I'd suggest sonication is not the best way to prep a whole brain. We use teflon grinders in a glass tube. See link for similar products. If your system is on ice, it all works out fine. We use 8-10 "up down strokes" to fully homogenise a brain.

You could also try grinding the brain manually in a traditional mortar and pestle, but in the presence of dry ice, so it's nice and hard. This works really well for us.

I also stumbled across this link, it looks useful and informative.

http://opsdiagnostics.com/applications/samplehomogenization/homogenizationguidepart1.html

http://www.thomassci.com/Equipment/Grinders/_/CAF5F85C-7178-497E-85C7-C3FA2F33A1B9?q=Homogenizing%20Pestles

http://opsdiagnostics.com/applications/samplehomogenization/homogenizationguidepart1.html

We use, for spinal cord and brain, the sartorius labsonic M. However, I do not think there is a way to avoid heat generation. Tissue disruption requires energy transfer, energy transfer typically results in heat. We protect against this by homogenizing tissues in tubes that are kept in an ice water bath during the process.

http://www.montreal-biotech.com/Brochures/Broch_Sonicators+Homogenizers_Product_Overview_SL-0013-e.pdf

Thank you Peter, Fernando and John for suggestions. I agree that teflon grinder can be best for this purpose. But every day I have lot of samples to do. I do not want to eliminate heat completely. I can keep samples on ice. It works well by my experience. I want to use probe sonicator because you can easily clean the sonicator probe and brains are in use and throw glass tubes. I never used teflon grinder. What is your opinion on it?

Nah teflon grinders aren't good for fatty tissue like mouse brain.