Thesis presented December 22, 2017

Abstract:
Over the last 25 years, the performance increase of lithium-ion batteries has been largely driven by the optimization of inactive components. With today’s environmental concerns, the pressure for more cost-effective and energy-dense batteries is enormous and new active materials should be developed to meet those challenges. Silicon’s great theoretical capacity makes it a promising candidate to replace graphite in negative electrodes in the mid-term. So far, Si-based electrodes have however suffered from the colossal volume changes silicon undergoes through its alloying reaction with Li. Si particles will be disconnected from the electrode’s percolating network and the solid electrolyte interface (SEI) continuously grows, causing poor capacity retention. A thorough understanding of both these phenomena, down to the scale of a single silicon nanoparticle (SiNP), is critical to the rational engineering of efficient Si-based electrodes. To this effect, we have developed STEM-EELS into a powerful and versatile toolbox for the study of sensitive materials and heterogeneous systems. Using the low-loss part of the EEL spectrum allows us to overcome the classical limitations of the technique. 

This is put to use to elucidate the first lithiation mechanism of crystalline SiNPs, revealing Li1.5Si @ Si core-shells which greatly differs from that of microparticles, and propose a comprehensive model to explain this size effect. The implications of that model regarding the stress that develops in the crystalline core of SiNPs are then challenged via stress measurements at the particle scale (nanobeam precession electron diffraction) for the first time, and reveal enormous compressions in excess of 4±2 GPa. Regarding the SEI, the phase-mapping capabilities of STEM-EELS are leveraged to outline the morphology of inorganic and organic components. We show that the latter contracts during electrode discharge in what is referred to as SEI breathing. As electrodes age, disconnection causes a diminishing number of SiNPs to bear the full capacity of the electrode. Over-lithiated particles will in turn suffer from larger volumes changes and cause further disconnection in a self-reinforcing detrimental effect. Under extreme conditions, we show that SiNPs even spontaneously turn into a network of thin silicon filaments. Thus an increased active surface will compound the reduction of the electrolyte and the accumulation of the SEI. This can be quantified by summing and averaging STEM-EELS data on 1104 particles. In half-cells, the SEI volume is shown to increase 4-fold after 100 cycles without significant changes in its composition, whereas in full cells the limited lithiation performance understandably leads to a mere 2-fold growth. In addition, as the operating potential of the silicon electrodes increases in full cells – potential slippage – organic products in the SEI switch from being carbonate-rich to oligomer-rich. Finally, we regroup these findings into an extensive aging model of our own, based on both local STEM-EELS analyses and the macro-scale gradients we derived from them as a whole.

Keywords:
Li-ion batteries, in situ TEM, Silicon, EELS

On-line thesis.