Si/Ni/In與Cu/Ag引腳架固晶接合研究

Abstract

近年來，LED產業蓬勃發展，跟其他電子產業一樣，LED也朝向高功率、高效能方向前進，但式LED晶片對於溫度範圍非常地敏感，因此其散熱問題變成目前需要克服之難題。LED封裝當中的「固晶」技術就是散熱設計一個重要的環節，目前的固晶製程大約分為三類，銀膠固晶法、銲錫固晶法和金錫共晶固晶法，但皆有一些不好的問題存在。本研究提供一新的製程方式─固液擴散接合法，利用矽晶片上分別鍍上Ni/In和鍍銀之銅引腳架與銀基板在低溫下進行界面反應產生介金屬化合物，並探討各接合溫度及接合時間所製程之試片其接點強度測試。 實驗結果顯示，利用固液擴散接合法進行銀銦界面反應，其引腳架上之鍍銀層與矽晶片上之鍍銦層之厚度必須要大於2 : 1，才能將薄膜銦完全反應。而Si/Ni/In晶片與銀基板接合的界面反應，在低溫短時間下會迅速形成AgIn2介金屬相，而在高溫長時間下，除了會形成Ag2In相外，In元素還會不斷擴散進入銀基板內形成Ag3In，但因為其反應產生之介金屬相較為複雜，且在掃描式電子顯微鏡下並無法分別各相差別，所以其反應動力學之n值算出並不是屬於擴散控制或界面控制；另外，在鎳銦界面，高溫長時間下也會反應形成Ni2In3之介金屬相。 銀銦接點強度來說，200℃接合10分鐘的條件下因為有AgIn2相之存再造成接點強度較弱；隨著加熱時間增高，反應界面之孔洞及裂縫數減少，使接點強度上升，但過長的接合時間，會造成介金屬層過厚，接點強度反而下降。

Recently, LED (Light Emitting Diode) has made a big progress. Like other electronic device, high power device have led to a movement towards LED. Because LED chip is sensitive to the temperature, thermal dissipation becomes the most important challenge to solve now. In LED package, die bonding is a key point of thermal dissipation. Nowadays, silver paste, soldering, and tin-gold eutectic bonding are usually used in LED die bonding, but these methods have some problems. This research provides a new die bonding method which called solid liquid inter-diffusion bonding (SLID). In this procedure silicon wafers (sputtering nickel and indium) assemble copper lead frames (sputtering silver) and silver substrate at low temperature to make indium and silver turn to intermetallic compound. This research also studies the die shear force at different bonding temperature and different bonding time. The result is as follows, in silver-indium SLID process, the thickness of silver on copper lead frame must be twice thicker than the thickness of indium on silicon wafer. Thus, we can make sure the reaction is complete. In interface reaction, silver and indium will form AgIn2 rapidly at lower bonding temperature. At higher bonding temperature and longer bonding time, besides Ag2In which is the main phase, indium will keep diffusing in silver substrate to form Ag3In. This reaction is complicate, so the N number of reaction kinetics is cast to neither diffusion control nor interface control. By the way, at nickel-indium interface, it will form Ni2In3 intermetallic compound at higher bonding temperature and longer bonding time. In shear force test, the joint formed at 200℃ and 10 minutes is too weak because AgIn2 exist at this bonding condition. As the bonding temperature increasing, the voids and cracks decrease at the reaction interface. This makes the joint become stronger. If the bonding time too long, the intermetallic compound become too thick. It makes the joint become weaker instead.