A Silicon Spintronic Memory That Lasts

10 January 2011—In what could prove to be a big development in the nascent field of spintronics, scientists have succeeded in storing information for almost 2 minutes using a magnetic property of phosphorus nuclei embedded in silicon. For this property, called spin, 2 minutes is an eternity. The breakthrough could lead to new kinds of silicon-based memories that might even work at the level of a single atom.

Unlike conventional electronics, which uses an electron’s charge-carrying property to create circuits, spintronics exploits the quantum mechanical property of electrons known as electron spin to create useful devices. (Electron spin is a form of magnetic moment that causes electrons to act like bar magnets.) However, electron spins typically have short lifetimes (microseconds), which make it challenging to create registers and other devices necessary to perform calculations, since those actions require that the information be stored for a relatively long period of time.

Now, a team of physicists led by Dane McCamey of the University of Sydney, in Australia, has succeeded in using the magnetic spins of phosphorus nuclei in phosphorus-doped silicon to store information for 112 seconds. Basically, it is a clever technique that allows the researchers to map and store the electronic spin information onto the nuclei of the phosphorus donors. The nuclear spins can be read out electronically repeatedly, and the information lasts much longer than the lifetimes of electron spin.

"Using silicon was a great way to do this, as donor nuclei in silicon have a well-understood interaction with electron spins, have long lifetimes, and the material is compatible with conventional electronics," says McCamey, who worked with Christoph Boehme at the University of Utah, Johan van Tol at Florida State University, and Gavin Morley at University College London.

It was not easy to achieve storage for this length of time, however, and doing so required the use of specialized equipment at extremely low temperatures at the National High Magnetic Field Laboratory in Tallahassee, Fla. It required an 8.9 tesla magnetic field (almost 200 000 times as strong as Earth’s magnetic field) to align the spins of the phosphorus electrons in a wafer of doped silicon chilled to 3.5 kelvin.

Two-hundred-forty-gigahertz electromagnetic pulses "wrote" spins onto the electrons surrounding the phosphorus atoms. Then, FM-range radio waves mapped the information stored in the electrons’ spins onto the phosphorus nuclei. After almost 2 minutes, the spins were mapped back onto electrons and then read out. McCamey says the technique also works for nuclei other than phosphorus, which extends the technique’s usability.

Experts are impressed. "What is most important is that they are measuring the nuclear spin state with an electrical current—it is a semiconductor device. That is what sets this work apart," says Stephen Lyon, a professor of electrical engineering at Princeton. He says it is impossible to know whether this technique will eventually make its way into the room-temperature electronics that we all use, "but it is a step in that direction."

He points out that while McCamey’s experiments were done at a low temperature, "one needs to remember that giant magnetoresistance (GMR) started out as only a low-temperature phenomenon, and now it is in all of our disk drives." Indeed, the researchers’ next goal is to find a way to make the memory work at warmer temperatures and with weaker magnetic fields.

Lyon also thinks that McCamey’s results could be useful for quantum computing. Many researchers have been intrigued by a still theoretical quantum computer concept that would use phosphorus nuclei in silicon to store and manipulate quantum information. For McCamey’s technique, "if we can get the sensitivity up to where we can measure a single nuclear spin, it could be the readout device of a quantum computer," Lyon says.

Andrew Dzurak of the University of New South Wales, in Australia, who was part of a team that recently read out the spin of a single electron in a silicon chip, agrees. "I think that the work has considerable potential for the development of spintronic devices in silicon," he says. "Electrical detection is important if you want to make large-scale integrated spintronic devices."