PORTLAND, Ore.—The ultimate memory chips of the future will encode bits on individual atoms, a capability recently demonstrated for iron atoms by IBM's Almaden Research Center in San Jose, Calif., which unveiled a new pulsed technique for scanning tunneling microscopes (STMs).
Pulsed-STMs yield nanosecond time-resolution, a requirement for designing the atomic-scale memory chips, solar panels and quantum computers of the future.
"My hope is that we can spawn a great following doing nanosecond time resolution and atomic-scale spatial resolution with their STMs," said Andreas Heinrich, a physicist in the IBM's Almaden Lab.
STMs, invented at IBM in the 1980s, have become the workhorse of the semiconductor materials industry. Their resolution extends all the way to the atomic scale, allowing individual atoms to be imaged. Unfortunately, STMs are slow at making such delicate measurements. Now IBM has perfected a new pulsed-STM technique that puts its ability to measure time on par with the nanoscale accuracy as its distance measurements.
IBM's pump-probe technique works in a manner similar to the way a pulsed laser works. First a pump signal is passed into the material from the STM tip to put the atom's electron spin in a known state, then after a waiting period a smaller probe signal is used to make a measurement. By repeating the process, each time extending the time between the pulses by a few nanoseconds, the process was able to accurately measure the electron spin relaxation time—or how long a bit of information is retained by a single iron atom.
Today's DRAM cells must have their bits refreshed every 50 milliseconds or so, but by using its new pulsed-STM technique, IBM has now observed that single iron atoms will need to be refreshed about every 250 nanoseconds—about 200,000 times faster.
Scanning tunneling microscope topograph of an iron atom (yellow) on a nitride-covered substrate (blue) which may someday enable single-atom bit-cells for memory ICs. Next to the iron are two more atoms and a missing atom defect in the nitride.
"We now know the answer to the question, 'What happens when you try to store information on a single iron atom?' And we hope that in the longer-term future we can make similar progress in answering questions about solar cell efficiency and quantum computers," said Heinrich.
The pulsed-STM technique will be adapted to measuring the efficiency of individual solar cells by using a light pulse as the pump to stimulate the solar cell, then probing it with the STM tip. Heinrich also hopes to reveal the inner workings of quantum computer gates using the pulsed-STM technique.
"If we can put quantum bits on surfaces so they have to interact with each other, then basically we will be showing a new way of performing quantum computations truly on the atomic scale. That's my vision of the future of quantum mechanics," said Heinrich.
Pixies: That STM will have to move very fast ;-)...agreed on IBM replacing Bell Labs as the only company worldwide that is doing fundamental research. I wonder how long this will last, I guess they see long term economical benefits...Kris
Colin, the readout circuit will be an STM operating under ultra high vacuum, that will be a huge boost to IBM's microscope business. :) That said, I think IBM has now replaced the defunct Bell Labs to be the leader in fundamental research.
You guys are absolutely right. In fact, I didn't include any calculations as to memory densities possible with a single-atom DRAM, because there is a lot more to it than just the bit cell material. But we must remember that IBM is currently just characterizing atomic-scale materials, not designing read-write mechanisms for atomic-scale bit cells. But as I've said before, characterization precedes realization, and its new pulsed-STM should facilitate the characterization of atomic-scale memory, photovoltaic and quantum-computing materials.
I agree with "selinz", this article is more about STM than memory implications. In all these futuristic discussion people seem to forget that you need to get a signal out somehow. So you build 10E15 (or 10E23 as "green_ee" asks) memory cells and then what??? Kris
Yes the long-term advantage of having nanoscale time resolution for STMs is the most important aspect of IBM's work, but their first example of its use revealed that DRAM bit cells based on a single iron atoms would have to be refreshed every 250 nanoseconds--which seems like an unreasonably short time. Nevertheless, characterization precedes realization, so this is at least a start toward single-atom bit cells for DRAM.
These are almost theoretical calculations, since even at the "ultimate" density of one bit per atom, you still have to have build support circuitry to set and read-out the bits. But now I'm remembering the clear polymer memory blocks used by the HAL 9000 computer in 2001: A Space Odyssey (film). I guess those could have held some kind of atomic lattice memory. What do you think?
IBM's recent characterization of an atomic-scale memory bit is the first step toward the realization of such atomically accurate semiconductor materials. The pulsed-STM essentially makes super slow-motion movies of semiconductor processes on the atomic scale. Check out how it works on this video at:
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.