Arrays of atoms could make faster, cheaper memory devices .
Memory chips that store data by using electrical pulses to rearrange atoms could revolutionize the next generation of mobile phones and digital cameras. So say researchers who have built a device that proves the idea can deliver faster, cheaper memory.
Computers use a binary code to store their information in capacitors that can hold electrons in two distinct states, like a switch that can be either 'on' or 'off'. But since electrons can leak out, each capacitor must be recharged thousands of times every second. And if the power supply dies, so does all your data.
Now Martijn Lankhorst and his colleagues at Philips Research Laboratories in Eindhoven, the Netherlands, have shown that instead of using electrons, it's possible to create two states using an ordered or disordered arrangement of atoms.
They use a material called antimony telluride, which starts off in an 'amorphous' state, with all its atoms jumbled up. But a small pulse of electricity provides enough heat to make the atoms line up into rows, creating an ordered, crystalline arrangement.
A second, higher-voltage pulse melts the crystalline structure, resetting the material back to its jumbled state. A computer could tell the difference between the two because the crystalline phase has a much lower electrical resistance.
“Imagine you could start your laptop and have it ready for you to work in less than a second. Matthias Wuttig , RWTH Aachen University, Germany”
Wiring lots of tiny pieces of antimony telluride together would create a memory chip that could store information in a stable way, without having to be continually charged up.
The approach has huge potential, says Matthias Wuttig, a materials scientist at the RWTH Aachen University in Germany. "Imagine you could start your laptop and have it ready for you to work in less than a second," he says, "or that you were able to record and watch full-length movies on your mobile phone."
The idea isn't new - Stanford Ovshinsky first proposed the concept for such 'Ovonic' devices in 19681. But it has taken researchers until now to find a material that can reliably change states millions of times without degrading, and to develop the techniques needed to wire such tiny components together.
The research is published online in Nature Materials2.
Flash memory is another attempt to solve the same problem. It too retains its data indefinitely, and is used in digital cameras and memory sticks.
It works by using many layers of mineral oxide, which are either full or empty of electrons. But flash memory sticks are an extremely expensive way to store large amounts of data. Each layer has to be individually wrapped, again to stop the electrons leaking out. It is also tricky to miniaturise them further because quantum effects start to interfere with their reliability.
Ovonic memory devices could work much better. Antimony telluride is relatively cheap and easy to use, and actually seems to perform faster when the device is miniaturised.
Wuttig says that a memory cell in a conventional flash memory device cannot be made smaller than about 65 nanometres across, whereas Ovonic memory cells could potentially get down to 10 nanometres. That could be enough to put Ovonic technology at the head of the field for the next two generations of electronic devices, he says.
But the key selling point is that the memory cell is remarkably simple to make, since it is essentially just a chunk of material hooked up to two electrical contacts. "I've spoken to a couple of companies and they're thrilled about this," says Wuttig.
There are still several hurdles to be overcome before the technology finds its way into your mobile phone, however. For example, shrinking the memory cell to even smaller scales could allow amorphous areas to become crystalline at much lower temperatures, so the data might get mixed up by accident.
OvshinksyS. R., et al. Phys. Rev. Lett. 21, 1450 - 1453 (1968).
LankhorstM. H. R., KetelaarsB. W. S. M. M. & WalktersR. A. M. Nature Materials published online, doi: 10.1038/nmat1350 (1968).
RWTH Aachen University, Germany