鎳鍺化物接觸之N+-P鍺淺接面及接觸電阻之研究

Abstract

在過去幾十年內，提升矽半導體元件的特性的研究已經越來越完整。但是隨著元件快速的微縮，以矽做為半導體材料的金氧半場效電晶體很快就會因為物理極限的限制而越來越難繼續微縮下去。因此，當前的課題就是發展其他可能方法來解決這個問題。鍺由於本身的較高的載子遷移率以及和矽製程有較大的相容性而被視為能取代矽基板做為未來半導體的材料。但是N型參雜在鍺基板裡面會快速地擴散，所以好的鍺淺接面特性不容易達成，此外，由於N型鍺與金屬界面上，費米能階會被鎖定在接近價帶的位置，會產生很大的蕭基位能障而導致較大的接觸組抗。本論文主要探討利用鎳鍺化物接面來形成較淺的接面，並同時降低對N型鍺的接觸阻抗。 由於鎳鍺化物的電阻係數較其他金屬鍺化物來的低，且形成溫度也相對較低，所以選用鎳鍺化物當作接面金屬。此外，本論文中使用了兩種製程來研究接面特性，其一是形成鍺化物之前打入參雜使參雜的離子堆積在界面，另一種是形成鍺化物之後打入參雜，接著再退火使參雜離子能夠在界面離析出來。 經過形成鍺化物之前打入參雜的製程後，在濃度較高的鍺基板上植入較淺較高濃度的參雜所形成的鎳鍺接面特性並不好，主要是因為鎳原子擴散到接面的邊緣，造成大量的漏電流。然而在形成鎳鍺化物之前植入氟離子卻能夠有效的抑制鎳原子的擴散而改善漏電流。另一方面來說，若在濃度較高的鍺基板上植入較深較低濃度的參雜，其所形成的鎳鍺接面特性較好，漏電流比前述之鎳鍺接面來的低。但是若在形成鎳鍺化物之前植入氟離子反而增進鎳原子的擴散進而造成漏電流增加。推測鎳原子的擴散深淺主要是靠植入的參雜濃度所產生的缺陷來決定。但在濃度較低的鍺基板上植入較淺較高濃度的參雜所形成的鎳鍺接面都會有好的特性，主要是因為接面深度較深而抑制鎳擴散所造成的破壞。而且就形成鎳鍺化物跟未形成鎳鍺化物的接面來比較，形成鎳鍺化物之後的接面，其順向電流都有明顯的增加，是因為在形成鎳鍺化物時，參雜的離子大量堆積在鎳鍺化物與鍺基板的界面，進而降低接觸阻抗。不論是磷參雜還是砷參雜都能看見此現象。 但在只經過形成鍺化物之後打入參雜的製程後，由於離析出來的接面深度太淺，受到鎳擴散的影響而造成大量的漏電流。因此結合以上這兩種製程，先做形成鍺化物之前打入參雜的製程然後再打入參雜後退火，能觀察到好的鎳鍺接面特性，且能同時提高順向電流。最後藉由量測接觸電阻來驗證此鎳鍺接面所觀察到的電流特性。在單純經過形成鍺化物之前打入參雜的製程的接面，其接觸電阻為2x10-5歐姆-平方公分 ; 而再打入參雜後退火的接面能使接觸電阻降到2x10-6歐姆-平方公分。 總而言之，與當今發表過的研究相比，本論文利用兩種使參雜堆積的製程來達到低接觸阻抗、低接面深度、以及好的接面特性。這個結果預期能夠改善N型鍺基板金氧半場效電晶體的特性。

In the past several decades, the research on Si-based devices progresses very fast. Furthermore, the Si-based MOSFETs have been successfully scaled down to 20 nm regime. However, scaling down the devices becomes more and more difficult and reaches the physics limits soon. Therefore, developing another ways to promote the device performance is necessary. Using new semiconducting materials is a way to improve performance. Because germanium has higher mobility and process compatibility for MOSFET fabrication process, germanium is considered to replace Si as the channel material in the future. Nevertheless, the n-type dopant diffusion in germanium is fast so that it is not easy to form shallow n+/p junction, and the high Schottky barrier height at the metal/n-Ge interface causes a high contact resistance due to the Fermi level pinning near the Ge valance band at the interface between metal and germanium. Therefore, the thesis would focus on the forming of shallow n+/p-Ge junction and low resistance metal/n-Ge contact. Because NiGe has the lowest resistivity and low temperature formation, NiGe is selected as the contact metal. The implantation before germanide (IBG) process means ion implantation is performed before germanide formation, and the implantation after germanide (IAG) process means ion implantation is performed after germanide formation. For the IBG junctions fabricated on heavily-doped substrate, very poor junction characteristic is observed by high dose phosphorous ion implantation due to the fast diffusion of Ni by virtue of defects which are generated by ion implantation. Fluorine ion implantation before NiGe formation could effectively suppress Ni diffusion and reduce the leakage current. Moreover, better junction characteristic can be obtained by low dose ion implantation due to the less defects resulting in less Ni diffusion. However, fluorine implantation before NiGe formation would enhance the Ni diffusion to degrade junction characteristic because the fluorine ion implantation induces extra defects. Next, on the lightly-doped substrate, good junction characteristic is more easily to be obtained than on heavily-doped substrate because the deeper junction depth on lightly-doped substrate so that the Ni diffusion would not destroy the junctions. In particular, after NiGe formation, the forward current obviously increases owing to the dopant segregation at the NiGe/Ge interface. Furthermore, either phosphorous or arsenic n+/p junction would have the dopant segregation effect. The arsenic n+/p junctions have relatively low activation concentration inferred by the I-V characteristic. Finally, because the IAG junctions have poor junction characteristic due to the segregated n+ layer is too thin to maintain good n-p junction and the Ni fast diffusion induces large leakage current, the IBG+IAG process is proposed. The IBG+IAG junction could achieve shallow junction depth and raise the forward current at the same time. Furthermore, the measured contact resistance of the IBG junction is about 2x10-5 -cm2 and the lowest contact resistance of IBG+IAG junction is 2x10-6 -cm2.Therefore, this thesis has formed a junction with shallower junction depth, lower leakage current, and lower contact resistance in comparison with previous studies. This achievement is expected to improve the performance of Ge nMOSFETs.