Longitudinal Recording

Longitudinal magnetization lies along a track, while PMR magnetization points normal to the surface. The change involves different head and medium geometry, not simply a new software command. [pmr-hgst-2007]

This is an orientation category, not the opposite of CMR. Non-shingled versus shingled describes track overlap. Historical use of the word conventional for longitudinal recording must not be read as the modern CMR/SMR taxonomy. [pmr-hgst-2007] [toshiba-pmr-announcement] [zoned-smr]

A December 2004 PMR product announcement scheduled production for 2005; a November 2007 technical paper describes another company's PMR introduction in 2006. These milestones describe adoption of the successor technology, not a single worldwide end date for longitudinal HDDs. [toshiba-pmr-announcement] [pmr-hgst-2007]

The medium is a layer of weakly coupled CoCr-alloy grains, with chromium segregated to the grain boundaries and several hundred grains per bit cell for low noise. Smaller grains keep the signal-to-noise ratio as density rises but can reverse thermally, so KuV must exceed about 55 kBT to hold a signal for five years at about 340 K. [fu2006] Modal grain size fell from about 20 nm to about 9 nm in a decade, a volume ratio of about 0.09. [plumer2011]

A February 2005 technical report identifies demagnetizing fields and thermal instability as constraints on increasing longitudinal density. It explains a PMR head/media response, not a measured industry-wide investment shift or a universal date at which longitudinal recording stopped improving. [tanaka-2005]

About 9.7 nm is calculated for a cubic Co-alloy grain at 300K only if Ku is 2.5 x 10^6 erg/cc and KuV/kBT is at least 60 for 10 years. The 1998 10Gbits/in^2 design cited beside that calculation used grains of about 12 ± 4 nm, so 9.7 nm is not that design's grain size and is not a measured modal size. [lu2001]

A 2011 review estimates that KuV/kBT should be greater than about 60 at operating HDD temperatures of about 340 K for recorded transitions. That estimate is not a five-year rule of larger than 55 kBT and not a numeric areal-density ceiling. The same review says that prior to 2006-7 all HDDs used longitudinal media and that the longitudinal mode prevailed until 2005. Those are the review's dates, not a worldwide last shipment. Its modal sizes, from about 20 nm to about 9 nm, are for the past decade relative to the review. Ni/Fe Permalloy with a saturation magnetization of 1.0 T was replaced by Co/Ni/Fe alloys with 1.3-2.1 T and finally by CoFe having 2.4 T in the late 1990's. Fig. 2 shows simulated transitions at 788.8 kfci. [plumer2011]

Each bit must cover about 100 grains so the information is reliably stored, and below a grain-size limit the magnetization may spontaneously reverse from thermal energy even at room temperature. Section 1.2 states no numeric areal-density ceiling. The paper says perpendicular recording was introduced in 2006 and that HGST's first perpendicular drive was the Travelstar 5K160. That 2006 introduction is not a worldwide last shipment of longitudinal drives. The title page says November 2007; the closing page prints a 2012 copyright and says the sample specifications do not constitute a warranty. [pmr-hgst-2007]

Toshiba's 14 December 2004 announcement prints an areal density of 206 megabits per square millimeter (133 gigabits per square inch) for the new 1.8-inch drives. It says the 40 GB platter is 33% more than its then-current 1.8-inch 30 GB MK3006GAL, and it calculates capacity with 1 GB equal to 1 billion bytes. Those figures belong to the announced perpendicular drives and the comparison with that model. They are not a longitudinal-media density and not a last-shipment date. [toshiba-pmr-announcement]

Iwasaki's 2009 review says Figure 11 shows that the paper described a flat response up to 100 kFRPI, even for the thickest medium when the gap loss was compensated, and that the maximum linear recording density of an IBM 3370 hard disk drive at the time was only 10 kFRPI. The 100 kFRPI figure is that described response. The 10 kFRPI figure is the 3370. [iwasaki-2009]