A UK team plans to develop a tiny sensor that can read the data from nano-scale magnetic circuits.
The researchers, atSheffield
universities, believe magnetic devices could replace silicon in some applications because they generally do not require power to retain data.
In some cases, magnetic nanotechnology devices may also offer higher device density, lower power consumption, improved reliability or additional functionality compared with silicon-based devices.
Chris Marrows, the principal investigator from Leeds, said his group hopes to develop a network of magnetic nanowires that could process information. The magnetic polarisation of various regions, or domains, of the wire would represent the binary numbers of digital information.
While there are many research groups around the world who have created similar magnetic nanowires, Marrows said all of them face a giant obstacle that keeps them from bringing the technology to commercialisation.
'The problem we're trying to solve in this project is how to get the magnetic information out of the circuit,' he said.
The researchers must develop a device than can read out the magnetic data in a form compatible with modern electronics.
'Dan Allwood [the principal investigator from Sheffield] can do that in the lab with a huge expensive instrument, but you need some way that you can do that on the chip to use these things as a technology,' he said.
The Leeds team is in charge of developing a tiny magnetoresistive sensor that would sit on the chip and read out the magnetic data.
The sensor would be similar to a hard disk read head. It would be made of a spin valve, which is a device that consists of two magnetic materials and a non-magnetic metal spacer layer in-between.
Marrows has already created such a device, but the new challenge will be to make one that is more sensitive and smaller.
'One of the problems is the magnetic field coming out of the wire is very weak so you need a very sensitive sensor,' he said. 'It also has to be extremely small. We have to fit our entire sensor into an area that's much smaller than a micron and very close to the edge of the wire we want to detect the magnetisation from.'
Another challenge is to make a sensor that can read out data at high speed. 'If you're sending a data stream along this wire and you want to read it out at any useful rate then you have to be able to detect magnetic events occurring at the wire at a timescale of a few nanoseconds,' said Marrows.
Yet speed is one major drawback of magnetic nanowire technology. The speed at which the magnetic wires are currently able to process data is extremely slow compared to silicon.
'Silicon runs happily at gigahertz — with magnetics you'd find it tough to do anything above 100mHz,' said Marrows.
That's why the researchers believe the magnetic technology would be better aimed at lower-end applications such as RFID tags where the cost of the technology is more important than its performance speed.
Marrows said the reason silicon-based technology is so expensive is because of its complex fabrication. An integrated circuit such as a complementary metal-oxide-semiconductor (CMOS) requires multiple levels of lithography, while a magnetic nanowire network could be created in one.
The magnetic network is built on a silicon wafer with a film of permalloy, a nickel iron magnetic alloy, deposited on top. The film is patterned into a network of wires.
The wires are shaped in a way that contain regions with different polarisations. Each region is called a domain. At the domain wall, which is where the regions meet, the magnetised direction switches.
'If you apply a magnetic field to the circuit, you can move the domain walls around and move the data through your network,' explained Marrows.
The project, which begins this autumn, was recently awarded a £500,000EPSRC
grant. It is also receiving industrial support from the global hard drive manufacturer Seagate Technology.
While the project is expected to last three years, Marrows said they could achieve results before then.
'We hope to demonstrate something that could be developed into a product in a couple of years time,' he said.