Actually thermoelectric devices would tend to be problematic unless you have a nice external cold sink (e.g. the arctic air) for them. If the external air is warmer than body temp its a no-go (or unless you are running your Si @ 105 deg F and sinking into the body @ 98 deg F. ) Much warmer than 105-110 deg F and you start to cook the surrounding cells.
<br><br>They are also notoriously inefficient.<br><br>Much better, as was suggested to me indirectly from E.D. @ Extro III, is to run them off of glucose based fuel cells. I think this is also covered in NM Vol I. There is of course the Gd-148 power source (but that only flies if you have real nanotech unless you want to try and do it with MEMS scale technology (probably possible but I doubt anyone has tried it yet). The Gd-148 heat engine is discussed in NM Vol. I as well.
<br><br>It largely depends on whether you anticipate driving it from the body or some self-contained or "plug-in" power source. Assume the brain is a 10W machine. So adding another 10-30W Si "drain" on the glucose energy the body can supply probably isn't going to stretch things too much. But if you want to add 100-1000W then you have some real problems.
<br><br>Robert<br><br>