佛罗里达电话: 407-834-7800

传真: 407-767-8675

小时: 东部时间晚上 8 点 30 分至 5 点

Case History No. 1: Hewlett-Packard, Cupertino Integrated Circuit Division

对不起。本文不存在所选语言版本。你可以用英文阅读。

Case History No. 1: Hewlett-Packard, Cupertino Integrated Circuit Division

By John Boucher, President, Thermocarbon, Inc., Casselberry, Florida

Reprinted from “The Resin-Bonded Diamond Blade for Dicing Hard, & Brittle Materials.” February, 1981

 

At various feed rates, an entire wafer could be diced with no indication of impending failure. With respect to the stated objectives, given feed rate, edge chipping developed beyond specification, and after subsequent cuts the diamond blade broke. We knew that all selected feed rates could not possibly be optimum and therefore proceeded as follows:

A feed rate of 0.100in/s (2.54mm/s) was selected and each cut was observed until the first indication of excessive chipping could be noted. It was in the initial cuts of the fourth consecutive wafer that edge chipping increased from less than 0.002 in (0.051mm) to 0.008 in (0.203mm) in size. The first three wafers exhibited consistent edge chipping, wafer to wafer, however, the blade wear was highly unstable. Where one wafer already diced reflected a wear of 0.001 in (0.025mm) off the blade’s radius the others reflected 0.005 in (0.127mm) and 0.003 in (0.076mm) respectively.

Power readings taken at the spindle motor drive indicated an ever increasing load on the system. The assumption was that the blade was becoming too hard due to under-feeding and therefore we increased the feed rate to 0.120 in/s (3.048mm/s) without changing any other variable in the system. The same blade, in its hardened condition, continued to cut succeeding lines in the sapphire substrate and showed significant improvement in edge chipping until it finally settled to a consistency of less than 0.0015 in (0.0381mm). This continued for an additional 12 wafers. The “steady state” for this combination of blade and parameters had been reached for the application. The process went to production and close study provided for finer tuning throughout the initial startup as more material was processed and data obtained. The final result was one of higher dicing efficiency and reduced costs.

Material: Sapphire, 0.017-in (0.432-mm) thick
Spindle Speed: 15,000 rpm
Blade Type: Resin-bond, 0.005-in (0.127-mm) thick
Diamond particle size - 54 μm
Blade Matrix: RMX2
Exposure: 0.040 in (1.016 mm)
Problem: Excessive edge chipping and blade breakage
Solution: Determine proper feed rate for “steady state”

Comment: Previous in-depth tests resulted in the above stated operating parameters as optimum for this application. The feed rate was the last variable to be determined. Other variables such as blade flanges care and torque, workpiece mounting, etc. have been addressed with regard to the three fundamental requirements outlined in article "Part I: Introduction And Fundamental Requirements Of Diamond Dicing".

分享文章