- 學位論文
高功率發光二極體固晶層受熱循環之疲勞壽命與可靠度評估
Thermal Fatigue Life and Reliability Evaluation of Die Attachment Layer of High-Power LED under Thermal Cycling Conditions
摘要
在講求節能的現代，由於高功率發光二極體(High Power Light Emitting Diode)具有壽命長、耗能低等優點，在照明應用上已越來越普及，然而其熱管理一直是個重要問題，且影響高功率發光二極體之可靠度甚劇。在高功率發光二極體封裝結構中，緊鄰發光晶片的固晶黏著層，簡稱固晶層(Die Attachment Layer)，就是一個影響熱管理與可靠度的重要元件。它不僅具有固定晶片的功能，更為晶片向外散熱的一條主要路徑，極端存在熱疲勞的隱憂，然而相關的可靠度研究卻很缺乏。就此，本研究以有限元素法模擬高功率發光二極體固晶層受熱循環負載下之力學行為，而後將有限元素數值分析結果帶入Coffin-Manson疲勞預估方程式，以得到高功率發光二極體的熱疲勞壽命。值得注意的是，以往藉助有限元素模擬高功率發光二極體封裝力學文獻所分析出來之壽命大多為一定值，無法反應實際產品測試所得一批相同二極體壽命往往具有離散的特性，也無法評估其壽命分佈與失效率等相關可靠度指標。有鑑於此，本研究考慮高功率發光二極體之固晶層幾何尺寸與機械性質等參數之不確定性，並將該些參數以隨機變數視之，利用蒙地卡羅法抽樣模擬該些參數之樣本個體數據後，帶入有限元素數值分析及其後之壽命預估方程式，以獲得一批代表抽樣樣本二極體之熱疲勞壽命，再經由機率圖紙嵌合與Anderson-Darling檢定找出樣本壽命的機率分佈函數，以進一步得到特定分佈下之平均壽命與失效率等相關可靠度指標。本研究結果發現，我們所探討的高功率發光二極體，在固晶層的寬度、厚度、楊氏係數與熱膨脹係數為隨機變數的考量下，所求得之熱疲勞壽命範圍介於731至880個循環數，經檢定後，三參數韋伯分佈最適合描述此壽命分佈情形，其形狀參數為2.6，尺度參數為105，位置參數即最小壽命為704，以上反映固晶層參數之不確定性對壽命分佈與可靠度的影響。
並列摘要
Power conservation is a very important aspect of modern day technology, and with its long life and low energy consumption, the High Power Light Emitting Diode has become very popular for lighting purposes. Thermal management plays a key role in the reliability of high power LEDs, which is essential for all of its applications. The die-attach layer is a important component of an HP LED package, not only does it hold the chip in place, it also provides the main path for heat dissipation in the structure. However, it is also susceptible to thermal fatigue failure, affecting the reliability of the HP LED itself. Seldom people discuss this issue. This study uses Finite Element Method to simulate and analyze the mechanical behavior of an HP LED die-attach layer under thermal cycling conditions, using the numerical results as the input for the Coffin-Manson relationship the predict its fatigue life. It is worth noting that past studies are mostly limited to finding a fixed value for the fatigue life of the package, which cannot truly reflect the discrete qualities of real life testing. Furthermore, they cannot provide vital information such as MTTF and the Hazard Rate of the HP LED. With this in mind, this study consider uncertainties of both geometric dimension and material properties of the die attachment layer, regarding them as random variables, which can be simulated by the Monte-Carlo method. The sampled data is then applied to the FEM analysis to evaluate its fatigue life, which can lead to the probability distribution of fatigue life by using the Anderson-Darling Test and probability plot to find the relevant variables. We can then obtain information such as the MTTF and failure rate function using this distribution.