Site: Sony Headquarters
(magnetic storage presentations)
Corporate R&D Strategy Department
6-7-35 Kitashinagawa, Shinagawa-ku
Tokyo 141-0001, Japan

Date Visited: 10 March 1998




Sony is a world leader in electronics and entertainment with 163,000 employees that in 1997 generated $45.7 billion in revenue (29% in the United States), and $1.1 billion in net income. Traditionally, Sony has been a leader in audio and video products and now has a new focus in digital and network technologies. Sony is a leading player in magnetic tape technology for data storage, participating in several formats including QIC, TRAVAN, DDS 19 mm and 8 mm. The R&D budget is $30 million, with some magnetics R&D carried out at the Yokohoma Research Lab, but the major effort is in the Sendai location.


High Data Rate Tape Recorder (Takayama)

Dr. Takayama pointed out that a data rate of 1.2 Gb/s is required for HDTV broadcast applications which will be compressed to 50 Mb/s for home use. He described two existing products:

  1. A 1" open reel recorder which achieves 1 Gb/s user data rate using 18 m thick, 1,500 Oe MP tape with 8 record, 8 reproduce and 2 erase conventional inductive heads on a 355 scanner. Track pitch is 37 m and the shortest recorded wavelength (2 bits) is 0.69 m (48 Mb/in2) with a head/tape speed of 51 m/sec.
  2. A 19 mm cassette recorder which achieves 512/400/256 Mb/s using 16/13 m thickness 850 Oe oxide media with 16 record and 16 reproduce heads to compensate for the smaller wrap angle of 180. The track pitch is 45 m, and the shortest recorded wavelength is 0.9 m (30 Mb/in2) with a head/tape speed of 40/31/20 m/sec. To push the user data rate and density to 1 Gb/s and 60 Mb/in2, double the number of heads to 32 each for record and reproduce, and halve the track pitch to 23 m. Increasing the writing speed was felt to be more difficult. To further increase the user data rate to 2.0 Gb/s, the media will be changed to 1,500 Oe MP with a track pitch of 23 m and a shortest recorded wavelength of 0.45 m (120 Mb/in2). Doubling the number of heads again to 64 each for record and reproduce was considered impractical.

An alternative considered was a multi-track linear serpentine system using the same media at 6 m/sec. This would require 100 channels/head for read/write/read and was felt to be impractical both because of expected low yield of the head and the need for 600 head connections.

The conclusion would seem to be that multiple heads in a helical system is a more attractive solution for data rates in this range.

MR Heads in Helical Tape Systems (Ozue)

Mr. Ozue showed that in 1991 the areal density of HDDs exceeded tape storage for the first time at 100 Mb/in2 and has continued to increase at a faster pace. To remain competitive, the rate of areal density increase in tape must parallel HDDs and reach 1 Gb/in2 in 2000. To accomplish this, Sony plans to continue to focus on helical systems but to adopt all the significant advances in HDD including thin film media, thin film record, MR/GMR reproduce heads, PRML or other advanced signal channels, and in-contact recording. An experimental system was described using a SAL/AMR reproduce head with NiZn ferrite shields, track width of 17 nm, azimuth of 25 on 100 nm thick, 1,500 Oe Co-O evaporated tape with a DLC overcoat, Mrt = 2.6 memu/cm2, and a squareness of 0.8. It realized 12 dB more signal at a wavelength of 0.5 m (141 Mb/in2) than an inductive head. Sony hosts were optimistic that the target of 1 Gb/in2 in 2000 would be achieved.

Wear of thin film heads and media is a major concern in the industry. Mr. Ozue said, however, at a wavelength of 0.5 m no signal loss was detected after 1,000 reads. The total magnetic spacing is 50 nm. The head is apparently designed with a deep throat height similar to ferrite heads, which is expected to accommodate some head wear. How this affects performance was not discussed.

Tape Media Technology (Onodera)

Mr. Onodera summarized recent advances in both MP and ME tape. Sony is using 80 nm MP particles in 100 nm thick coatings and experimenting with 60 nm particles. Dispersion is the major technical challenge to achieve low media noise. A 200 MB floppy was expected to be announced the end of the summer (1998). Sony researchers believe the reliability problem associated in the past with ME has been solved using a DLC overcoat. Helical systems require recording only in one direction, so only a single angle-of-incidence layer is needed. However, Mr. Onodera speculated that two layers might be used to reduce noise. The cost differential between MP and ME will be narrowed as the reduced moment needed for MR heads can be achieved with thinner films, resulting in higher web throughput.

ASET Research Program (Miyauchi)

The ASET program is sponsoring work at Sony on a vertical GMR head, which was reviewed by T. Miyauchi. A comparison through simulation of horizontal and vertical heads including the bottom conductor height revealed several important points. The output for both types depends on the flux decay length lambdaB, which defines the length of the sensor over which the field from the media is effective. This is particularly important for vertical heads where lambdaB depends on e1/2 where e = 4piMs/He is the effective permeability in the sensor, Ms is the saturation magnetization of the sensor and He is the effective anisotropy field including the induced anisotropy and the bias field from the permanent magnet stabilization field. For a gap length of 0.13 m, lambdaB = 0.55 m for e = 1,000 and decreases to 0.28 m for e = 250. The gap length is consistent with a linear density of 700 kbpi, which at 57 ktpi will achieve 40 Gb/in2. The analysis showed that when the vertical height of a horizontal sensor h < 2lambdaB, the output is independent of h. In the vertical case, when h > 2lambdaB, the output is independent of W, the sensor width. The latter is considered a significant advantage for vertical heads as track widths are decreased below 1 m. If 2lambdaB > W, the output from a vertical head will exceed the output from a horizontal head.

The output from an experimental device using an NiFeCr/(NiFe/CuNi)n/NiFe/Cr multilayer with W = 2.0 m and a sensor length of 8 m, gave good agreement with the model below 3 mA (25x106 A/cm2) where the output was ~6 mV. Above 3 mA, Joule heating dominated the behavior, which could be improved by heat sinking. The characteristic dependence of output on an applied field was shown for another device with W=0.5 m and sensor length=2 m. It gave a full width-half max of 16.8 Oe with a sensitivity of 215 V/Oe. To produce these devices, Sony used an inductively coupled plasma magnetron sputtering system capable of simultaneous deposition from six 2" targets onto 4" substrates. Vertical stripes down to 0-1 m have been fabricated with electron beam lithography and ion beam etching. Future vertical devices include three GMR layers with both outside layers pinned with adjacent antiferromagnetic layers and an abutted hard magnet for domain stabilization.


There was no discussion of future storage systems due to a shortage of time.

Published: June 1999; WTEC Hyper-Librarian