As hard drives store data at ever-higher densities, their heads need to be able to resolve smaller magnetic domains.
© 2010 iStockphoto/janrysavy
The exponential growth in the complexity and performance of magnetic storage technologies mirrors that of computer processors, which was famously described by Moore’s Law. Unlike computer processors, however, magnetic devices face strong competition from alternative technologies, such as flash memory. To maintain the cost and capacity advantages of hard drives, physicists need to increase their areal density — bits per area — by 30% to 40% per year. And to retain competency, hard drives need to reach a density of 10 Terabits per square inch (Tb/in2) by 2015.
Using current technologies, hard-drive read heads are unlikely to scale past 2 Tb/in2. Many of the technologies under consideration as viable replacements share a common feature with existing read heads: the reading element is placed between two magnetic shields, helping the head to ignore nearby data bits that could confuse the signal from the bit of interest. Now, Guchang Han and co-workers at the A*STAR Data Storage Institute in Singapore have calculated that all shielded read head designs will also fall short of 10 Tb/in2, and have proposed an alternative shieldless read head design with superior scaling characteristics.
The alternative design is based on a ‘differential dual spin valve’ (DDSV) approach, by which electron ‘spin’ is used to reorient the magnetization of a thin magnetic layer. While existing read heads measure the strength of the magnetic field and therefore require shielding to ensure they are reading the correct bit, the DDSV responds only to gradients in magnetic field. Data are then encoded in the transitions between adjacent magnetic domains in a magnetic medium, rather than in the domains themselves as in conventional read heads. Surrounding domains will still produce stray fields as usual, but because these stray fields are uniform, the DDSV head can ignore them without the aid of shielding layers, according to the team. This in turn allows the DDSV head to scale to smaller sizes than conventional shielded designs.
Han and his team numerically evaluated the signal-to-noise ratio of data produced by a DDSV head when reading at a density of 10 Tb/in2, and considered both magnetic noise caused by thermal fluctuations and noise caused by spins exerting a torque on one another. They found that while the former is unlikely to be a critical issue, spin torque noise places a strict limit on the magnitude of the writing current that could be passed through the head.
“While the differential dual spin valve is a very promising design for high-density read heads,” says Han, “the spin torque effect will be a big challenge to its real implementation.”
The A*STAR-affiliated researchers contributing to this research are from the Data Storage Institute.
