晶圓打線檢測系統之研究

Abstract

在半導體產業中，晶圓封裝往往被認為是技術層次較低的一環。然而由於半導體線寬走向0.1微米以下製程，封裝已成為良率的瓶頸。在晶圓封裝過程中，打線是最精密，最複雜，也是最容易發生錯誤的製程。客戶在評估封裝廠能力時，打線的品質是最大的重點。在打線製程中，必須對金線及金線球的位置、線弧高度和形狀等進行詳細的檢驗，一但發生錯誤則會造成大量的報廢及損失。目前封裝廠絕大多數是以人工對打線完成之半成品進行抽檢，然而人工目視不但容易失誤且耗時過久，勢必無法滿足未來客戶100%品檢的要求。 本論文的目的，在發展出一套適用於封裝廠打線製程的電腦視覺檢測系統，能檢測出已知所有可能會發生的錯誤，包括漏線、線偏、斷線、短路和線塌。金線由於線徑極細加上背景之晶圓及導線架反射性高，造成從上視角取像的二維影像中，不易對金線進行定位及檢測。因此本論文先對金線及金線球的反射特性進行分析研究。並以此為依據，發展出一套可自動程式控制光線入射角的結構光源系統。使用此一結構光源系統，金線的位置資訊將能清晰的表現在二維檢測影像中，因此漏線、線偏、斷線及短路現象都能有效的被觀測。而本論文將立體線塌瑕疵的檢測方法，由傳統的高度量測轉換為斜率檢測；以程式精密的控制光線入射角範圍大小，使立體的線塌瑕疵也能被突顯在二維的檢測影像中。 本論文並開發出一系列的演算法，能快速且有效的檢測出所有打線製程中的瑕疵。經由實驗結果可證明，本論文提出之金線檢測系統，其檢測速度遠高於目前現有之打線機台。此檢測系統將可應用於實際晶圓打線生產線，達成100%全檢的目標。

Wire bonding is the process that makes the connection between the IC chip and the substrate. The minimum diameter of a bonding wire, which appears as a 3D loop in an IC chip, is about 20 μm. This kind of bonding wire easily bends or sags in the middle segment. Bonding wire defects can be classified as 2D-type and 3D-type. The 2D-type defects include broken, lost, shifted or shorted wires. These defects can be inspected from a 2D top-view image of the wire. The 3D-type defect is wire sagging, which is difficult to be inspected from a 2D top-view image. In this dissertation, a structured lighting system was designed and developed to facilitate all 2D and 3D-type defect inspection. The devised lighting system can be programmed to turn the structured LEDs on or off independently. For the 2D-type defect, the lighting system can provide light in proper incident angle to illuminate the bonding wire and the position of bonding wire will be represented in the image clearly. For the 3D-type defect, we supplant the traditional height measurement by the slope inspection. The structured lighting system can provide light in proper incident range and the variance of slope, the sagged wire segment, will be highlighted in a 2D image. Experiments show that the devised illumination system is effective and robust for wire bonding defect inspection. We developed a fast and robust set of machine vision algorithms for wire bonding inspection. The defect of a bonding wire being broken, lost, shifted, shorted or sagged in an IC chip can be inspected automatically. Some illustrations are given to show the efficiency and effectiveness of this proposed novel wire bonding vision inspection system.