|Date||Disk Coating - Iron Oxide|
Why It's Important
1950's-1980's Iron Oxide Disk Coating
The RAMAC was the first rotating magnetic memory disk device, and therefore pioneered many of the engineering features of the entire industry utilizing magnetic disks. Fortunately, a lot of recording technology groundwork had been accomplished in Magnetic Tape devices, notably in the German Magnetopon of the 1930’s. Common features between tape devices and disk drives (over time) include:
- Longitudinal recording using a “ring head” strucature
- Particulate iron oxide recording media in a plastic binder
- Durability enhancing abrasive materials (IBM-1316)
- Oxide particle orientation (IBM-3336model2)
Magnetic recording requires a magnetizable surface, originally a metal wire (early wire recorders), which evolved to an oxide coated tape (German magnetophon tape deck), and coated (or "painted") disks. The first ten magnetic disk products, the RAMAC 305 prototypes, utilized magnetic paint, which probably came from 3M, a known supplier of specialty magnetic products. The exact formulation is unknown other than the magnetic particles were likely gamma ferric oxide as used in audio and computer tape.
The basis of magnetic coatings is a magnetizable surface consisting of magnetic particles in a resin binder selected for desirable physical properties (e.g. flexibility, adhesion, mechanical durability, etc.). The iron oxide used for magnetic recording is a non-natural crystalline material originally derived from roasting the mineral magnetite (gamma Fe3O4). A patent often cited for volume manufacture of magnetic iron oxide was filed 1n 1947 by Marvin Camras of Armour Research and issued in 1954 (US Patent # ,694,656)
although the German magnetophon used similar iron oxide coatings on paper and plastic tape in the mid-1930's
The RAMAC demonstrated viability of coated disks, and two IBM patents were clearly intended to optimize coating properties for the new disk form factor. Hagopian's melamine-based magnetic coating patent (2,914,480) filed in June of 1955 may be representative of material used on the first RAMAC production disks shipped a year later. The subsequent Don Johnson et al disk coating patent, first filed in August 1957 disclosed an epoxy-phenolic formulation that became the basis for all subsequent coated disks on aluminum at IBM (plus a few licensees such as Caelus). I became familiar with it during my 1961-1966 tenure at IBM, working with two of the coating inventors Don Johnson and Marcel Vogel during investigations of various disk coating problems (see anecdotes). For a discussion of an urban legend about the RAMAC disk coating see: Golden Gate Bridge Paint Myth.
ROCKS IN THE COATING
Iron oxide was dispersed into the plastic coating ingredients using 210 gallon ball mills. The impact between zirconia containing aluminum oxide balls about 3/4" in diameter sheared and separated magnetic oxide particles, providing a uniform mix of oxide and plastic ingredients after a day or so of milling. Routine chemical analysis verified the ratio of plastic to oxide (about 50%-50% by weight), but there was always a little bit of white residue at the end of the analysis procedure, about 1%. There was not much interest in this residue material, which was thought to be insoluble combustion products of the plastics … or maybe ball mill debris. It was saved by Ed Kettman's building 5 chemistry lab in test tubes, for possible futurre reference. Sometime in 1965 Dr. Leo Diricco was investigating new coatings and durability evaluation methods in the building 13 development lab. One evaluation method was the “tap-tap” tester, impacting a stainless steel button (about ½” diameter) into the spinning disk repetitively until the coating failed and electgrical contact was made between the button and disk substrate. The development coatings always failed prematurely compared to material obtained from the manufacturing operation. One difference I observed was the absence of white debris in chemical analysis of the development coatings. Suspecting the white residue might have something to do with it, I collected a quantity of residue from the manufacturing QC lab and added it to our development coatings ... and voila! durability was now equivalent. This and subsequent work eventually led to a patent for the use of abrasive particles in disk coating, filed on 18 september 1972 by Joe Haefele and Ron Kubec, issued as patent number 3,843,404. This turned out to be "old news" since the tape industry had been adding ceramics for years to their formulations for the same reason, to improve wear resistance.
