I suspect that lunar temperature would make electrochemical options tricky, if running at ambient temperature most reactions would slow to a crawl; while if heated by sunlight or system waste heat they could end up well above comfortable operating temperature, having no atmosphere is obnoxious like that(though you might be able to exploit a thermal gradient between a sunlit and shadowed or dark-side area for quite a while before your exploitation equilibrated temperatures and shut down that energy source; not sure how viable geothermal heat pump arrangements in lunar regolith would be).
In terms of locateability your suggestion of a conductive and/or RF reflective structure seems like a very plausible one. In percentage terms there’s a decent amount of metal on the moon; but not in bulk metallic form with suspiciously artificial shape.
As an accessible archive I’m curious about the viability of a radio equivalent of what they have etched on these disks(the small but unencoded pictures telling you about how to interpret the higher density encoded data sections): you’d want a first-stage accessible with as little knowledge and equipment as possible(our hypothetical listener would probably need to know where to point a fairly directional antenna, I suspect that preposterous power would be required if the lunar transmitter was omnidirectional enough for a ‘just point at moon from somewhere in earth’ case to work): have a first stage in something simple and retro like a looping recording in several languages, broadcast as AM audio, describing the encoding of the actually efficient and high speed transmission; with the bulk of the data being sent out over and over on that channel so that anyone who can scavenge enough circuitry to build the receiver out of the nuclear rubble could obtain a copy just by keeping their antenna on target and waiting. No need to actually retrieve the hard copy(indeed, the distance would keep it safe).
The RF beacon would obviously fail for one reason or another long before the disks decay(I’m just curious if a realistic target given the environmental constraints would be 10 years? 100? 1,000?); but being accessible with comparatively modest RF gear would be useful to a lot more hypothetical futures than needing some billions of dollars worth of at least 1970-level rocket science(since any disaster that leaves you with that will be virtually certain to spare some of the terrestrial copies).
