Sign In

Communications of the ACM

ACM News

IBM Boasts Single-Atom Two-Bit Memory

View as: Print Mobile App Share:
Artist's rendering of the nuclear magnetism of a single copper atom.

IBM Research says it has magnetically encoded two bits on the nuclear polarization (spin) of a single atom.

Credit: IBM

IBM Research says it has magnetically encoded two bits on the nuclear polarization (spin) of a single atom.

In comparison, today's magnetic random access memories (MRAMs) require about 100,000 atoms to hold one bit.

Christopher Lutz and Kai Yang, scientists at IBM Research's Almaden facility in San Jose, CA, demonstrated how to change the magnetic spin of a single copper atom among its four possible quantum polarization states, essentially clearing the road to a new type of memory similar to spin-transfer torque (STT) magnetoresistive random access memory (MRAM), but which can be demonstrably reduced to the atomic scale.

Bulk copper is not known for its magnetic properties, but when individual copper atoms are isolated on an carefully chosen insulating substrate, the magnetic properties of their nuclei and electrons exhibit hyperfine interaction (the ability to align themselves), according to their research paper in Nature Nanotechnology titled Electrically controlled nuclear polarization of individual atoms.

IBM researchres explain how the strong nuclear magnetic resonance (NMR) of isolated copper nuclei can be controlled by an electrical current oscillating at a radio frequency and aimed at the atom's tunneling electrons. In a demonstration of this newly discovered mechanism, the controlling current was applied through the use of an earlier IBM invention (for which inventors Gerd Binnig and Heinrich Rohrer shared the Nobel Prize in Physics for 1986 with German physicist Ernst Ruska), the scanning tunneling microscope (STM). The STM was used to perform electron spin resonance (ESR, a way to study materials with unpaired electrons) by tuning in to the natural frequency (~1.5 GHz) of the copper atom's "nuclear antenna," as they put it.

"For the first time, we have performed NMR on a single atom's nucleus, controlling its magnetism among four states, using a STM. This breakthrough may one day lead to the development of extremely small magnetic devices with very low power consumption requirements," said Lutz. "This is related to the ever-increasing new field of nuclear spintronics—developing means of storing and manipulating information using nuclear spins in solids. It may also become highly useful to chemists, because single-atom NMR turns the nucleus into a nuclear magnetometer for measuring structures of molecules at the single-atom scale."

NMR is already used to uncover the structural properties of individual molecules at the atomic scale. NMR also is the central mechanism in magnetic resonance imaging (MRI), the technology used in non-invasive medical imaging devices.  The MRI makes use of the NMR principle by applying a very powerful magnetic field to "reset" the atoms inside the body into a single magnetic orientation, after which they return to the assortment of magnetic spin orientations in which the body's different parts are normally oriented. In the case of IBM's latest development, the RF electrical signal from the STM is applied to an orbiting copper electron to switch the atom's nucleus among the four-possible quantum spin states of copper (write), as well as to sense its current orientation (read).

To maximize the magnetic properties of IBM's material for NMR, individual copper atoms were deposited atop an insulating magnesium-oxide substrate.

To adapt this development for commercial use, this new type of atomic-scale STT-MRAM-like memory chip would have to be recast into a mechanism that can address each atom. The STM itself has sub-nanometer (angstrom-level) precision, capable of addressing individual atoms, but its individual atom addressability is too slow, bulky, and expensive for commercialization.

IBM Research's next step will be to improve the current STM mechanism by thickening the magnesium oxide substrate, as well as by experimenting with copper-based molecules, rather than individual atoms, which may be easier to manipulate and stabilize. IBM also aims to experiment with combining its newfound magnetic spin read/write capability with its already demonstrated ability to additively construct an entire electronic device atom by atom—aiming for atomic-scale addressable memory cells. It also plans to combine its new single-atom nuclear spintronics capability with the invention of atomic-scale nano-magnets it demonstrated last year as capable of storing a single bit per atom.

In addition to working toward atomic-scale devices for storing digital bits, IBM will also work toward storing quantum bits in the magnetic spin of these same atomic nuclei. This separate effort will run in parallel to its current IBM Q quantum computer development efforts, which stores qubits in the spin of electrons, which have a shorter coherence time (before collapsing into a digital bit) than atomic nuclei qubits.

"The technique described here does tap into quantum properties of copper atom nuclei, but this research, at its fundamental stage, is not part of IBM's current quantum computing strategy," said Lutz. "However, this research has implications for materials science, quantum physics, spintronics, nuclear memory, and quantum information processing. Being able to sense and control the magnetic orientation of a single nucleus in one of four directions may lead to the development of nanoscale devices that store and process information inside of atoms. Nuclear spins generally have much longer coherence times than the spin of electrons, so they are an attractive candidate for holding and manipulating quantum information. Since we sense and control atoms and their nuclei electrically, these results bring together the quantum properties of individual nuclei with the versatile and readily accessible world of electronics."

R. Colin Johnson is a Kyoto Prize Fellow who ​​has worked as a technology journalist ​for two decades.


No entries found