原子尺度下的鋰離子電池矽基負極鋰化行為 以及 LAGP/PEO 固態電解質之鋰離子擴散機制

Abstract

鋰離子電池為現今重要的能源儲存裝置，其中能量密度以及安全性都是重要的議題。為了改善這些問題，人們發展了矽基負極與固態電解質，但目前它們都有一些問題因此無法用於實際電池中。矽負極在充電的過程中形成鋰矽合金，稱為鋰化(lithiation)。結晶相或非晶相的矽都可以作為矽基負極。實驗觀察到結晶矽會有不等向鋰化的行為(anisotropic lithiation)，而導致鋰化較慢的面上產生的鋰矽合金出現應力集中而破裂。除此之外，結晶矽的鋰化速度會隨著時間降低，稱為自減速鋰化(self-limiting lithiation)。非晶矽的鋰化行為和結晶矽有一些差異，非晶矽的鋰化為兩階段鋰化，在第一階段形成有邊界的a-Li2.5Si (c-Si/a-Li2.5Si)，而在第二階段形成沒有邊界的c-Li15Si4。更重要的是，結晶矽所形成的鋰矽合金比非晶矽的更容易破裂，也就是說結晶矽的破裂臨界半徑(critical size of fracture)小於非晶矽。這些現象發生的機制尚不清楚。本論文的第一部份我們用平板狀的矽研究不等向鋰化，我們重新發展ReaxFF Li-Si的參數以解決先前參數的問題，接著我們使用分子動力學(molecular dynamics, MD)模擬找出Si(100)、Si(110)、Si(111)與Si(112)面的鋰化行為差異。我們成功重現實驗上觀察到的不等向鋰化行為，並發現它是動力學上的能量障礙差異所導致的結果。論文的第二部份我們同樣以分子動力學MD計算結晶矽奈米線與奈米顆粒的鋰化機制與自減速行為，我們發現這些奈米結構曲率會稍微影響鋰化行為，而自減速的行為的起因是平面法向量壓應力阻礙矽的斷鍵。論文的第三部份，我們以分子動力學MD計算非晶矽的鋰化，我們重現了二階段鋰化並和結晶矽的行為比較，整體的鋰化反應可以視為一個動力學上的反應-擴散系統，而矽的斷鍵速率直接導致了非晶矽的二階段鋰化以及更大的破裂臨界半徑。 LAGP固態電解質主要面臨的問題是較低的鋰擴散係數，它藉由在LGP固態電解質中摻雜鋁(Al)而形成。研究發現在LGP中摻雜較大顆的元素例如錫(Sn)可以增加鋰擴散係數，但是在LAGP中做同樣摻雜卻可能降低鋰擴散係數。論文的第四部份，我們以分子動力學與彈性帶方法(nudged elastic band, NEB)研究LGP/LGSP/LSP與LAGP/LAGSP/LASP中鋰的擴散能量障礙，我們發現到錫(Sn)在LAGP的摻雜穩定了擴散的起始與最終穩定狀態，因而提高鋰擴散的能量障礙。PEO (polyethylene oxide)高分子電解質相對於陶瓷的LAGP有較好的延展性，但它卻比LAGP有更低的鋰離子擴散係數。研究發現在PEO中混合LAGP可以略為增加鋰擴散係數，但人們尚未理解這部份的機制。論文的第五部份，我們以分子動力學模擬LAGP/PEO的界面並與純PEO比較，我們發現稍微遠離LAGP/PEO界面的PEO顯示些微的鋰擴散係數提升，這部份的現象可能由於LAGP/PEO界面限縮高分子鏈在法向量方向運動，而加速高分子鏈在切面方向的共同運動有關。

Li-ion batteries are important energy storage materials nowadays. The specific capacity of energy and safety of Li-ion batteries are always an important topic. Silicon anodes and solid-state electrolytes (SSEs) are possible candidates to solve these problems. However, there are still many problems unsolved in these materials. Silicon anode stores lithium ion by forming Li-Si alloy during its charging process, called the lithiation of silicon anode. Both crystalline and amorphous silicon can be used as silicon anode. The lithiation is anisotropic for crystalline silicon (c-Si), which concentrates stress on Li-Si alloy generated by slower-lithiated facets. Besides, crystalline silicon lithiates is slowered down when lithiation proceeds, called the self-limiting lithiation. The lithiation behavior is very different for amorphous silicon anode than crystalline silicon. Amorphous silicon undergoes two-stage lithiation, which lithiates to a-Li2.5Si with a clear boundary (c-Si/a-Li2.5Si), and then lithiate to c-Li15Si4 without a clear boundary. More importantly, the Li-Si alloy formed by crystalline silicon cracks much more easily than the one by amorphous silicon. That is, crystalline silicon has a much smaller critcal size for fracturing than the amorphous one. The mechanisms for these phenomenon for silicon anode is not fully understood. In the first part of this work, we focued on the anisotropic lithiation of crystalline silicon slab. We developed a ReaxFF parameter to correct some failures in previous ReaxFF parameter. After that, a molecular dynamics (MD) simulation is performed to find out the lithiation behavior of Si(100), Si(110), Si(111), and Si(112) slabs. We successfully reproduced the anisotropic lithiation and concluded that it is caused by the Li insertion energy barrier difference between each facet. In the second part, we performed MD simulation on crystalline silicon nanowire/nanoparticle to clarify its microscopic behavior and self-limiting lithiation. We found that the lithiation behavior is altered by the curvature, and the self-limiting lithiation is caused by a suppression of silicon-silicon bond-breaking reaction by normal compressive stress. In the third part, we performed similar MD simulation on amorphous silicon nanowire/nanoparticle. We reproduced the two-stage lithiation of amorphous silicon, and compared the microscopic lithiation between crystalline and amorphous silicon. The overall reaction is identifed as a reaction-diffusion system, and the silicon-silicon bond-breaking rate difference directly leads to the two-stage lithiation and larger critical size of amorphous silicon.

Solid-state electrolyte has the problem of lower Li diffusivity when comparing with liquid electrolyte. LAGP is a kind of solid-state electrolyte formed by doping Al in LGP. Researchers found that doping a larger ion such as Sn into LGP as LSP increases the Li diffusivity, by increasing bottleneck size for Li diffusion. However, doping Sn does not increase and even hamper the Li diffusivity of LAGP. In the fourth part. we performed MD and nudged-elastic band (NEB) analysis on LGP/LGSP/LSP and LAGP/LAGSP/LASP. We find out that Sn doped LAGP increases the stability of stable sites and increase the Li diffusion energy barrier. The polyethlyene oxide (PEO) is a solid polymer electrolyte (SPE) which is more flexible than ceramic LAGP electrolyte. However, it suffers from more serious Li diffusivity problem. Researchers found that mixing LAGP ceramic electrolytes into PEO slightly increases the Li mobility, but its reason is unknown. In the final topic, we perfomed a MD simulation on LAGP/PEO interface and sampled the Li diffusivity. We found a slight Li diffusivity enhance at some distance farther from the interface. The phenomenon is related to the enhance of cooperative polymer chain mobility in tangential directions, due to the normal confinement of the interface.