>Given that you work on space stuff, I'm sure you are correct, Alan,
>but I had always heard the opposite: that SMALLER geometry made you
>more radiation resistant (since smaller area meant that any one device
>would receive a far smaller dose).

Yes, but I am at the "user end", i.e. building devices into instruments. My 
understanding is that devices with geometries of (IIRC) 0.25um and smaller 
are inherently radiation resistant. It would appear that the reason is the 
geometry is too small to trap a passing particle, and so the device does not 
get upset. A larger feature will trap a particle and this produces a change 
in the device characteristics that upsets operation.

>There are two kinds of events, I think: single-particle events and
>multi-particle (cumulative) effects. A smaller geometry device will
>be more susceptible to the first (as less energy is needed in the
>single particles to influence the devices) but less susceptible
>to the cumulative effects (smaller area=lower dose per time).

single particle events are known as an SEU (Single Event Upset) and this can 
have several bad effects, the worst being that the parasitic devices that 
exist between the desired features and the substrate get activated, causing 
an SCR type action that typically causes very high current draw. An 
electronic fuse arrangement is often provided to turn off the power to the 
device(s) when this occurs to protect both the device and the power supply. 
This type of action is similar to that being done to PIC chips when people 
make the protection diodes conduct under normal operation, causing injection 
of electrons and holes into parasitic devices that are normally 
non-operational as they are reverse biased, and producing unwanted side 
effects.

The other long term radiation exposure produces degradation in device 
characteristics, being particularly bad in opto devices like opto-isolators 
and fibre optics, but also affecting FET devices, so part of the analysis 
requires "end of life" analysis to ensure that degraded parameters will 
still allow correct operation on the circuit. We have needed to do 
accelerated radiation testing on devices at times when there is no history 
of use. It is the expense of this sort of testing that means older devices 
are still made for radiation hardened use, and new devices are not available 
as no-one has committed to the time and expense of testing them, so the 
market keeps on using the old devices. 

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist