N+-P鍺淺接面暨接觸電阻研究

Abstract

透過微縮來提升矽基電晶體效能將會因為物理極限而碰到瓶頸。因此，高遷移率材料或者新的元件結構成為近年來熱門的研究主題。鍺擁有比矽更高的電子以及電洞遷移率而被視為下一世代的材料。然而因為有很多的介面能態集中在鍺的價帶，導致費米能階會被鎖定在價帶附近，讓金屬與N型鍺的蕭特基能障有一個幾近鍺能隙的量值，造成很高的蕭特基能障進而導致高接觸阻抗。同時因為N型摻雜物在鍺中很容易擴散，也不容易形成一個淺接面來避免短通道效應。 本篇論文一開始先採用傳統鋁接觸的N+-P接面製程，過低的植入能量或植入劑量將導致過淺的接面深度以及過淡的N+區域會使得漏電流偏大，接面特性不佳。接著將鋁接觸置換成鎳鍺化物接觸，並利用摻雜堆積技術來改善接面。摻雜堆積技術主要可以分成形成鎳鍺化物之前摻雜或者形成鎳鍺化物之後摻雜。當採用形成鎳鍺化物之前摻雜製程完成接面時，由於鎳原子在鍺中的擴散係數高，過淺的接面會使得鎳原子容易到達空乏區，造成很大的漏電流。接著我們也探討了鎳鍺化物的熱穩定性，我們發現接面深度淺於100奈米(82.06nm)的鎳鍺化物的接面大約只能再承受450℃持續30秒或者500℃持續10秒的熱製程，若再持續將溫度上升或者時間拉長，接面將會產生由於鎳原子擴散並和逆偏電壓相關的激發電流。最後，我們再加上形成鎳鍺化物之後離子植入的製程以增加鎳鍺化物及鍺接面的濃度以降低接觸阻抗，大約可以降低平均值至3×10^(-6)Ω-cm^2，相較於單純只有鎳鍺化物之前離子植入接面的接觸阻抗平均值為2×10^(-5)Ω-cm^2，效果相當顯著，同時接面也保持在10^-3 (A/cm^2)的低漏電流特性，不會因為再進行一次離子植入而產生額外的漏電。 另一方面，本論文也探討了植入高劑量的鋁、硒、錫對鎳鍺化物跟鍺的蕭特基能障的影響，由第一原理模擬計算得知硒對鎳鍺化物N型鍺的蕭特基能障可降低最多的量值、鋁幾乎不改變，而錫則會提高能障量值。實作蕭特基二極體後確實發現硒對鎳鍺化物N型鍺的蕭特基能障約可降低0.15 eV左右。 簡言之，本篇論文主要採用在形成鎳鍺化物的先後各一次的離子植入來形成一個低漏電且低接觸阻抗的接面，相較於其他已發表的降低接觸阻抗論文相比，我們的製程相對簡單、量測採用的阻抗測試結構更為精準，若能應用在N型鍺金氧半場效電晶體上，預期可看到顯著的性能提升。

Scaling down geometry to enhance transistor’s performance will encounter difficulty due to physical limitation. Therefore, high mobility materials and novel device structures become the emerging research topic. Germanium (Ge) has higher electron and hole bulk mobility than Silicon (Si) so that it is considered as the material for the next generation. However, because there are a lot of interface states distributing near the valence band of germanium, Fermi level would be pinned near valence band. It causes a high Schottky barrier height value nearly about a bandgap of germanium and a large contact resistance for a metal/n-Ge Schottky contact. Moreover, due to the high diffusivity of n-type dopants in Ge, it is hard to form a shallow junction to suppress short channel effect. In this thesis, we fabricate conventional Al-contacted N+-P junction at first. Low ion implantation energy or dose cause a very shallow junction but the leakage current increases. The junction characteristics is not good. Then, the Al-contact is replaced by NiGe-contact and dopant segregation is used to improve junction characteristics. There are two types of dopant segregation process. One is implantation before germanidation (IBG), the other one is implantation after germanidation (IAG). For the IBG process, due to high diffusivity of Ni in Ge, it would make nickel diffuse to the depletion region and cause a high leakage current if junction depth is too shallow. The NiGe-contacted junction can tolerate at 450℃ for 30 sec or 500℃ for 10sec at most as the junction depth becomes shallower than 100 nm (82.06nm). If increasing the temperature or extending the annealing time, the junction would exhibit voltage dependence generation leakage current due to Ni diffusion. Finally, we use the IBG+IAG process to enhance the doping concentration at the NiGe/Ge interface and lower the specific contact resistance to 3×10^(-6)Ω-cm^2 for the mean value. The specific contact resistance of the pure IBG process is 2×10^(-5) Ω-cm^2 for the mean value. Besides, the additional IAG process would not induce more leakage current. The leakage current for the IBG+IAG process keeps as low as that of the IBG junction. This thesis also demonstrates the effect of high dose implantation of Aluminum (Al), Selenium (Se), and Stannum (Sn) on the Schottky barrier height. It is predicted by first-principles calculations that the Se atom can lower the Schottky barrier height of NiGe/Ge the most, the Al atom is hardly to modify the SBH, and the Sn atom can raise the SBH. For the implementation of the Schottky junctions, the Se ion implantation can lower the SBH about 0.15 eV. In brief, this thesis focus on the IBG+IAG process to form a low leakage current and low contact resistance junction. Compared with other studies about contact resistance issue, our process is easier and our test structure of the contact resistance is more accurate. Employing Se ion implantation is expected to further reduce the contact resistance.