Evolving affinity between Coulombic reversibility and hysteretic phase transformations in nano-structured silicon-based lithium-ion batteries

Nano-structured silicon is an attractive alternative anode material to conventional graphite in lithium-ion batteries. However, the anode designs with higher silicon concentrations remain to be commercialized despite recent remarkable progress. One of the most critical issues is the fundamental understanding of the lithium–silicon Coulombic efficiency. Particularly, this is the key to resolve subtle yet accumulatively significant alterations of Coulombic efficiency by various paths of lithium–silicon processes over cycles. Here, we provide quantitative and qualitative insight into how the irreversible behaviors are altered by the processes under amorphous volume changes and hysteretic amorphous–crystalline phase transformations. Repeated latter transformations over cycles, typically featured as a degradation factor, can govern the reversibility behaviors, improving the irreversibility and eventually minimizing cumulative irreversible lithium consumption. This is clearly different from repeated amorphous volume changes with different lithiation depths. The mechanism behind the correlations is elucidated by electrochemical and structural probing.


Supplementary Fig. 5
Li/Li + potential in 2032-type coin half-cells as a function of the specific capacity, for type-A (left two lines) and -B electrodes (right two lines) at the cycle previous to the probing point (surrounded by blue rectangle) and the probing points (surrounded by red rectangle) . Note that at the probing points (inserted every 20 cycles), all the electrodes are cycled at 0.1 C regardless of the depth of discharge (DOD) cycling history while at the other cycling points cycled at 1 C under the given DOD controls. a-e The cycles previous to the probing points for type-A electrodes, (f-j) the probing points for type-A electrodes, (k-o) the cycles previous to the probing points for type-B electrodes, and (p-t) the probing points for type-B electrodes. Black, pink, blue, and green profiles correspond to cycling under DOD100%, 90%, 80%, and 70%, respectively.

Supplementary Fig. 6
Dependence of Li-Si processes on current rates at probing points cycled under depth of discharge (DOD)100% with different DOD history for type-A electrodes in 2032-type coin half-cells; Evolution of Si#c-1-3 processes on delithiation at the probing points is shown by dQ/dV profiles.
Note that at the probing points (inserted every 20 cycles), all the electrodes are cycled at slower current rates regardless of the DOD cycling history while at the other cycling points cycled at 1 C under the given DOD controls. The Li-Si processes (Si#d-X/Si#c-X represents the X th Li-Si discharge/charge process) are summarized in Fig. 3a and Supplementary Table 5 (also see Methods > 'Reference electrochemistry'). At the probing points, three lower current rates (0.02, 0.05, and 0.1 C) are used consecutively to observe the rate dependence of the processes, shown as blue, green, and red profiles, respectively.

Supplementary Fig. 7
Dependence of Li-Si processes on current rates at probing points cycled under depth of discharge (DOD)100% with different DOD history for type-B electrodes in 2032-type coin half-cells; Evolution of Si#c-1-3 processes on delithiation at the probing points is shown by dQ/dV profiles.
Note that at the probing points (inserted every 20 cycles), all the electrodes are cycled at slower current rates regardless of the DOD cycling history while at the other cycling points cycled at 1 C under the given DOD controls. The Li-Si processes (Si#d-X/Si#c-X represents the X th Li-Si discharge/charge process) are summarized in Fig. 3a and Supplementary Table 5 (also see Methods > 'Reference electrochemistry'). At the probing points, three lower current rates (0.02, 0.05, and 0.1 C) are used consecutively to observe the rate dependence of the processes, shown as blue, green, and red profiles, respectively.

