Chip-on-Glass（COG）彈性凸塊電性與可靠度之研究

Abstract

COG使用NCF接合，能夠達到較低的製程溫度，降低成本，所以顯示器業者逐漸廣泛使用這些製程。COG經過30~100oC熱循環之後的凸塊電阻，經過六次熱循環使用四點量測結果，電阻約從1.6Ω上升到2.15Ω並未超過5Ω為可靠度範圍內，電阻值的貢獻大都來自接觸電阻，因為此接合為物理接合，期間電阻上升為膠材劣化所導致。COG試片經過80oC約1000小時之後觀察電阻上升情形，一片約從0.48Ω上升至1.26Ω另一片從0.73Ω到1.52Ω也符合可靠範圍內，但是上升的原因不完全是膠材劣化導致，因為凸塊本身具有彈性，所以在壓合之後還會回彈造成電阻上升，因此會使用另兩片未熱時效的試片來比較，得知結果大概65％的電阻上升是來自本身具彈性導致。COG 室溫耐電流約65mA，破壞位置發生在玻璃端鋁導線，中間可以觀察到45mA開始電阻有下降的趨勢，可能是因為溫升之後造成金表面與 ITO 接合更完美所引起的。COG 膠材在85oC 85%以及外加30V或40V偏壓測試後，發現電阻都在109Ω以上符合可靠度範圍內，電阻有上升下降的趨勢可能是膠材吸收水氣造成或量測的電阻過高已經超過儀器解析度，所以上升下降可能為儀器干擾所導致。

Non conductive film（NCF）is widely adopted by flat panel display（FPD）industry in chip-on-glass（COG）bonding method it provides low processing temperature and low product cost. As revealed by four-point test, the contact resistances of bumps subjected to six thermal cycles between 30~100oC raised form 1.6 to 2.5Ω, which is lower than 5Ω as require by the reliability standard. The resistance increment is largely resulted form the change of contact resistance. For the COG joint subjected to thermal storage test at 80 oC for 1000hours, the resistance raises from 0.48 to 1.26Ω while in the sample, the resistance raises from 0.73 to 1.52Ω; all test results fit the reliability requirements. But the reason of increment is not only NCF material degradation cause, but also the bump is elastically. Therefore elastic bump resile caused the resistance to increase after pressing. Consequently I used another two samples which is not thermal aging compared to test samples, the results were show bump resistances increasing about 65%, its come form elastic bumps itself. The best current stressing of COG was 65mA at room temperature. The failure position took place in aluminum wire on glass. Observed the resistance to decrease in 45mA, may be the contact was perfectly between Au surface and ITO after temperature increasing. The resistances of NCF of COG were above 109Ω all accorded with reliability range after 85 oC 85% humidity and 30V or 40V bias test. The resistance of NCF sometimes increased and decreased may be cased by NCF absorbing vapor or the resistance of measurement is higher than instrumental analysis. Therefore the increase or decrease of resistances of NCF was caused possibility by instrumental noise.