PORTLAND, Ore. The design of atomic-scale magnetic memories got a boost Friday (Feb. 22) with an announcement from IBM Corp. researchers that they have manipulated individual cobalt atoms.
The first successful characterization of the force required to move magnetic atoms on a surface suggests that IBM's technique is a prelude to future bit-cells holding just a few atoms.
Even the densest magnetic memories use 1 million or more atoms to store a single bit. However, last fall IBM's Almaden Research Center (San Jose, Calif.) demonstrated how to measure a single atom's ability to store a bit, called magnetic anisotropy. That demonstration used a scanning tunneling microscope (STM) to send a current through a cobalt atom. However, sending a current tunneling through an atom using STM requires placing it on an insulating monolayer measuring only one-atom thick.
|Atomic force microscope tip (top) exerts force on single cobalt atom (yellow) to position it anywhere on crystalline surface lattice (bottom).|
Since then, IBM researchers discovered that such thin insulating layers will not work for storing bits on cobalt atoms. Instead, thick insulators are required that preclude the tunneling action of STMs. Hence, Almaden scientists had to devise a different method for moving cobalt atoms into place for future atomic-scale memories.
That new method, which IBM described Friday, involves using an atomic-force microscope (AFM) developed by IBM scientist Gerd Binnig and colleagues in 1986, the year he received the Nobel prize for developing the STM with fellow IBM scientist Heinrich Rohrer.
"The main reason we wanted to learn how to use an atomic-force microscope for these operations is because we want to build magnetic structures on insulators that are too thick to be using an STM," said IBM scientist Andreas Heinrich. "We have found that the magnetic nanostructures we want to build require a thicker insulator. So we have begun using an [AFM], which does not require a tunneling current.
"This should allow us to build magnetic storage structures using very few magnetic atoms--say about five atoms per bit," Heinrich added.
By comparison, magnetic domains on current hard disks are about 20-nm long with a track width of about 100 nm (2,000-nm2). IBM's stated goal for future atomic scale memories is 500-times smaller, just 2-by-2 nm (4-nm2). Such tiny magnetic domains require precise characterization of the amount of force needed to move individual cobalt atoms into place.
Moving an individual magnetic cobalt atom into place on a smooth platinum surface totals 210 pico-newtons; moving a cobalt atom on a copper surface takes just 17 pico-newtons. By comparison, it takes about 10 times as much force to break the chemical bond between two atoms in a molecule.
"If the motion of an atom is a part of the functional design of your device, like it was for our molecular switch, then we now have a much better idea of the forces required to make such as device work," said Heinrich. "We are no longer flying blind, because we have quantified how much force is required."
To measure the force, IBM teamed with Franz Giessibl at the University of Regensburg, Germany, who invented a technique that uses a tiny quartz tuning fork on the business end of the AFM, along with its normal tip. As the AFM applies force to move the cobalt atom, the tuning fork's frequency changes, allowing the scientist to read out the force required to move the atom.
The researchers said one surprise was that applying an upward force to reduce the load--akin to lifting a box in order to make it easier to slide it across the floor--had no effect at the atomic scale. Only lateral force could pry the atom loose, making it jump over to the next position on a cystalline lattice.
Hence, future equipment designed to move cobalt atoms in memory devices need only apply lateral forces, the researchers found.