Sidewall Chipping Investigation & Challenges on 100um Thin Low-K Wafer with DAF

The significance of wafer dicing to the quality of the final product is prominent in the semiconductor manufacturing process. Any dicing defect will impact the electrical properties of the chip. At present, common wafer dicing methods are as follows: blade dicing, hybrid process (laser grooving and then blade dicing), plasma dicing. With plasma dicing, near zero chipping, but high cost and it requires special requirement in the saw lane. Blade dicing, a mature and widely used dicing process, especially for silicon (Si) wafer, risk of die chipping due to its mechanical stress, but lower cost compared with plasma dicing. However, chipping could be reduced through process optimization.Wafer fabrication house continually faces challenges in manufacturing semiconductor devices because the devices getting thinner and more complicated for integrated circuits increasing in circuit density. In this paper, it focuses on the research through different blade types and dicing parameters (spindle speed, feed rate) in dealing with thin low-k wafers (100 microns) with die attach films (DAF) tape and narrow saw lane, to solve the sidewall chipping quality issue.Our process characterization including four steps: 1) feasibility study; 2) pre-screen Design of Experiment (DOE); 3) full DOE; 4) validation trial. In feasibility study, we used standard dicing method, blade, and parameters to know the chipping performance differences between normal versus (vs) low-k wafer, DAF vs non-DAF tape. This first step study helped to screen dicing method and blade selection. In second step pre-screen DOE, we selected two different kinds of blade based on wafer thickness and wafer material of the project, also including different dicing method. In full DOE, we selected the key factors from pre-screen DOE learnt in the dicing study matrix and collected more than 1000 die samples in different area to measure the chipping size. The collected data was analyzed by Minitab – a professional data processing software and decided the optimum dicing parameters and key factors from data graph. The verification experiments in confirmation trial and final validation are necessary for the consideration of the accuracy and rigor of the experimental results.