以第一原理計算 氧化鋯/矽烯/氧化鋯之三明治結構

Abstract

現今技術的發展使得元件越做越小，而傳統使用的矽基金氧半場效電晶體也逐漸接近其微縮極限。因此，尋找合適的替代材料是現階段非常急切的課題。二維材料由單層原子構成，對於現今元件金氧半場效電晶體通道微縮是不錯的選項之一。其中，矽的二維同素異形體矽烯，在現今金氧半場效電晶體製程上不會造成汙染與不相容的問題，還能延用矽製程的經驗。矽烯載子遷移率比所有的塊材半導體高兩個數量級，若應用在元件中，預期能有令人滿意的驅動電流。然而，將矽烯實現在元件中有兩項主要挑戰：矽烯具有與石墨烯一樣的零能隙狄拉克錐能帶輪廓，需要近一步的處理使能隙打開。此外矽烯的π鍵化學活性高，使矽烯接觸空氣後會劣化。 在我的研究中，以第一原理理論計算，看不同應力下矽烯的能隙變化與矽烯夾在介電材料之間時電子結構(尤其是遷移率)受到的變化。在我施加不同應力的矽烯中，能隙打開的最大值為0.17eV。計算特定應力下單層矽烯結構的載子遷移率，其電子與電洞的載子遷移率分別為16.6m2V-1s-1、14.4m2V-1s-1與沒受應力的矽烯載子遷移率同數量級。我選擇半導體元件中常見的高介電常數材料氧化鋯建構氧化鋯/矽烯/氧化鋯的三明治結構。我計算在三明治結構介面處沒氫化與氫化的電子結構，發現前者沒有狄拉克錐，而後者雖具有狄拉克錐卻沒有能隙。我更進一步施加應力在氫化的三明治結構中，使矽烯恢復原本的對稱性，並發現其具有能隙。透過Bader分析這項方法，推測氧化鋯轉移到矽烯的靜電荷，是造成無施加應力氫化三明治結構中矽烯對稱性被破壞的主因。 總結來說，從我的第一原理計算中可得知：矽烯可藉由施加應力打開能隙；在氧化鋯/矽烯/氧化鋯的三明治結構中，必須讓接面氫化以保持矽烯π鍵，並且也要維持矽烯原本的對稱性，如此能隙才會打開。同時我也透過第一原理計算遷移率，發現其仍然維持在10m2V-1s-1的數量級。預期這些計算結果能為開發矽烯電晶體提供初步的方向。

As the state-of-the-art technology in semiconductor industry drives devices toward remarkably tiny dimension, the traditional Si-based MOSFETs will soon approach their scaling limits. An emerging development in this field is to find a replacement for the Si channel. A two-dimensional material consists of only a single atomic layer, being a good candidate of the MOSFET channel upon the ongoing scaling down. A two-dimensional silicon allotrope, silicene has its great advantage in being significantly compatible with the Si-based process. Silicene has a mobility two order of magnitude higher than bulk semiconductors and is expected to maintain a satisfactory device on-current. Yet there are also two major challenges to make it a feasible device. A stress-free stand-alone silicene, like graphene, has a gapless Dirac-cone band dispersion and needs additional treatments to open its gap. Besides, the silicene π bonds are chemically active, making it unstable in the air. In this thesis, I perform first-principles calculations to look for the strain-induced energy gap of silicene and how its electronic structures, especially the mobility, are affected when being sandwiched between dielectric materials. I find the energy gap is opened up to 0.17eV among the strains I have applied. We also calculate the electron and hole mobility of silicene under a particular strain to be 16.6 and 14.4m2V-1s-1, respectively, preserving the same order of magnitude of the free-standing mobility. I choose ZrO2/silicene/ZrO2 to be my model sandwich structure, where ZrO2 is a widely used high-k material in conventional semiconductor devices. I have considered both unpassivated and H-passivated interfaces of the above sandwich structure, and find that the former has no Dirac cone while the latter contains a Dirac cone and have indirect gap closing. I further apply a strain to the silicene in the H-passivated case such that the silicene restores its stand-alone symmetry, and find that the gap is re-opened. The symmetry-breaking mechanism of the unstrained (relaxed) silicene in a H-passivated sandwich is very likely due to the electrostatic charge that is transferred from ZrO2 to silicene, as calculated by the Bader analysis. In summary, my first-principles calculations show that silicene can have strain-induced gaps, and the ZrO2/silicene/ZrO2 sandwich structure needs both the H passivation (preserving the π bonds) and the silicene-symmetry restoration to open the gap. The mobility is also calculated from first principles and are found to maintain a high value in the order 10m2V-1s-1. These results provide preliminary guidance to develop silicene-based transistors.