Supplementary Fig. 8
Electrochemical impedance spectroscopy (EIS) data for fully lithiated type-A electrodes in 2032type coin half-cells cycled under depth of discharge (DOD)80-100% over ~150 cycles. The cell is cycled at 1 C under the subjected DOD controls until it reaches the cycle just prior to the probing point (inserted every 20 cycles). Then, the cell is cycled at 0.05 C, holding the voltage at 10 mV on lithiation for at least 24 h until the residual current becomes less than 0.001 C. The EIS frequency is swept from 1 MHz to 0.1 Hz with a fluctuating voltage of ±5 mV. a-c Impedance component of a real part plotted over an imaginary part for DOD100, 90, and 80%, respectively. d-f Measurement frequency plotted over impedance component of a real part for DOD100, 90, and 80%. These profiles under DOD100, 90, and 80% are encircled by black, pink, and blue rectangles, respectively. Red, green, purple, pink, orange, and blue lines correspond to the profiles at the 23 rd , 44 th , 65 th , 86 th , 107 th , and 146 th cycle, respectively.

Supplementary Fig. 9
Electrochemical impedance spectroscopy (EIS) data for fully lithiated type-A electrodes in 2032type symmetric cells cycled under depth of discharge (DOD) 80-100%. Two identical cells are cycled at 1 C under the subjected DOD protocols, until they reach the cycle just prior to the probing point (inserted every 20 cycles). Then, the cell is cycled at 0.05 C, holding the voltage at 10 mV on lithiation for at least 24 h until the residual current becomes less than 0.001 C. The two coin cells are then disassembled in Ar-filled glovebox and re-assembled into one symmetric coin cell with new electrolyte and separator. The EIS frequency is swept from 1 MHz to 0.1 Hz with a fluctuating voltage of ±5 mV. a-c Impedance component of a real part plotted over an imaginary part for DOD100, 90, and 80%, respectively. d-f Measurement frequency plotted over impedance component of a real part for DOD100, 90, and 80%. These profiles under DOD100, 90, and 80% are encircled by black, pink, and blue, respectively. Red, green, and blue lines correspond to the profiles at the 23 rd , 65 th , and 107 th cycle, respectively.

Supplementary Fig. 10
Examination of Li-metal resistivity effect in half coin-cells, accumulated over cycles, on Li-Si electrochemical processes. a Schematics of the experimental flow; Type-A electrode is cycled in 2032-type coin half-cell for 107 cycles under DOD100% protocol. Afterwards, the cell is disassembled in Ar-filled glovebox and reassembled with fresh, polished Li-metal with new electrolyte. b dQ/dV profiles on delithiation of type-A electrodes at 0.05 C (green) and 0.02 C (blue) in the reassembled 2032-type half-cell. Green and blue profiles correspond to cycling at 0.05 and 0.02 C, respectively. The sharp peak at 430 mV (Si#c-3, a signature of c-Li 3.75 Si decomposition into a-Li <~1.1 Si) is absent, and instead the Li-Si processes are dominated by Si#c-2 and Si#c-4. For definition of the Li-Si processes see Fig. 3a (also see Methods > 'Reference electrochemistry').

Supplementary Fig. 11
Effect of electrode/electrolyte exposure time at constant voltage (CV) domain (10 mV) on Coulombic efficiency (CE) for type-A electrodes. Li/Li + potential in 2032-type coin half-cells as a function of (a, c) specific capacity (mAh g -1 ) and (b, d) cycling time in the 10 th cycle at a current rates of (a, b) 1 C and (c, d) 0.1 C. The total duration of lithiation is not proportional to the depth of discharge (DOD) percentages at 1 C, due to the nature of the constant current constant voltage (CCCV) mode, while they are proportional to each other at 0.1 C. e Specific capacity and (f) CE cycled at 0.1 C with different DOD controls. g CE as a function of specific capacity cycled at 0.1 C under different DOD controls. Black solid line corresponds to DOD100%, and pink, blue, and green profiles/dots correspond to DOD90%, 80%, and 70%, respectively. CE error bars are all within a range of ±0.1% and omitted in the figure.

Supplementary Fig. 12
Effect of electrode/electrolyte exposure time at constant voltage (CV) domain (10 mV) on Coulombic efficiency (CE) for type-B electrodes. Li/Li + potential in 2032-type coin half-cells as a function of (a, c) specific capacity (mAh g -1 ) and (b, d) cycling time in the 10 th cycle at a current rate of (a, b) 1 C and (c, d) 0.1 C. The total duration of lithiation is not proportional to the depth of discharge (DOD) percentages at 1 C, due to the nature of the constant current constant voltage (CCCV) mode, while they are proportional to each other at 0.1 C. e Specific capacity and (f) CE cycled at 0.1 C with different DOD controls. g CE as a function of specific capacity cycled at 0.1 C under different DOD controls. Black solid line corresponds to DOD100%, and pink and blue profiles/dots correspond to DOD90% and 80%, respectively. CE error bars are all within a range of ±0.1% and omitted in the figure.

