相場模式在多晶矽生長行為之研究

Abstract

在光電產業中，多晶矽(poly-Si) 被廣泛地使用在製作太陽能電池的材料上，然而，由於多晶物質裡的晶界密度(grain boundary density) 將會影響太陽能電池的能量儲蓄的效率，當密度越高時越不利於能源的儲存。因此過多的晶界是不被期望的，如何藉由減少多晶塊材上的晶界分布來提升電池的效率也就成為了一個重要的課題。一般而言，控制晶體生長的環境藉以改良材料品質是最常使用的方式，而電腦模擬則提供了一個相當容易的方法來探討控制晶體生長的因素。在本論文中，我們使用由Warren等人所發展出的多晶相相場模式(polycrystalline phase field model)並結合由Eggleston所提出模擬小面生長(faceted growth)的相場模式，首次用以探討多晶矽在不同過冷度下的熔體生長行為。根據Fujiwara等人所進行的矽晶粒競爭生長實驗，當過冷度很小時，生長出的多晶矽中，(111)晶向的矽晶占了大多數；然而在過冷度較大時，(100)晶向的矽晶則為多數。這其中，晶粒選擇的機制扮演了很重要的角色。 根據Fujiwara等人的解釋，低過冷時晶粒選擇的機制主要取決於界面能效應；而在高過冷方面，根據Atwater等人的解釋，則由動力學效應主導。在我們的相場模擬結果上，我們能夠得到與實驗上相似的生長行為與界面形狀。此外，由我們在定性上的分析得知，晶粒選擇的機制主要是由界面能效應與動力學效應兩種不同的效應競爭的結果，過冷度則為調整動力學效應的控制變因。這些結論與Fujiwara和Atwater在解釋矽晶粒的選擇機制上，具有良好的一致性。

In photovoltaic industry, polycrystalline silicon (poly-Si) is widely used as the material to manufacture the solar cell. However, the energy conservation of the solar cell is usually affected by the grain boundary density in the poly-Si due to the grain boundary will act as the photo-carrier in the cell. Thus, to decrease the grain boundary density is crucial to produce good quality poly-Si for solar cell. In general, the crystal growth environment will affect the quality of the crystal, and the computational simulation is used for studying the growth dynamics of the crystal. In this thesis, we combine the polycrystalline phase field model proposed by Warren et al., with the highly anisotropic surface energy phase field model proposed by Eggleston et al. to study the growth behaviors of the polycrystalline silicon under different undercooling of the melt. According to the experimental observations from Fujiwara et al, when the undercooling is low enough, the (111) silicon will be the dominant orientation in the polycrystalline silicon. As the undercooling becomes much higher, the (100) will be the dominant orientation. To explain the behaviors, the mechanism of the grain selection between grains is the key to elucidate the whole problem. According to Fujiwara et al., when the undercooling is low enough, the effect of the surface energy will dominate the grain selection mechanism. For the higher undercooling case, Atwater et al. explained that the kinetics effect will dominate the growth. In our simulations, we can reproduce the similar growth behaviors and morphologies from the experiments. Beside, from our qualitative analysis of the growth behaviors, the mechanism of the grain selection is mainly caused by the competition between the effect of the surface energy and the kinetic effect, and the undercooling serves as a factor to control the relative strength of the kinetic effect. In explaining the competitive growth behaviors of silicon, our conclusion shows good agreement with the explanations by Fujiwara et al. and Atwater et al.