of an electrochemical approach – depth of discharge test – to separate the charge-discharge capacities of crystalline silicon electrodes into four contributions: (1) SEI formation and electrolyte decomposition, (2) lithium accommodation in carbon and binder, (3) lithiation and delithiation of the Si active material, and (4) capacity loss associated with particle cracking and detachment.

Silicon electrodes using bulk silicon particles (Sigma Aldrich), acetylene black (AB) and carboxymethyl cellulose (CMC) binder were made with a ratio of 4:3:3, 6:2:2 and 8:1:1. The electrodes are assembled into 2032-type coin cells with Celgard separator and Li metal as the counter electrode. 1M LiPF₆ in fluorinated ethylene carbonate (FEC) and diethyl carbonate (DEC) = 1:1 by volume is used as the electrolyte. For the depth of discharge test, multiple cells were made, and each of them is initially discharged to different cutoff capacity (250, 500, 1000, 1500, 2000, 2500, 3000 and 3500mAh/g) and charged back to 1V.

Figure 1a shows a summary of the initial discharge-charge curves of different cells with a composition of Si:AB:CMC = 4:3:3. The respective first charge capacity vs. first discharge capacity is summarized in Figure 1b, which shows a linear behavior that is shifted from the origin. The slope of the line corresponds to the intrinsic first cycle efficiency of Si, which is about 90% and is independent on the degree of lithiation. This suggests that no matter how much lithium is initially inserted, 10% of them remains in the lattice and cannot be removed. This is attributed to partial irreversibility of the crystalline-to-amorphous transition of Si during lithiation, and is consistent with a core-shell model of lithiation during first discharge. The x-intercept of Figure 1b is non-zero, which corresponds to the irreversible capacity of the electrode. The x-intercept depends on the composition of the electrode (see inset of Figure 1b) – larger x-intercept for large amount of carbon and binder, which is due to irreversibility from these two components. Accounting for all of these, we estimate the capacity from SEI formation and electrolyte decomposition to be about 16 mAh/(g Si) for bulk Si (with BET = 1.6 m²/g).

SEI formation is expected to depend on particle size and surface area. We verify this by performing the depth of discharge test with ballmilled Si (in Ar at 200rpm for 24 hours) with BET = 6.6 m²/g. The capacity due to SEI formation during 1^st discharge is increased to about 68 mAh/(g Si), and scales with BET surface area. We estimate that the SEI formation in our system to be about 10 mAh per square metres of active material surface. This value would depend on the type of electrolyte used in the system.

Mechanical issues and particle isolations are observed in fully discharged electrode when the amount of binder is less than 20%. This is represented in the bend in the depth of discharge curve for the CMC-811 electrode, as in Figure 1b. Our results suggest that there is a threshold to mechanical breakdown. In the case of the CMC-811, high reversibility is still obtained when the discharge capacity is below 2000 mAh/g. With larger discharge capacity, volume expansion is larger, leading to severe capacity fading.

In subsequent cycles, cycle stability is governed by the coulombic efficiency (CE) of the electrode - the larger the coulombic efficiency, the better the stability. The irreversible capacity is not only due to continuous SEI formation, but also to Li trapping in the active material and mechanical breakdown of the electrode. CE can be increased by limiting capacity, reducing particle size, or changing to a stronger binder such as polyimide. Some of results can be found in our recent publication¹. More results will be shown during the meeting.

[1] P.-K. Lee, Y. Li, D. Y. W. Yu, J. Electrochem. Soc. 164, A6206 (2017).