摘要:電致遷移效應是一種因為導電電子與導線原子之間碰撞的動量轉移所造成之質量傳遞。這種效應在高電流密度負荷下特別的重要。當這些元件的尺寸逐漸縮小時,其中的電流密度隨之上升,會造成更顯著的電致遷移效應。這些因電遷移所造成之質量傳遞,會在沿著電子流的方向上發展出一定程度的壓應力,而此一壓應力會引發金屬鬚晶的產生。金屬鬚晶會造成電路短路,此一現象已被證實了是造成通訊衛星、核能電廠、重要醫療器材失效的因素。根據美國太空總署的資料指出,光是金屬鬚晶就造成至少四例的太空運行中衛星的損毀及四例部分損傷[1]。
電遷移效應也逐漸成為影響微電子與能源相關應用上可靠度的主要因素。短期內,覆晶銲點內之電流密度將達到104 A/cm2。在如此高的電流密度下,電遷移效應將引發另一重要之物理效應 : 金屬導線上的焦耳熱效應,這會使得電遷移效應更加複雜。簡單地說,此時電遷移效應與焦耳熱效應會交互作用而偶合成一複雜現象;而到目前為止,關於這兩者效應的交互作用尚未浮現出一個明顯的輪廓。因此在理論的了解上與實驗的確認上,需要更進一步的探索。
在極高的電流密度作用下,覆晶封裝銲點的數種失效機制已有所報導。在所謂孔洞生成與傳遞機制中[2,3],一個小孔洞首先起始於高電流密度聚集區域,接下來便沿著金屬墊層與銲料之間的界面擴展,並在其界面逐漸形成一個缺口,最後造成電路的斷路因而失效。這種失效機制過程最主要的便是電遷移。焦耳熱效應造成的溫度上升在此處扮演的僅是次要的角色,但在其他狀況下,焦耳熱效應卻可能扮演更主要的角色。例如,我們及其他實驗團隊已發現由焦耳熱效應所造成的銲料局部鎔融[4,5]。最近,Alam 等人[5]更指出由我們實驗團隊提出的銅原子快速溶解[6,7]事實上是因為銅金屬墊層與鎔融銲料接觸所造成的。因為在極高的電流密度的狀態下銲料會變得局部鎔融,所以在多相材料之電遷移行為的研究就顯的格外重要。
本研究的目標在於探索電遷移效應與焦耳熱效應的交互偶合作用,以及研究在多相材料中的電遷移效應。而在目前階段,對如何抑制高電流密度下電遷移失效或是鬚晶的生長仍然無解,本研究目標著眼於發展出一系列關於釐清電遷移之基本性質的了解,只有在建立其基本機制之後,才可進一步做適當的設計來克服元件可靠度的問題。
Abstract: Electromigration is the mass transport caused by the momentum transfer between conducting electrons and the metal atoms of a current-carrying metal line. The effect is important in applications where high direct current densities are applied. As the feature size in these devices shrinks, the current density increases, resulting in much more pronounced electromigration. The mass movements caused by electromigration can develop significant compressive stresses along the direction of electron flow. These compressive stresses induce the formation of metallic whiskers. Metallic whisker can result in electrical shorts, which is documented to cause major failures in communication satellites, nuclear power plants, and critical medical devices. According to NASA, metallic whiskers alone were responsible for at least four complete satellite losses and four partial losses [1].
Electromigration is also becoming a major reliability concern in microelectronics and solar energy applications. It is anticipated that the current density through flip-chip interconnects will reach 104 A/cm2 soon. At such a current density level, electromigration is further complicated by a second physical process: Joule heating of the metal lines. In short, two fundamental processes, electromigration and Joule heating, are coupled under such a current density level. A clear picture regarding this cross-interaction has not emerged so far. Fundamental advancements in both theoretical understanding and experimental verification have to be achieved.
Several failure mechanisms have been proposed for flip-chip solder joints under the stressing of a very high current density. In the void formation-and-propagation mechanism [2, 3], a void first initiates at the current density crowding region. Then it extends along the UBM (Under Bump Metallurgy)/solder interface, and eventually forms a gap between the UBM and the solder, resulting in an open circuit failure. The main process responsible for this mechanism is electromigration. Joule heating here only plays a secondary role in raising the local temperature. Nevertheless, Joule heating can play a more dominating role under certain conditions. Recently, the local melting of the solder induced by Joule heating was identified as an operative failure mechanism by our research group and others [4, 5]. Moreover, Alam et al [5] pointed out the rapid Cu dissolution reported by us [6, 7] was in fact caused by the contact of the Cu UBM with the molten solder. Because under certain conditions solder can become locally molten during current stressing, the knowledge of electromigration behaviors in multi-phase materials becomes important.
The objective of the proposed study is to examine the combined effect of electromigration and Joule heating, and to investigate electromigration in multi-phase materials. At this stage there are no known solutions to control neither electromigration nor growth of whiskers at such high current density level. The proposed experimental studies are aimed to develop a clear understanding of the fundamental issues contributing to electromigration. Only after establishing the fundamental mechanism, can approaches to overcome the reliability problems be properly devised.