Longitudinal to Perpendicular Recording Transition (1975-2010)

Short answer: disk makers switched because longitudinal recording, in which each bit is a small magnet lying along the track, was running out of room to shrink. Shun-ichi Iwasaki of Tohoku University proposed standing the magnets upright, first presented the idea in 1977, and his students were on the teams at Toshiba and Hitachi that shipped the first drives. Toshiba announced the first perpendicular drives on 14 December 2004, Seagate announced shipments in January 2006, Hitachi GST announced its first in May 2006, and the IEEE milestone text says that by 2010 all production of conventional-recording drives had been replaced.

1975-1979: Iwasaki's idea. Iwasaki says he had found by 1975 that recording was limited because short, closely spaced longitudinal magnets repel each other and turn into eddy-like shapes that put less flux outside the medium, which lowers the read signal. An experiment on high-density tape, in which he applied a direct-current field perpendicular to the tape and restored missing read pulses, made him conclude that a perpendicular mode existed in which neighbouring bits attract. He says he decided to start research in 1976 and that the Japan Society for the Promotion of Science's Committee 144 on Magnetic Recording, set up in August 1976, became the open industry and university forum for it, with more than 190 bimonthly meetings by 2009. He found in 1975 that a cobalt-chromium film sputtered with columnar grains had perpendicular anisotropy, by luck, he says, because his laboratory was also doing magneto-optical recording research. He reported the work at the Magnetics Society of Japan's first topical symposium in May 1977 and at the Intermag conference in Los Angeles in June 1977. The paper he and Yoshihisa Nakamura published in IEEE Transactions on Magnetics (volume 13, September 1977) is the reference that the IEEE milestone cites. A 1978 paper on the field of a perpendicular head, a 1978 paper with Kazuhiro Ouchi on Co-Cr films, and a 1979 paper on a double-layer medium followed (IEEE milestone reference list). Iwasaki says the double layer, a Co-Cr film over a soft magnetic layer, raised write and read sensitivity about ten times over a single layer (Iwasaki), and the same combination of single-pole head and soft underlayer is what the IEEE milestone text calls indispensable. Hitachi's 2005 press release quotes Iwasaki as saying that around 1975 he began to feel the vertical direction was right for high density.

Why it was not adopted in the 1980s. Iwasaki reports that the first prototype, a flexible-disk drive with a single-pole head and the auxiliary pole on the far side of the disk, proved the principle. He also says that at the 1978 Intermag conference Dennis Speliotis pointed out that single-pole heads read poorly, which he addressed by showing that read output rose with write sensitivity in the double-layer medium. The IEEE milestone text says the industry took a strong interest after the 1977 paper, and that perpendicular recording was seen as the next-generation disk technology in the 1980s, citing a 1981 IEEE Spectrum article. Iwasaki says venture companies for perpendicular recording started in the US around 1982. Then, in his account, IBM proposed a magneto-resistive read head that made thin longitudinal media readable, a strong majority followed it, and publications on perpendicular recording show a deep valley in the mid-1990s that he calls the "Death Valley". The IEEE milestone text agrees there was a long halt during the 1990s, and IBM's 2001 release says IBM introduced the first giant magneto-resistive read element for disk drives in 1997. The mechanism of the detour is Iwasaki's own, and no independent source opened here corroborates his explanation of why the industry followed IBM.

The thermal wall. The decisive argument, Iwasaki says, came from an American scientist who insisted that longitudinal recording had a physical limit: a thin medium loses its magnetisation to thermal fluctuation, and the quantity that matters is the product of the anisotropy constant and the grain volume relative to thermal energy. The IEEE milestone text states the same physics, that the bit's magnetic energy is anisotropy energy times bit volume and that smaller bits are more exposed to thermal agitation. IBM researchers D. Weller and A. Moser reported in 1999 that the thermal decay in longitudinal media set in at a stability ratio (KuV over kBT) of about 35 plus or minus 2, that the requirement rises to about 60 for a projected 40 gigabits per square inch drive with ten-year storage, in a paper that reviews the prediction by Charap and colleagues (1997) that thermal effects would limit longitudinal recording at about 40 gigabits per square inch (abstract as retrieved). Their abstract lists perpendicular recording among the options that may shift thermal effects to higher densities, along with squarer bits, better signal processing and patterned media. IBM's 2001 release says the superparamagnetic effect had long been predicted to appear at 20 to 40 gigabits per square inch, near the density of products at the time. The sources therefore agree that a limit was expected in the tens of gigabits per square inch, and they disagree on where it finally fell: Iwasaki puts the longitudinal ceiling at about 100 gigabits per square inch, and Hitachi's April 2005 release says longitudinal recording loses its ability to keep data integrity much beyond 120 gigabits per square inch and was within two product generations of its practical limit (vendor claim). A Hitachi GST white paper of 2007 adds a rule of thumb, that each bit must cover about 100 grains, and that grains cannot be made smaller without risking spontaneous reversal.

