Nice idea, but this instance would still need to follow the laws of
chemistry. Atoms don't form generic, linear chains like that by
themselves, and they wouldn't be particularly stable (energetically,
electronically, mechanically, etc) if you forced them.
Think of the capacity. We could mimic and surpass the storage capacity of DNA (base 4).
DNA is base four if you look at only one strand. Because DNA is
redundant, with two strands, though, each pair of nucleotides can take
on four states, so I have a hard time classifying it as truly base
four, in my mind. (Yes, I am aware of the single-stranded nature of
RNA, but RNA is not a permanent storage form of information ni the
same fashion as DNA.)
There may be more material-efficient free-standing strands than DNA,
but if there are, I am not aware of them, and to design them from
scratch would be... something of an undertaking, to say the least.
However, there are many many more practical schemes for molecular or
atomic memory schemes, as long as you allow for infrastructure to
support those smallest components. For instance, UCLA researchers can
flip single electron spins inside transistors now: Technically, that
is a form of memory dependent on a single electron if you view it one
way, but it also depends on the structure of the integrated circuit
transistor which contains the electron viewed in another way.
Similarly, IBM researchers can selectively charge gold ions by one
atom and detect the change, but a bunch of free floating gold ions
isn't useful without some substrate or matrix to sit on. IBM (in a
different breaktrough) can image molecules with sufficient precision
to map out the "locations" (scare quotes required due to the nature of
quantum mechanics) of particular electrons within.
I know of at least one nanotube scheme which is claimed to be able to
store one bit per tube, and I would be positively shocked if there
were not more, or if they were limited (by nature, not by current
research) to bi-state devices.
Then there are molecules like rotaxane, which act as molecular
switches, changing properties (in this case, mechanical) depending on
the presence or absence of an electron. There's a nanowire scheme
where a nanowire, augmented with a molecular coating of sorts, can
store up to three bits per wire based on the conductance changes that
result because of the oxidation/reduction chemistry that occurs during
the application of negative voltage. (I like that one. It's
But there are some general points to be made:
First, these are all laboratory toys. Despite every researcher's
ardent plans to bring these things to market, few if any of these will
really succeed. *Some* form of molecular storage will succeed, almost
certainly, but how many different formats do we need? More than one,
but probably not terribly many. (The ones easiest and most
efficiently retrofitted to present technologies will be the ones to
make the cut.)
Second, related, some of these are researchers with more hype than
real hope. A good way to tell is the buzzword count-- in these
schemes, a great many are described breathlessly as avenues for
quantum computation. That sets off the same flags in my mind that "A
potential cure for cancer!" sets off in John Larkin's mind.
Third, each and every one of these methods, as I mentioned before,
uses molecules or atoms as the primary elements of memory or storage,
but in order to make them generally useful, they still rely on
substrates or infrastructure of supporting material. A bunch of
rotaxanes, for instances, still need to be held in a regular array
of some sort, so that the read/write equipment (which is its own
equipment, of course) knows where to look when performing its
operation. You can do a lot of things in theory, as lab toys--
repeating them a billion times and using all billion of them together
is quite another issue.