晶圓打線腳位之自動驗證與確認

Abstract

在IC製程中，實體載板是根據CAD模型所做出的，然而製程限制造成實體載板與CAD模型有出入，主要是製程問題造成腳位有平移、露金、變形。因此為了協助IC自動打線，需先知道CAD上IC腳位的打線資訊，以及與其相對應實體IC載板的腳位資訊，方能將正確的打線資訊載入到打線機台。為此，本研究針對運用AUTOCAD繪出的載板圖(載板設計圖)，與利用CCD攝影機所取得的實體IC載板圖(載板實體影像圖)做比對，提出兩者間之IC腳位自動對照表演算法；透過連通元件分析協助腳位自動編號、再配合最佳化重疊比對方法，協助載板設計圖與載板實體影像圖之間的腳位對應辨識。 IC腳位自動對照表演算法以重疊比對為核心方法，透過Affine Transformation(縮放，旋轉與平移)，進行影像對應之調整， 並以載板設計圖與實體IC載板圖的腳位重疊面積為最大且逼近載板設計圖的腳位個數做為判斷依據。 研究結果顯示自動腳位對照表演算法，可在不同載板樣本，完成腳位對照，以協助打線腳位之自動驗證與確認，提升工廠的工作效率、設備的附加價值與生產力。未來亦可提供協助自動打線端點的校正與檢驗等相關研究。

To wire-bond automatically, the bonding lead position from CAD model and the corresponding lead position of actual base material should be pre-given. In the manufacturing process, base material is produced based on CAD-model. However, due to production technique limitation, the actual base material often differs from CAD-model specification. That is, some lead on the base material may be shifted, exposed, and distorted. This thesis proposed a lead matching algorithm which could find the lead positions between the leads on CAD-model and the corresponding leads on actual base material. First, a CAD-graph of base material was used as a template, and the image of actual corresponding base material was grabbed by using a CCD camera. Second, the Connected Component Labeling (CCL) method was applied on each of the leads of the image to have each lead be treated as a complete component. Third, the particle swarm optimization (PSO) method was used to find a set of best affine transformation parameters between the leads of the CAD-model and the actual corresponding base material image. The thesis would output a table which shows the matched corresponding leads positions. Experimentations revealed that the developed algorithm responds good lead matching and could be applied to various base material types. The proposed method can also be applied in mass production environment to reduce the manual preparation work before actual wire-bonding process.