Physicists in Japan have shown how to make a semiconductor magnetic simply by applying a fairly modest voltage across the material at room temperature. Although the effect had been seen before, it previously required ultralow temperatures and massive voltages. Masashi Kawasaki of the University of Tokyo and colleagues say that their discovery could help to make MRAM memory chips more energy efficient, because a current would no longer need to be applied when writing data to the chips.
Magnetism plays a key role in much of condensed-matter physics, with metals such as iron and cobalt displaying permanent magnetism, or ferromagnetism, because the magnetic spins of their constituent electrons naturally align with one another. Semiconductors such as silicon, however, are instead paramagnetic, which means that their spins line up only when exposed to an external magnetic field.
But what Kawasaki’s team has done is to show that the semiconductor titanium dioxide, doped with about 10% cobalt impurities, can be transformed from a paramagnet into a ferromagnet (and back again) when housed in an electrolytic cell and a voltage applied across it.
In the absence of an applied voltage, the three spins inside each cobalt ion align with one another but there is no alignment between ions, as would be expected in a paramagnet. But with an applied voltage, extra electrons can enter the material, conveying information about the electron spins within the cobalt ions from one ion to the next. The spins of the ions now line up with one another, which in turn orients the spins of the mobile electrons in the same direction.
Although the chameleon-like ability to turn magnetism on and off in a semiconductor had been previously demonstrated by Hideo Ohno and colleagues at Tohoku University in Japan in 2000, using a thin-film semiconducting alloy, they were only able to do it at the ultralow temperature of 25 K and with a massive 125 V.
But by incorporating an electrochemical cell into a field-effect transistor, Kawasaki’s group has been able to add much larger densities of electrons to the semiconductor and so switch the material’s magnetism on and off at room temperature using a potential difference of just 4 V.
The researchers confirmed the presence of ferromagnetism in the doped titanium dioxide by passing a current through the material and measuring the voltage generated across it. The predominance of one spin direction over the other forced electrons to scatter more to the left than to the right (or vice versa), thereby generating a potential difference at right angles to their path (a phenomenon known as the anomalous Hall effect).
Kawasaki says that his team’s finding could boost the energy efficiency of MRAM memory chips, which consist of millions of pairs of tiny parallel ferromagnetic plates, with the plates in each pair separated by an insulator. The electrical resistance of these pairs is lower when the spins in each of the two plates are lined up in the same direction and higher when they are aligned in opposite directions, corresponding to a “1” and “0”.
Writing data to the chips involves altering the relative orientation of the magnetic spins, which is currently achieved by sending a current down a wire and exposing the plates to a magnetic field. If, however, the spins could be made to flip simply by applying a voltage to the pairs of plates then this flipping could be carried out at much lower energies, says Kawasaki. Although he has not yet actually managed to flip the spins in a semiconductor, Kawasaki says this is next on his agenda.
He admits that energy consumption in MRAM chips is not currently a huge issue, simply because this kind of memory is still not widely used, being much more expensive than other types of storage technology, such as flash memory. But he says that if prices come down and MRAM devices become more popular (they are fast and durable) then their hunger for energy will become a problem.
At that point, he maintains, it would make more sense to use semiconductors that can be made ferromagnetic by applying a voltage. “IBM is aiming at doing away with hard drives and using MRAM instead,” he says. “But its devices need a current to flip the memory.”
Other researchers also believe that the latest work may have significant practical pay-offs. Yuan Ping Feng of the National University of Singapore thinks it could “lead to technological applications in semiconductor spintronics”, while Igor Žutić and John Cerně of the State University of New York in the US, writing in a “perspectives” piece accompanying the paper, argue that the new ferromagnets could “help us make more versatile transistors and bring us closer to the seamless integration of memory and logic”.
Žutić and Cerně also point out that, unlike normal ferromagnets, in which heat is a problem because it tends to break up the spin alignment, the new material might actually enhance magnetism at higher temperatures because additional heat increases the number of charge carriers. “This could strengthen their role as magnetic messengers and could conceivably overcome the usual role of heat as the main foe of ferromagnetism,” they say.
The research is reported in Science 332 1065.