The detour: antiferromagnetically coupled media. Before perpendicular drives shipped, IBM stretched longitudinal media. Its release of 21 May 2001 describes antiferromagnetically coupled (AFC) media, informally called pixie dust: a ruthenium layer three atoms thick between two magnetic layers makes them orient in opposite directions, so the stack acts as thinner for writing and reading while its total volume stays large enough to resist thermal decay. IBM said AFC media was shipping in Travelstar notebook drives at up to 25.7 gigabits per square inch and was expected to allow 100 gigabits per square inch by 2003 (vendor claim; a projection, not a measurement). The American Institute of Physics credits Eric Fullerton and the IBM media team, says AFC was first used commercially in IBM Travelstar drives in 2001, and says Fullerton later developed laminated AFC media at Hitachi GST. IBM also says its scientists discovered in 1990 that ruthenium gives the strongest antiparallel coupling of any spacer element and that the same structure was in IBM's first giant magneto-resistive read element of 1997. AFC was a way to keep a longitudinal medium thermally stable, not a step toward perpendicular recording. Iwasaki separately says a thin ruthenium interlayer is used in perpendicular media to stop exchange interaction between the recording layer and the soft underlayer, which is a different use of the same element. No source opened here says how far AFC pushed the highest longitudinal density; Hitachi's April 2005 statement that its 230 gigabits per square inch perpendicular demonstration was a doubling of today's highest longitudinal densities implies roughly 115 gigabits per square inch (this card's arithmetic).

2000-2005: from prototype to first product. Iwasaki shows a Hitachi prototype 2.5-inch perpendicular drive of 2000 at 52.5 gigabits per square inch, built by Hisashi Takano and colleagues who studied in his laboratory, and the IEEE milestone text says a perpendicular prototype drive was demonstrated at the Intermag conference in 2000 by Hitachi and Tohoku University, with the industry resuming development around 2000. Iwasaki credits Yoichiro Tanaka's group at Toshiba for a CoPtCrO/Ru perpendicular medium with a square hysteresis loop and grains of 5 to 10 nanometres. Toshiba's press release of 14 December 2004 announced what it called the world's first perpendicular-recording hard drives (vendor claim): the 1.8-inch MK4007GAL with 40 gigabytes on one platter and 5 millimetres thickness, and the 80 gigabyte MK8007GAH, at 133 gigabits per square inch, which is 206 megabits per square millimetre, and 33 percent more per platter than the company's MK3006GAL. Mass production was planned for April to June 2005 for the 40 gigabyte model and July to September 2005 for the 80 gigabyte model. Iwasaki says Toshiba announced the shipment of a music player with a perpendicular drive in May 2005, and the IEEE milestone reproduces a headline from the Asahi Shimbun of 21 May 2005 on commercialisation. Hitachi GST's release of 4 April 2005 says its testers had been using computers with perpendicular drives since December 2004, that it had demonstrated 230 gigabits per square inch, that it expected to ship a first perpendicular product in 2005 and that mass-market adoption would gain momentum in 2006. The Hitachi GST white paper of November 2007 describes its field test as 2005 and as first-generation technology, with a year-long field test before the Travelstar 5K160.

2006-2007: the first wave of products. Seagate announced in January 2006 that it had started shipping the Momentus 5400.3, a 2.5-inch 5,400 rpm drive of up to 160 gigabytes at 132 gigabits per square inch, which it called the first 2.5-inch notebook drive built on perpendicular recording (vendor claim). On 26 April 2006 it announced the 3.5-inch Barracuda 7200.10 family, 200 to 750 gigabytes at 130 gigabits per square inch and up to 188 gigabytes per disc, and said it now offered perpendicular recording across its desktop, notebook, enterprise, consumer electronics and retail drives (vendor claim). Computerworld reported in May 2006 that Hitachi GST was about to sell its first perpendicular drives, the 2.5-inch Travelstar 5K160, up to 160 gigabytes against up to 100 gigabytes for the longitudinal 5K100, and quoted chief technologist John Best saying prototypes had been in the field for one and a half years with no returns (vendor statement). Iwasaki lists Seagate in January 2006, Hitachi GST in May 2006 and Fujitsu in December 2006 as the next entrants after Toshiba. The Hitachi GST white paper says that in January 2007 the company announced the Deskstar 7K1000, which it called the first terabyte hard drive (vendor claim). Hitachi's white paper also notes that perpendicular read-back waveforms differ from longitudinal ones, with every frequency component shifted 90 degrees in phase and more low-frequency energy, so the read/write electronics had to change as well.

Completion. The IEEE milestone text says that by 2010 all drive production with conventional recording had been replaced by perpendicular recording and that shipments of perpendicular drives reached 600 million units in 2013 (a milestone nomination claim without a table in the text opened here). Iwasaki estimated in 2008-2009 that about 70 percent, or about 75 percent in the abstract, of about 500 million drives shipped that year would be perpendicular, and he lists 360 million drives shipped in 2005. Those are his estimates from conversations with manufacturers and are not measured shipments.

What the sources do not settle. The year of the idea is given as 1975 by Iwasaki and Hitachi's quote, 1976 for the start of dedicated research and 1977 for the first presentation, and the IEEE milestone dates the invention to 1977. Which product was the first perpendicular drive depends on definition. Toshiba's December 2004 announcement is a vendor claim, the sources opened do not give an exact first shipment date for the MK4007GAL, and Seagate's and Hitachi's claims are about their own product classes. Hitachi's April 2005 release attributes the first perpendicular sound recording to Valdemar Poulsen in the late nineteenth century, while Iwasaki treats the steel-wire era as longitudinal; this card adopts neither claim. Iwasaki, the IEEE milestone and the Toshiba and Hitachi pages overlap in authorship, since the milestone cites Iwasaki's review, and some of Iwasaki's laboratory alumni led the Toshiba and Hitachi teams, so they are not fully independent.