Metallic Disk Recording Media Survey

Survey Article on Disk Metallic Recording Media

DRAFT of a potential article, never updated

This is an outline and partial draft of an article prepared for
Computer History Museum's
Storage Special Interest Group

Prepared by W. Carlson
Posted in 2012
It has never been updated


Proposed outline and some text of a brief history of metal media development and importance
Transition to Metal Media Outline

Why it’s important
All current media is metal, how did that happen?
Problems to overcome with any new technology
Technology drivers, reasons to adopt
Oxide inefficient, low output
Reduce spacing loss
Early metal surface recordings
Wire recorders
IBM drum
Bryant disk
Conversion Layers
IBM “Steam Homo”
Electroless Plating
Kanigen process
Ampex Alar
Seagate in Fremont,
Domain in Milpitas, large scale “Mint”
Protective Layers
Precious metals
Chemical conversions
Carbon sputtering variations
Tape analogs, fatty acid, etc.
Fluorinated oils (Krytox)
Bonded & non-bonded lubricant
Contamination issues

Current state of the art
Preferred alloys in use today
Carbon overcoat, DLC preferred?
HAMR & patterned media


Why its important
The move to metal film disk media began in the 1960’s but was not commercially successful until the mid 1980’s. The approximately 20 years of engineering work had a number of problems and failures, including some very significant such as Ampex Alar. The driving force behind adoption of metallic thin films is magnetic performance from better efficiency and higher energy, leading to higher areal density and ultimately lower consumer cost due to fewer components. Coated disks, such as RAMAC through 3380 used magnetic iron oxide, γ-Fe2O3 (gamma structure) in a paint-like binder, similar in principal to that used in magnetic tape. Unlike tape, which is in intimate contact with the head, the disk must be recorded to the bottom of the magnetic layer from a distance dictated by the coating thickness plus the flying height of the head, subject to “spacing loss” which is an exponential decrease in energy transfer with increased distance. Over time recording heads were made to fly lower, but a limiting factor remained the thickness of the coating itself, in the range of 50-100 microinches. Additionally, the disk coating was magnetically inefficient. Disk coatings are about 50% by weight of iron oxide, and iron content is about 70% by weight of iron oxide. Due to density difference of 5.24 grams/cm^3 for Fe2O3 vs about 2.0 grams/cm^3 for binder material) the oxide represents <30% of the volume of the magnetic layer. Taking these factors together results in a low magnetic contribution, approximately 20% of the coating’s volume. Gamma iron oxide also has a coercivity limitation of about 350 Oersteds, which limits its resistance to demagnetization. Metal magnetic films, in contrast, are only a few microinches in thickness, are virtually 100% magnetizable, and have adjustable coercivity. These advantages held the promise of a far more magnetically efficient recording layer.

Early magnetic metal films were electroplated nickel-cobalt alloys. The principle drawback of these metal films was lack of corrosion resistance. Gradual growth of corrosion particles could exceed the flying height of a head, resulting in mechanical interference commonly referred to as a “HDI” (head disk interference), leading to a “head crash”. Various techniques were attempted to address the problem, but were mostly unsuccessful. A 1960’s plated drum project at IBM (model # ____) utilized a drum prepared in San Jose’s development machine shop (use Bobby Deardorff commentary), which was then sent to Endicott in New York for magnetic plating. Endicott used a technique called “periodic reverse” plating current to make the film smooth. The process used a modulated sine wave, with more power applied in the deposition part of the cycle, the other half-cycle (reverse direction) removed high spot material due to a high voltage gradient of any peaks, effectively “electro-polishing” while plating. The drum was returned to San Jose, where it was post treated to resist corrosion before installation into a machine. One attempt used a Sodium Dichromate dip described in the literature (Bill Carlson), which slightly oxidizes the surface, leaving a corrosion resistant film … but unfortunately does not have sufficient mechanical durability. The next attempt was use of a hard-chrome over-plating (Ignatious Tsu, Bob Black), commonly used in industry from hydraulic cylinders to bores of rifle barrels. The plating lab in Building-13 was a fully operational prototype shop run by Dr. Ignatius Tsu and Dr. Robert Black, with dirty work performed by technician Jerry Leary. The plated drum project involved flying a large number of curved ceramic heads over a cylindrical surface, which had more than its share of mechanical issues, and the project was cancelled due to reliability problems. Only a handful of commercial electroplated disk makers supplied the drive industry, one of which was Eikon in Southern California (Ron Dennison), providing Maxtor with media for development projects. Eventually this technology was discarded by most as impractical.

