三維積體電路異質整合技術開發之聚醯亞胺材料與鈍化層附著強度以及鈷金屬超薄緩衝層應用於混合接合之研究

Abstract

三維積體電路(3D IC)被認為是滿足半導體領域中摩爾定律有希望的方法之一。3D IC的關鍵技術包括矽通孔（TSV）製程，晶圓薄化和晶圓接合。本論文的3D IC異質整合關鍵技術開發為聚合物 - 金屬混合接合，一種實現3D金屬連線和結構增強的接合方法。研究上所使用的金屬為銅和錫；而聚合物則以聚醯亞胺為主。由於聚合物沉積於鈍化層上方，為了確保混合結合能順利進行，鈍化層與聚合物的附著性是極其重要。因此，第一個研究項目旨在探討聚酰亞胺對各種可能的鈍化層的粘附強度研究。使用四點彎曲儀（4PB）對各個附著強度進行定量測試。 另一方面，為了確保混和接合的品質與良率，低溫金屬接合被廣泛的採用，尤其又以接合溫度大約250°C的銅錫(Cu-Sn)金屬接合為業界所使用的標準接合材料。然而，銅錫結合的潛在缺點是形成金屬間化合物（IMC），因此在銅和錫之間需要超薄緩衝層（UBL）以延遲或防止IMC形成。在本論文中所採用的UBL材料為鈷（Co），因為鈷已廣泛用為TSV製程當中的潤濕層。因此本研究旨在確定利用鈷作為UBL的可能性。針對含有/不含有UBL的銅錫接合材料進行比較。從實驗結果發現十奈米的鈷能夠有效改善銅錫結合的品質。

3D integrated circuit (IC) is recognized as one of the promising approach in keeping up with Moore’s law in the semiconductor field. Some of the key technologies of the approach include bonding, through silicon via (TSV), thinning and handling. The focused key technology in this thesis is polymer-metal hybrid bonding, a bonding approach to achieve electrical connection and structural enhancement simultaneously. In this thesis, polyimide is used for polymer bonding while Cu and Sn are used for metal bonding. Since hybrid bonding is often carried out on top of a passivation layer, to ensure the success of hybrid bonding, the adhesion of the materials used respective to the passivation layer is important. Thus the first research topic aims to study on the adhesion strength of polyimide against various possible passivation layers. The respective adhesion strengths are quantified using the Four Point Bending system (4PB). On the other hand, to ensure the success of hybrid bonding, low temperature metal bonding is always preferred. Copper-tin (Cu-Sn) metal bonding has recently been widely researched on because of its low bonding temperature requirement. However, the disadvantage of direct Cu-Sn bonding is the formation of intermetallic compound (IMC) thus an ultra-thin buffer layer (UBL) is required in between the Cu and Sn layer to delay or prevent the IMC formation. The UBL material researched in this thesis is Cobalt (Co) since Co has recently been widely used as a wetting layer for the TSV process thus this research is conducted with the aim to identify the possibility of utilizing Co as the UBL. A comparison between Cu-Sn with and without UBL is conducted. It is found that 10nm Co is capable of improving the Cu-Sn bonding quality.