Original URL: http://www.theregister.co.uk/2008/09/30/hdd_areal_density_improvements/

How many terabytes can you fit on a 2.5-inch hard drive?

Fun with areal densities

By Chris Mellor

Posted in Storage, 30th September 2008 16:55 GMT

Can we expect 2.5TB 2.5-inch hard drives and 5TB 3.5-inch drives by 2012? It seems realistic if the claims of hard disk drive toolmaker MII, Hitachi GST, and others are realised.

To reach these levels, platter areal density needs to increase and read/write head capabilities also need to improve. Current areal density mass production drives are in the 250Gbit/sq in areal density area, but best practices are much higher.

Toshiba has a 1.8-inch drive with 250GB from one platter and two heads - that's 125GB/platter surface and an areal density of 378.8 Gbit per square inch. Seagate has a generation 6 Momentus 2.5-inch drive rated at 500GB, using 2 double-sided platters with 125GB/surface and a Seagate-claimed areal density of 394Gbytes/sq in. This should probably be 394Gbit/sq in.

Let's say current best areal density practice in production HDDs is 378-394Gbit/sq in with 4th generation perpendicular magnetic recording (PMR) technology.

Areal density improvements

The PMR problem is that the technology is running out of steam and cannot continue delivering 50 to 100 per cent or more annual increases in areal density. People can't see it progressing beyond 1Tbit/sq in, and it may not even attain that.

Several sources have contributed to a picture of the HDD industry using new recording techniques involving Discrete Track Recording (DTR) and Bit Patterned Media (BPM) to progress beyond fifth and possibly sixth PMR generations and into the 1Tbit/sq in plus areal density area, meaning HDD capacities three times higher than current levels.

A product manager in a firm using HDDs to build systems said his information was that he expects to be using 750GB 2.5-inch drives by mid-2009 and 1TB units by early 2010, if not the end of 2009. No particular supplier was mentioned but Fujitsu and Western Digital are current suppliers while Seagate, Hitachi GST, and others are qualified. A Western Digital spokesperson said the firm viewed these capacities and dates with incredulity.

Hitachi GST has been quite open about its wish to match and surpass Seagate and has revealed information about both its media and its head technology research efforts. It is working with University of Wisconsin researchers and has attained terabit/sq in areal densities using patterned media, and lithographic tools to pattern magnetic domains and a self-assembling block co-polymer to divide each track into quarters.

Wachovia analyst Aaron Rakers has written about Xyratex, which makes equipment used in HDD manufacturing processes. In early September, Rakers wrote: "Xyratex did note that the HDD industry looks to face some meaningful technology challenges beyond PMR with technologies such as Discrete Track Recording and Patterned Media over the longer-term...The company believes these technologies will not materialize until 2011-2012 timeframe - though some R&D focused equipment could begin to ship as early as 2009."

The Xyratex picture is of HDD suppliers researching DTR and BPM techniques up to and including 2010 with product shipping in the 2011/12 timeframe.

Disk platter tool supplier Molecular Imprints Inc., of Austin, Texas, has been prolifically vocal about next-generation HDD production technology using nano-imprint lithography. It has received orders for its Imprio 2200 tool from two HDD vendors for DTR pilot production purposes, one of them Hitachi GST, and seven R&D Imprio 1100 tools have shipped, at least three to HDD suppliers. Obducat AB of Malmo, Sweden has also received an order for its production-ready Sindre HDD tool from an HDD media supplier plus orders for R&D-level tools.

Nano-imprint technology is used to define discrete concentric tracks (DTR) on a platter surface. The BPM technology involves creating nano-scale pillars of magnetic material, around 12-20nm high, on a disk platter's surface. The pillars are surrounded by insulating material. It is thought that DTR will arrive first followed by BPM.

The imprinting equipment is the same in each case, it being the template used for imprinting which changes. The template is the size of the target HDD platter and is pressed against it to imprint a pattern into both of its coated surfaces, much like the way a vinyl long-playing record was made.

Paul Hofemann, a marketing and business development VP at Molecular Imprints, has said: "Patterned media is a huge transition for the HDD industry. The first adoption will be in mobile for the 65mm disks ... They know they will soon run out of the different tricks used to increase the aerial density with the unpatterned substrates. That is the reason why patterned media will come in, because everybody wants to stay on this track of 100 per cent annual increases in aerial density." I think this 100 per cent areal density increase figure is a tad aggressive and have used a 50 per cent annual increase in estimates that follow.

Hofemann thinks HDD and media suppliers will engage in pilot production this year and next year, then move on to volume production in the 2010/11 period. This agrees with the Xyratex view.

Beyond DTR and BPM is Heat-Assisted Magnetic Recording (HAMR) with Seagate thinking it might attain 10Tbit/sq in with that technology. Seagate sources have indicated that it might use HAMR before BPR. HAMR will involve large changes in head technology to provide the heat needed for writing data.

Hofemann has said: "Most present technology extensions are running out of steam, and going to 1Tb/in.2 won’t be easy," Hofemann said. "We’re at about 700-800Gb/sq in, and these extensions are hitting limits. Discrete track media will get us past 1.5Tb/sq in, and then it’ll be necessary to go to bit patterning. Heat-assisted recording theoretically could work at a 10-30Tb/sq in level, taking us to the end of the decade."

Head density improvements

Hitachi GST has said it hoped to achieve a 2.5 TB 3.5-inch drive in the 2010 period with existing tunnelling magneto-resistive (TMR) head technology. A TMR head will be given a Wrap-Around Shield (WAS) to prevent the head's own magnetic forces corrupting perpendicularly recorded data on adjacent tracks to the read track. Then it expects to move to CPP-GMR (Current-Perpendicular-to-Plane Giant Magneto-Resistance), a more advanced technology, able to deliver recording densities of 500Gbit/sq in to 1Tbit/sq in.

This means, Hitachi GST avers, a 625Gbit/sq in areal density would be practicable instead of today's 250Gbit/sq in, giving us a 2.5TB, 3-platter, 3.5-inch drive by 2010/11. The next-generation CPP-GMR heads would take Hitachi GST through to terabit areal densities, still using PMR media, Hitachi GST says, and a 5TB, 3-platter, 3.5-inch drive by 2013.

These numbers are against a base line of a supposed 250Gbit/sq in areal density today. As we have seen, advanced notebook drives are already way beyond this, at 380GBit/sq in or higher. But they rotate more slowly than enterprise 3.5-inch drives and thus can pack their data bits closer together. They would meet the PMR limit sooner and so move to DTR sooner. They are also facing burgeoning competition from solid-state drives (SSD) - Toshiba has just introduced a 250GB SSD in the 2.5-inch form factor.

Applying the same 2.5 times increase to their areal densities gets us to a 1.25TB 2.5-inch drive in 2010. What happens if we apply a 50 per cent annual increase in areal density to today's highest-capacity 1.8-inch, 2.50inch, and 3.5-inch drives?

The probability seems to be that we will get a fifth PMR generation next year, a sixth in 2010, the first DTR drives in 2011/12 with BPR appearing possibly in 2013/14. All this is very speculative and individual manufacturer's mileage may vary, but a kind of consensus is emerging that both and head technology developments will enable us to bypass the areal density roadblock approaching us as HDD suppliers push PMR densities higher. Whether the 2010 and 2012 HDD capacity points above are realistic only time will tell. ®