Supplementary Fig. 13
Electrochemical reversibility outputs for type-B electrode. a Specific capacity and (b)   Si-2p, C-1s, Li-1s, P-2p, O-1s, and F-1s XPS spectra with different ion beam sputtering times for type-A electrodes before cycling (the 1 st row), after the amorphization of polycrystalline Si nanoparticles (the 2 nd row), and delithiated electrodes after the 44 th and 107 th cycles under depth of discharge (DOD) 80-100% (the 3 rd -8 th rows). Prior to the observation, the coin cells are fully delithiated, holding the potential at 1.5 V until the current decays to less than 0.001 C. The electrodes are then washed using dimethyl carbonate (DMC) in Ar-filled glovebox, vacuum dried, and transferred to in-house XPS holder without exposure to the ambient air (see details in Methods > 'XPS').

Supplementary Fig. 21
Ex situ measured volume expansion rate for fully delithiated type-A electrodes at the probing points cycled under depth of discharge (DOD)80-100% controls; the value is tabulated from electrode thickness before/after the cycles. Black, pink, and blue dots correspond to the expansion rate under DOD100%, 90%, and 80% cycling protocols, respectively. The electrode thickness is measured by a commercial micrometer (Mitsutoyo) with an accuracy of 1 μm.

Supplementary Fig. 22
X-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure

Supplementary Table 1
A list of electrochemical control parameters (brown), electrochemical outputs (blue), and structural analysis methods (purple) used in this study. The corresponding figures are indicated in the right column. The experimental scheme based on the following list is also shown in Fig. 1. The repeated a-c phase transformations accelerate the alteration of Li-Si electrochemical pathways from asymmetric to symmetric reaction sequences. In the former, c-Li 3.75(+δ) Si formed around ~50 mV asymmetrically transforms back into a-Li <~1.1 Si on delithiation, with the well-known large hysteresis at 430 mV. In the latter, c-Li 3.75(+δ) Si can reform a-Li~3 .2-3.75 Si at as low as ~150 mV via a newly elucidated process, followed by the subsequent symmetric a-a transformations at 300 mV (a-Li~3 .5 Sia-Li~2 .0 Si) and 550 mV (a-Li~2 .0 Sia-Si). Repeating the a-c transformations under depth of discharge (DOD)100%, the regime shift occurs at around the 60 th cycle, while it is postponed to after the 107 th cycle when cycled under the lower DOD%.

CE dQ/dV 2 Quantitative and qualitative characterization of the evolving CE alterations by different Li-Si structural changes
Repeating the c-Li 3.75(+δ) Si formation/decomposition under DOD100%, typically featured as a capacity degradation factor, can increase CE up to ~99.9% in the most efficient manner among the given DOD controls, and minimize the cumulative irreversible Li consumption after incurring a certain amount of sacrificial Li consumption and capacity loss. This is quantitatively distinguished from that by mere amorphous Li-Si volume changes under DOD70-90%.

Clarifying inherent nature of CE behaviours in different electrochemical Li-Si process regimes
When Si surface is fully exposed to the electrolyte, Coulombic efficiency (CE) has a strong correlation with the associated electrochemical regimes in the electrodes and the DOD controls. The asymmetric-to-symmetric shift can alter the susceptibility of CE. In the asymmetric regime, CE is more susceptible to the presence/absence of c-Li 3.75(+δ) Si upon (de)lithiation by the DOD controls. In contrast, in the symmetric regime, CE is more stabilized/saturated and higher in value regardless of the presence.

CE dQ/dV
Supplementary Table 4 Physical parameters and composition ratios of materials in type-A and -B secondary particles.