I was doing some calculations yesterday about blowing holes in things with lasers from DVD writers. My estimate is that you can blow a micron-diameter hole in just about anything with those 400-milliwatt red lasers if you get them properly focused, in about a microsecond.
That would store a few gigabits in a fairly permanent way into a 3.5" square of material, in a few gigamicroseconds, so a few kiloseconds. This is the first long-term archival technology that I think is feasible to do on a shoestring; a DVD-writer is a lot cheaper than a focused-ion-beam etching machine. One of the most appealing media to use for this purpose would actually be old floppy discs, because the Mylar medium is extremely chemically stable, and the oxide coating is nice and optically absorbent, but glass or aluminum seems like it would work fine.
An even simpler alternative that would work with metal foils or even metal surfaces: blow holes in them using sparks from a graphite point (of a radius on the order of a micron, about as sharp as a new scalpel, a bit sharper than the sharpness of a razor blade) driven by a capacitor. However, most metals are not very stable in Earth’s atmosphere.
Gold is an exception; it’s stable in Earth’s atmosphere. I can buy 100 sheets of 140mm square gold leaf for AR$350 (US$20), which works out to about 2 million square millimeters, or 100 000 square millimeters per US$.
The data density we’re talking about here is about a megabit per square millimeter, so 2 million square millimeters is about 250 gigabytes. Hard disks are a bit cheaper at this point, by a factor of maybe 4, but they weigh more and won’t last as long.
Gold boils at 2970°, occupies 19.30 g/cc, and consumes 12.55 kJ/mol of latent heat of melting and 342 kJ/mol latent heat of vaporization; its molar heat capacity is some 25.42 J/mol/K. At the 650-nm wavelength of red lasers, it’s about 98% reflective, which makes red lasers kind of a shitty way to boil it, but an arc in air deposits about 90% of its energy at the negative electrode (the one that’s releasing electrons and being bombarded with positive ions). Gold leaf can be as thin as about 100 nanometers thick or a bit more, but 200 to 400 nanometers and thicker is also sold. Its molar mass is 197.0 g/mol.
The per-mole quantities above work out to 63.7 kJ/kg latent heat of melting, 1.74 MJ/kg latent heat of vaporization, and 129 J/kg/K of specific heat. So boiling gold from something like room temperature should take, roughly, 383 kJ/kg of heating plus the latent heats, or 2.19 MJ/kg, or 2.19 J/mg. A cubic micron of gold weighs 19.3 picograms, so boiling it requires about 42.3 nJ. But our gold leaf is a fraction of a micron thick, perhaps 200 nanometers, so you only need a fifth of that; if you get the polarity right and the pulse quick enough, you need maybe 10 nJ per spark to blow holes in a sheet of gold foil, or about 500 nJ per laser pulse, which would be 1¼ microseconds of 400 mW — I hope that’s fast enough to keep the heat from conducting away.
The 250 gigabytes on 100 pages of 140-mm gold leaf would occupy, hypothetically, some 20 microns of thickness, if you laid them atop one another, for a volumetric density of about 10 exabits per cubic meter or 10 gigabits per cubic millimeter. But you can't do that because they would cold-weld together and then you wouldn't be able to separate them in order to read a particular page.
Gold is kind of a worst-case material in some ways — you lose a factor of 25 or 50 compared to any substance that is reasonably emissive in the relevant wavelength range, and it has very high thermal conductivity.
Nearly any other material will be a factor of 2 or more less dense, boil at a factor of 2 lower temperature, and have a factor of 10 or higher emissivity, have similar or lower heat of vaporization, and have worse thermal conductivity, so will be easier to blow holes in. Even aluminum has a third lower thermal conductivity of gold, though it has higher heat of vaporization.