August seemed to be the month for coating problems at the San Jose manufacturing facility in the 1960's. Two years in a row the manufacturing process went "out of control", shutting down disk pack production ... or at least severely curtailing it. The problems included "coating runs" (little rivulets of material radially streaking the disk), and "fish eyes" (circular spots with lack of coating coverage). Since disk packs were highly profitable and required to support disk drives, the problem got a lot of corporate attention. John DelFavero was in charge of the problem solving effort, calling daily meetings to go over data, plan experiments, and make assignments. It was politically wise to pull out all the stops so one could not be accused ot insufficient effort, so the inventors of the coating were called in as consultants (although they had no recent experience), plus any development and manufacturing personnel who could contribute. One potential culprit was usually "improvements" made to the process by manufacturing engineers to simplify the operation by omitting some steps in the original implementation ... which changes of course were reversed in hopes of a fast fix. The next suspect included materials being used, but of course IBM was a pioneer in disk making ... so the vendors had limited insight into what was going on at our end. Suspicion turned at one point to a flow control agent supplied by Dow Corning, since infrared analysis showed a slight lot to lot variation. John DelFavero called Dow Corning one morning and insisted on talking to the "head man" responsible for Paint Additive #11, a flow control agent. When Dow's manager came on the line, he was advised of the seriousness of the problem. Upon finding out that we were spin-coating magnetic paint at a few hundred microinches in thickness, Dow's scientist fell silent. When John asked what his quality control methods involved, he waited a bit, then advised they mixed a small batch of conventional paint, applied it to an inclined plane, and made sure that it "looked good" prior to releasing that lot. Now there was silence at our end, clearly we were using the material well beyond the maker's intended purpose. The lesson learned is that a new industry is highly subject to consistency of their suppliers, and unannounced (or even unintended) changes are usually bad news. Something similar happened with the adhesive used in 1302 drives (got gooey & ran over the disk), and years later to Read-Rite when a vendor changed formulations on photo resist which put them out of business for a while.
Too Much Oxide
In the late 1960's a number of new vendors of disk media put a strain on the oxide supply. The principal vendor was C.K.Williams & Co., who announced rationing of oxide to distribute what was available to meet an overwhelming demand. Since the ration was a percentage of the oxide on order, customers were motivated to increase their orders such that the ration would be close to what they wanted in the first place. Many customers greatly exaggerated their needs in hopes of building a "nest egg" of this critical material. Several months later a semi-truck and trailer pulled into the Caelus parking lot with hundreds of bags of iron oxide. We were told that the vendor responded to the huge demand by renovating and enlarging his facility, and was now able to fill all the orders. This meant that a 5-10 year supply was now sitting on the back dock, and the company had to build a shed in the parking lot to hold it all. I often wondered what the vendor thought when the record quantitiy of orders dropped to near-zero for those who had overindulged. The lesson learned is to keep in touch with your vendor!
Tape coating is applied to flexible Mylar material using rollers in an automated process which may contain a mile or more of coated plastic film at one time. At one point in the process, shortly after the wet coating is applied, the tape is passed over a magnet array which aligns the acicular magnetic particles into the same direction before the coating hardens. This alignment maximizes output by eliminating random particle orientations which would otherwise leave some particles perpendicular to the field of the recording head, and therefore not contribute to the magnetic recording process. Disk coating was applied using a "spin coating" method wherein the coating material is poured from a container (or dispensed from an automated spigot) onto the inside edge of the disk, letting centrifugal force from the spinning disk spread the material out over the rest of the surface. An alternative method used by Memorex applied the coating via a spray-paint technique, but neither method facilitated particle orientation. In late 1965, after reviewing tape orientation benefits, I thought it would be interesting to try particle orientation on a disk, so we used a permanent magnet's circular field above and below the disk to orient particles in all directions from a central spot. After curing and polishing the disks, Engineers evaluated the different orientation directions, concluding that longitudinal orientation looked best. The challenge was then to circumferentially orient the entire disk surface coating, which had to be done while the disk was both wet and spinning. The solution was a modified magnetic disk brake electro-magnet from IBM's drive production line. The original brake acted by placing a metal disk on a drive shaft in a slot in the electromagnet, which slowed the shaft when current was applied to the electromagnet. By placing such a disk brake above and below the wet disk during coating application, a uniform horizontal field was obtained in the circumferential direction ... just what we needed. Disks so prepared were evaluated and found superior (they also had a nice irridescent appearance), so this technology was applied to the 3336-II generation of media ... but probably never patented.
Other Magnetic Oxides
Tape makers (e.g. 3M, BASF, Memorex) tried variations on red ferric oxide, including black magnetite, Chromium Dioxide, and cobalt substituted iron oxides. Chromium Dioxide became a secondary standard for cassette tapes, but required a higher intensity magnetic field for optimum performance, requiring special settings on tape recorders. The "HD" flexible disk (1.2 MB 5.25 inch) used a cobalt doped iron oxide to improve output at higher track density, which made them incompatible with the earlier 360KB disks using conventional magnetic iron oxide. Mainstram disk drives never used other than conventional magnetic gamma iron oxide.
Comment - this is overwhelming as told an IBM RAMAC story - UNIVAC made coated drums (maybe before RAMAC), certainly plated drums predate RAMAC, (even at IBM??) Shouldn't there be a topic that covers Ferranti, UNIVAC, Vermont Research, LFE, Burroughs? How about card equipment that had drums in it?? Grant
Provenance note: This article was originally written by Bill Carlson with contributions from Grant Saviers and Tom Gardner.
Revision 8 of the article was reviewed at the June 20, 2012, meeting of the Computer History Museum's Storage SIG and it was decided that the article was not suitable for inclusion on the main timeline but should be evolved into an oxide media survey article with links to and from oxide media events on the main timeline.