現在雷射切割已被廣泛應用在半導體封裝製程,而雷射切割製程目前遭遇的最大問題是熱影響,導致經過雷射切割後的晶片強度會減弱。加工目標經過雷射高溫照射後,理論上要完全汽化,並由其物質汽化所產生的交互作用讓高溫消散不會殘留於晶片。但實際切割情況會受到加工物材質不同與其參數配置不適,導致加工物未能完全汽化,而產生一些缺陷,如雷射後汽化未完全的矽碎片形成回填(recast)、或在晶片側面形成孔洞(void)及細小裂紋(micro crack),這些問題就可能導致晶片在後續製程發生缺陷(Defect)、功能失效(Function fail)。因此合適的參數設計非常重要,可避免上述之雷射切割異常發生。 實驗設計可以有系統及有效地提高製程良率、降低製程變異以及減少生產成本。本研究是依照機台參數特性配合實驗設計(Design of experiments, DOE),將量測數據輸入統計分析軟體JMP找出對結果有顯著影響性的主要參數與雷射切割最適化參數模型。 最後驗證最適化參數雷射切割所得之量測數據,皆於預測模型的誤差項範圍內。實驗結果證實了本次實驗設計所得之參數預測模型確實為雷射切割最適化參數,以及實驗設計可以成功應用於製程開發。
Nowadays, the laser grooving had been widely applied on semiconductor package process, and the biggest challenge to laser grooving process is the HAZ (Heat affect zone) that will impact chip strength. Theory, the TEGs and low-k layer of wafer should be vaporized completely, and the high temperature should be released and not remain on chip by the interaction of vaporization. But the actual process, we should consider the recipe whether well match with the wafer. If not, there are some defects will generate, like the recast, sidewall void and micro-crack formation by metallic layer incompletely vaporization, and these defect will leads chip function fail. For this purpose, the proper laser grooving recipe design is very important to the quality of chip, which can prevent above defects occurrence. The design of experiments is a system method with high efficient on the process control, cost reduction, and yield improvement. For this study, we'll according to the machine controlled parameter(factors) to perform experiments by DOE then measure the laser grooving width and depth by the 3D profiler, and then statistics analyze the measurement data by JMP to determine the main effect factor to laser grooving process and the proper recipe module. After verification run, the response result is same as recipe module prediction, and also proves the DOE is workable and it can be applied successfully to the process set up.