Opportunities, Challenges and the Road Ahead for Field-Coupled Nanocomputing: A Panel Discussion - Field-Coupled Nanocomputing: Paradigms, Progress, and Perspectives - page 385

Bhanja: Most groups don't have the infrastructure to monolithically grow some-
thing like an MRAM device where you start with CMOS, then you add metal 1,
metal 2, metal 3, and then you anneal it, and then you put the magnetic logic
down, and then you do metal 4. We don't have that. It is scientifically possible,
but it is not economically possible. You also need the team.
Wang : So actually there is no technical problem to manufacture.
Bhanja: No, MRAMs are in market now. That's why as researchers we are
focusing on different things.
Anderson: So we have only a few minutes left. Any closing comments or
remarks?
Snider: Just a question. One of the things that made electronics so attractive
for the last forty five years is Moore's law, which is this nice exponential thing.
And an exponential function like that, as Gordon Moore realized, is a license to
print money. So that has driven the industry for forty years. Is there any sort of
exponential curve that you guys can see anywhere that would be an equivalent
sort of thing, that would be an exponential increase in productivity of something
even over a short term. (I don't think anybody is going to say forty years.)
Wo lkow: It's such a funny question. It's essentially saying that even if you
can make this fantastic new technology, it might not be that attractive from a
business point of view. It might be delivered and mature, but if it did not have
more and more stages to come it might not be that attractive. It is a funny
question, but I guess a very practical one. I do not know the answer. I think if
we made the first thing just work, and it was worth something to someone, then
they would buy it, and that would be the first generation. And then there would
be so many improvements, not in the way we follow Moore's law by making more
or less the same thing better, but there would be more refinements. So I think
it could display that trend for a good long while.
Lent: It seems to me that if you start at the limits of small, so we are talking
about molecular or atomic size, then what you scale first is the number of devices.
So you will be at the order of 10 24 in terms of device density. Seems to me that
is a reasonable scaling if you are trying to get more perfection, more control. At
those kinds of densities you can go quite a distance by just having more control
and perfection.
If the power dissipation is low enough, you could think about building them
up. I have always had a hard time seeing [management of] the power dissipation
in 3-D, but the answer is essentially the same answer in the orthogonal direction.
If the units are small, then all you can really do is to add more of them and do
it more perfectly and more reliably.
Wo lkow: There are twenty seven atoms smaller than silicon so we can go
through the periodic table. [Laughter.]
Snider: That goes to my followup question. Do we ask Bob [Wolkow] to write
the paper “We are out of room at the bottom”?