edfas.org 27 ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 26 NO. 1 (continued on page 30) This should result in a 50 ± 2 micron thick sample with a perfect surface and minimal thickness variation. Evaluations can be easily made on these samples to identify areas of interest. AREA-OF-INTEREST PROCESSING The area of interest can easily be thinned to the desired thickness. Assuming the target is 5 microns and the area is 4 microns square; the process would be as follows: 1. Coarse lap with a 1.5 mm diameter tool and 30-micron film to a thickness of 18 microns. 2. Fine lap with a 1.5 mm diameter tool, a coarse pad, and 9-micron slurry to a thickness of 9 microns. 3. Coarse polish with a coarse pad and 3-micron slurry to a thickness of 6 microns. 4. Fine polish with a fine pad and 1-micron slurry to a thickness of 5 microns. 5. Final polish with a final polish pad and colloidal silica. The first step must thin an area greater than 7 mm square, as the tool inside edge must be moved beyond the edge of the area of interest. The tool travel of each step needs to be somewhat less than that of the previous step to insure uniformity in thickness. A reduction of 100 to 200 microns in over travel for each step will give acceptable results. This requires that the first step have an over travel equal to the tool diameter plus 300 to 600 microns. Using the high value to be safe, the over travel for coarse lap needs to be 2.1 mm making an 8.2 mm square hole. This value reduced by 0.20 mm (0.4 mm total) for fine lap, coarse polish, and fine polish. Final polish can use the same over travel as fine polish. There are some considerations for the selection and size of the area of interest. As some type of profiling must be done prior to each process step, the area of interest must not be so large or positioned as to prevent measuring the mean plane of the die. This is necessary to adjust the tool path for any plane rotation resulting from removing, cleaning, and reinstalling the sample and holder. As there can be no contact measurements of the surface made after the coarse lap step, measuring the 50-micron RST surface can indicate a surface plane rotation. DEALING WITH THE VARIABLES As stated above, there are variables to be addressed. How well can the sample surface profile be measured and how well does the tool travel reproduce the profile? Most machine manufacturers confuse the terms accuracy, resolution, and reproducibility. Accuracy is an absolute and is related to NIST standards. That is, any movement of the tool will result in a tool position that is at an absolute position that can be traced to measurement standards. Most misuse “resolution” as accuracy. The two are very different. A voltmeter with a 1% accuracy may measure a 1-volt source as anywhere from 0.99 to 1.01 volts. The meter could have microvolt resolution resulting in measurements of 0.990000 to 1.010000, but the additional digits are meaningless as the readings are only good to two digits. The difference between accuracy and reproducibility is best indicated in Fig. 2. A profile is relative, not absolute. The capability of reproducing a profile depends on reproducibility, not accuracy or resolution. How consistent is the material removal process, both in average and across the die surface? Material removal rates are dependent on contact area, down force, and the material characteristics. As the die has a curved profile, the contact area changes according to the local slope of the die surface. Although the lapping film or a polishing pad has some compliance, the greater the slope, the lower the contact area. On a flat surface that is parallel to the tool face, the entire face is active in material removal. On a curved surface, only some portion Fig. 2 The difference between accuracy and repeatability. Accuracy is absolute; repeatability is relative but can be moved to the center. "ANY MOVEMENT OF THE TOOL WILL RESULT IN A TOOL POSITION THAT IS AT AN ABSOLUTE POSITION THAT CAN BE TRACED TO MEASUREMENT STANDARDS."
RkJQdWJsaXNoZXIy MTYyMzk3NQ==