The basements of large buildings made from concrete spring to mind, then geographical areas that are at or below sea level.
Random thought: anti-coincidence detector? A detector above the RAM chips and one below with data processing set so that if the detector above gets a count and the one below does not or sees a count with significantly lower energy, you know some energy has been absorbed somewhere near the RAM.
This won't work because all particle detectors are noisy. Pretty much all tracking detectors need to use multiple coincidences to determine that a muon has indeed passed by. Individual counts by themselves are not reliable. My past life as a particle physicist included design and build of muon detectors using scintillating plastic and (much more expensive) multi-anode photomultipliers to look at nuclear composition of volumes. Something like an underfunded British version of decision sciences http://www.decisionsciences.com/ :-)
The energy involved in radioactive decay is far less than that of cosmic rays. If the goal is to protect ICs from bit flips then indeed putting them under concrete or natural rock will reduce them.
You don't have a background in nuclear spectroscopy, and I claim my £5.
The irony of this: what is typically used for nuclear spectroscopy? Semiconductor detectors? If you get one out -- albeit probably a geranium^Wgermanium detector rather than silicon, if you do gamma rays -- potassium-40 products stand out in a typical environment. (I've mostly done that over sandstone -- why the lab was there; it may be different somewhere like Edinburgh.) Concentrated in a calibration source rather than a concrete block, it would be controlled.
Random thought: anti-coincidence detector? A detector above the RAM chips and one below with data processing set so that if the detector above gets a count and the one below does not or sees a count with significantly lower energy, you know some energy has been absorbed somewhere near the RAM.