| 摘要: | 為了解決鑽石刀片切割(Saw-Dicing)時因振動造成晶片背面破裂(back side chipping)問題，本研究應用藍芽晶片感測器(感測器)先收集其振動源的數據，再藉由振動值的分析曲線來判斷振動起伏的主要來源，再由循環式品質管理(PDCA)先針對振動的來源，列舉影響振動的不同來源進行分析，分別考慮幾項條件，包含1.不同水流量、2.法蘭表面平坦度(TTV(Total Thickness Variation)、 3.刀子總類的選擇等以上三項進行實驗，以比較不同加工條件的晶背破裂程度，來找出主要的振動源。實驗結果顯示:
在相同因素的情況下只改變水流量大小的實驗，水流量2.0L/min跟1.5L/min得到的，合格公差內碎裂深度的晶片比率分別為0.2206%.跟0.2352%碎裂深度約為1:1.066，由振動曲線來看兩者相當接近。
在相同因素的情況下只改變SD#3000跟SD#4500型號的鑽石刀片進行晶圓切割實驗，合格公差內碎裂深度的晶片比率分別為0.2026%跟0.2113%，碎裂深度約為1:1.043 ，由振動曲線來看兩者相當接近。
在相同因素的情況下只改變全新的法蘭平坦度跟舊的法蘭平坦度進行晶圓切割實驗，合格公差內碎裂深度的晶片比率分別為0.0340%跟0.6267%，碎裂深度約為1:18 由振動曲線來看差異非常大。
由實驗結果可知，平坦度破裂深度的比例為1:18遠遠大於改變水流量的破裂深度1:1.043和不同鑽石刀片的破裂深度1:1.066。
由本實驗看出破裂深度與振動加速度成正比，可以知道法蘭表面平坦度是影響振動的主要來源，經由振動晶片感測器測得破裂深度異常與正常比例相差18倍以上，晶片切割寬度為0.1mm容許破裂公差10µm以下，本研究證實當振動加速度小於2.5m/s2時，可達到破裂公差10µm以下的標準，良率可以達到100%。
;To address the issue of back side chipping during Saw-Dicing due to vibration, this study employed Bluetooth chip sensors to first collect data on the vibration sources. Subsequently, the main sources of vibration fluctuations were determined through the analysis of vibration values. Then, through a Plan-Do-Check-Act (PDCA) approach, the study focused on the vibration sources, listing and analyzing various factors influencing vibration, including different water flow rates, Total Thickness Variation (TTV) of flange surfaces, and the selection of blade types. Experiments were conducted under these conditions to compare the extent of backside chipping under different processing conditions and identify the primary sources of vibration.
The experimental results showed that:
When only the water flow rate was changed under the same conditions, the chip fracture depth within the acceptable tolerance ratio for water flow rates of 2.0L/min and 1.5L/min was 0.2206% and 0.2352%, respectively, with a fracture depth ratio of approximately 1:1.066. The vibration curves indicated that these two were quite close.
When only the types of diamond blades (SD#3000 and SD#4500) were changed under the same conditions, the chip fracture depth within the acceptable tolerance ratio was 0.2026% and 0.2113%, respectively, with a fracture depth ratio of approximately 1:1.043. The vibration curves indicated that these two were also quite close.
When only the flatness of the new and old flanges was changed under the same conditions, the chip fracture depth within the acceptable tolerance ratio was 0.0340% and 0.6267%, respectively, with a fracture depth ratio of approximately 1:18. The vibration curves indicated a significant difference between the two.
From the experimental results, it can be observed that the ratio of fracture depth to flatness is 1:18, which is much greater than the ratios for changing water flow (1:1.043) and different diamond blades (1:1.066). It is evident that the flatness of the flange surface is the primary source affecting vibration. Through the measurement of abnormal and normal fracture depth ratios obtained by vibration chip sensors, with a chip cutting width of 0.1mm allowing for fracture tolerance below 10µm, this study confirms that when the vibration intensity is less than 2.5, the standard for fracture tolerance below 10µm can be achieved, and the yield can reach 100% |