One successful implementation was Data Disk, utilizing a single surface for instant replay during sports telecasts. We examined one at IBM in the mid 1960’s, its design included starting and stopping with heads in media contact. The heads were very lightly loaded (a few grams) compared to hundreds of grams for a disk pack design. The sliding/flying head surface included 3 sapphire pads, one of which contained the read-write element. This lightly loaded head was a likely precursor to the Winchester design. Details of the metal film disk are unknown, but most likely electroplated.

A more successful interim process was practiced by several disk makers, including SAE in Milpitas and Ampex in Redwood City. The basic process was a variation on the patented ‘Kanigen” process for nickel plating objects without use of electricity. Typical applications were nickel plated pistols, and hubs in disk packs. The method relies on instability of nickel salts in presence of hot sodium hypophosphite solution, which reduces the nickel ions to a metallic film in presence of a catalyst. A process depending on unstable chemistry is bound to have problems, both in control and waste disposal. The resulting metallic films had good magnetic properties, and began to be adopted by leading edge drive makers for advantages in higher signal output leading to improved areal densities. Ampex in 1979 helped lead the charge into metal media with its electroless “Alar” media, achieving design wins with a number of drive companies including Rodime's R0352 (3.5"), at least one model from IMI, and intended for use in Maxtor's XT1000 (high performance 5.25").By July 1984 Ampex had shipped 1 million disks, but problems started to become apparent with disk failures traced to media problems. (lubricant issues?, ask Gil Argentina for details). In the end adopters of Alar had to recall and/or replace their products using more reliable media. In June 1985 Alar production ceased, and electroless plating of disk media was eventually abandoned.

The surviving method for producing metal media in volume includes RF heating and evaporation of a suitable alloy in a vacuum, forming a plasma which deposits metal with cooling on the desired object. Before use in disks, the technique was (and is) employed to coat reflective or semi-transparent films on large architectural windows to manage sunlight. Other uses are consumer goods including mirrors, and metal films on plastics. Domain Corporation ( see David Pierce) in Milpitas CA was a pioneer in this method for making magnetic media on a large scale, with company founders building their own sputtering and RF equipment based on experience coating window glass. The original idea was to plate a large window size object, then cut the object into disks … not too practical for something with microinch tolerances, so arrays of many pre-finished disk substrates on a carrier were plated instead. The working equipment was called “The Mint”, since they expected to “make a mint” building disks. Domain was acquired by Conner Peripherals in ______, and Conner subsequently acquired by Seagate in _____ with volume manufacturing of sputtered media one of the acquisition goals in both cases. The advantages of sputtering were many; plating could be deposited on any material, operation in a vacuum was inherently clean, and most importantly a protective layer could be sputtered on the magnetic film while still in the machine. A wide variety of alloys with variable magnetic properties can be sputtered, which could not be done with electrolytic or electroless plating. The addition of sputtered carbon overcoat provided both corrosion protection and a slippery surface for heads landing in contact with the media. Sputtering processes evolved to include glass substrate disks, which became mainstream in small form factor drives. Sputtering has become the most universal method of making magnetic disk recording surfaces.

Early coated disks developed at IBM used no lubricants (1311, 2311, 2314), although experiments had shown improved (sometimes temporary) reliability. One experiment involved oleic acid, which provided exceptional wear resistance … but eventually evaporated away. Subsequent lubrication of coated disks included materials used by NASA for space applications due to zero vapor pressure, including highly fluorinated materials (e.g. Dupont Krytox). With advent of metallic media, which is arguably more fragile due to very thin films, lubrication is an important issue. Initial attempts with very inert fluorinated oils were replaced by “bonded” lubes which had chemical attraction to the metallic layer. Attempts at fighting corrosion also involved use of noble metals deposited over recording films. Georgio Bonzano of Olivetti found (as did others) that the fresh Rhodium plated surface became a catalyst for creating “goo” out of airborne organic contaminants. Noble metals such as Palladium, Rhodium, and Platinum are well known chemical catalysts, used today in automobile catalytic converters … but no longer in disks. The most successful solution has been sputtered carbon overcoats; offering lubricity, inertness, non-magnetic properties, and a corrosion barrier. A synthetic lubricant is usually applied over the carbon for additional benefits.

More pages