It can be done. First, it requires a good bond strength - if you have poor bonding, it of course won't work. I've diced stuff with pretty mediocre bond strength, so it doesn't have to be extraordinary, but it should be pretty solid. I've successfully cut a triple stack of bonded silicon (total thickness ~75 mils). First, you need to get a "high exposure" blade. The exposure on a dicing blade determines how deep you can cut with a given blade. Most standard blades have exposures of .030" or so, which may be inadequate for you. Contact the maker of your dicing saw, and see what they have to offer. (I did this on a Disco saw, and got a special kit to use a high exposure hubless blade.) When it comes to actually cutting, you will want to do each cut in multiple passes. Use a high spindle speed, a slow cutting speed, and cut in multiple steps. I believe I cut in .010" increments, but it may have been deeper than that, I don't remember. Disco saws (again, the only one I have experience with) can be programmed to do a cut in multiple passes. I set up in semi-automatic mode. A few design considerations: 1) The kerf for these deep cuts is wider than for a standard cut. (The kerf is the width of the cut). If the exact edge placement is critical, you will have to take this into account. Don't put your devices too close together. 2) Make sure that "dead" areas of your wafers are bonded together, so that when you are cutting, pieces of the top wafer don't come free. This can break the dicing blade, which is prone to breakage with high exposures. Good Luck! David Nemeth -----Original Message----- From: mems-talk-bounces@memsnet.org [mailto:mems-talk-bounces@memsnet.org] On Behalf Of kris Sent: Wednesday, May 09, 2007 11:32 AM To: mems-talk@memsnet.org Subject: [mems-talk] bonded wafers dice Hello, I was wondering if we can dice a Au-Au thermocompression bonded wafer pair using the diamond saw. I need individual dies from the wafer pair to go for the etch trails. Your suggestions are highly appreciated.