Ok, so i'll be the first one to bite. What does this device actually do?
To me, it looks like a two terminal tunneling device with electrodes at two different work functions. The difference in work function means that the I/V behavior will be asymmetric, so it could show a weak diode like behavior, but probably still a lot different to what we are used to from semiconductor diodes.
Since there is no semiconductor involved, switching delays caused by minority carrier recombination (diffusion capacitance) should be absent. That is probabably the reason for the claim for speed. However there could be a disadvantage due to the extremely high capacitance of the MIM structure.
There are some other drawbacks too: The way quantum tunneling behaves in such a two layer devices means that there are some second order slope changes in the I/V behavior. The material used (Al2O3 and HfO2) are also prone to electron trapping. This means that there are most likely additional conduction mechanisms, which will give rise to device leakage at low bias, dielectric relaxation (1/t decaying current after switch) and noise.
So here are my questions:
What is the exact application of this device?
What are the relevant figures of merit that are optimized in respect to semiconductor diodes? How does it compare in the other ones?
How is this supposed to "beat CMOS" or lead to transistors? My understanding so far is, that this is a two- not three terminal device. So it can not replace a transistor.
This is a major technological achievement. Hope they work more to make it apply to practical devices. It will interesting to know quantitative comparision - gain in speed, saving in power, voltage level required to operate it.
Blog Doing Math in FPGAs Tom Burke 15 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...