 |
|
||
|
Revolutionary inventions in previous decades have enabled the manufacturing of inexpensive magnetic disk drives, low-cost devices with properties that even computing centers did not have in the 1980s. To perpetuate the gains in storage capacity, the magnetic storage industry is in pursuit of “planned revolutions.” Some of these revolutionary technologies are bit-patterned magnetic media, schemes of energy-assisted (heat, microwave) magnetic recording and some others. However, these are all difficult and expensive in practice. Recently, the pace of growth in areal density of recording (hence, capacity per disk drive) has shown signs of dramatic slowing. This is characteristic of the transition to a fully mature technology, often described in terms of the “S-Curve” of product maturity [1]. New types of non-rotating data storage technologies may become able to replace an increasing part of the market now occupied by standard disk drives. This article will take a look at some of the issues in extending existing magnetic recording and the potential impact of new storage technologies. When IBM invented the hard disk drive in 1956 it was a revolutionary new product. Compared to the existing punched card systems or tape drives for data storage, the new rotating platter storage enabled faster storage and recovery and virtually instant access to any individual file. The RAMAC (Random Access Method of Accounting and Control) stored a whopping 5 Mbytes in a system the size of a large refrigerator, but heavier. The areal density of storage was about 2,000 bits per square inch (100 bits per inch and 20 tracks per inch) [2]. Access time to any file was about half a second (600 msec). The cost in 2009 dollars would be more than $100,000. Fifty-three years later, disk drives descended from that first product are incredible masterpieces of science and engineering. A drive for a laptop isn’t much bigger than a deck of playing cards. You can buy one for $100 that’s the size of a hard back book and holds 1 Terabyte of data (that’s about 200,000 times the capacity of the first RAMAC) and transfers data over USB at up to 60 MB/sec. The degree of improvement is obscured by the change from refrigerator-sized systems to book-sized or smaller. Data storage areal density was originally 2,000 bits/in2; products are now shipping at 330 Gbit/in2 (165 million times more data in each square inch) and all the other attributes are similarly improved. Laboratory demonstrations have shown technical feasibility of as much as 800 Gb/in2. During the last 20 years, the products introduced each year have shown a growth rate in areal density of approximately 40 percent, except for a four year period from 1998 to 2002 where growth rate was 100 percent per year [3]. At this rate, areal density will double in just two years. With the industry having started shipping products at 330 Gb/in2 just this summer, it will need to ship products at 1 Tb/in2 in late 2012, just over three years from now. However, there are already indications that the rate of areal density growth in shipping products is slowing down. Since 2006 to 2007 it appears that shipping products have fallen below 40 percent/year (190Gb/in2 in early 2006, maybe 280 Gb/in2 early 2008, but about 330Gb/in2 in mid-2009 suggests an average growth rate below 40 percent for the past two years). In order to hit an average 40 percent rate from 2006 at 190 to 200 Gb/in2 up to 1 Tb/in2, the industry would have to ship that density of products by late 2010 or early 2011. Already, one can see that things have slowed down if the more recent projection suggests 2012 instead of 2011. Frankly, it seems unlikely to happen, requiring a more than doubling of areal density in the next year and a half. This further supports the observation that the hard drive industry is at the top of the S-Curve, with rate of areal density improvement slowing down. But we mustn’t count this industry out just yet.
To explain why, let’s briefly review the problems of competing effects. The magnetic coating on the hard disks is a metal alloy that is micro-structurally composed of small crystalline grains (See Figure 1). The basic problem is that increasing areal density means making the magnetically stored bits smaller. Making the bits smaller though, means also making the individual grains smaller. This is because the signal to noise ratio is improved by having more grains in the recorded bit. However, smaller grains become thermally unstable and prone to spontaneously lose their magnetized state, or erase the data. This is a basic effect of physics called superparamagnetism [4, 5]: the thermal energy in the magnetized material causes the magnetic spins of the atoms to jitter and reverse their direction. To overcome this effect, there are two approaches. One can make the grains bigger, until each bit of data is made up of just one grain. And one can make the coercivity of the recording medium larger, so the magnetic spins are less likely to be reversed by heat. The first approach has led researchers to “bit patterned media” and the second approach to “energy assisted writing.” In Bit Patterned Media, (See Figure 2) the recording media is patterned into a regular array of separate pillars and the bits are recorded in the pillars. To achieve this at, for instance 1-10Tb/in2 requires pillars of diameters 12 nm down to 4 nm, spaced about an equal distance apart to prevent interactions. Right now, and for the next 10 years, this far exceeds the ITRS Semiconductor Roadmap. Still, it’s believed that it may be achievable by the use of “nano-imprint lithography.” This is a technique in which a master pattern mold is made using slow processes, and daughter molds are used to press the desired pattern into millions of disks. It remains to be seen whether this can be done economically; there’s a fierce debate on that subject [6] because the new processing equipment is very expensive. Energy Assisted Writing is the other technology being developed. With the current technology the writers can’t produce enough magnetic field to write effectively when the media’s coercivity is increased enough to maintain thermal stability. However, by adding some energy to the recording medium to assist the magnetic field, effective writing can be achieved. There are two approaches to adding this energy. In one, called Heat Assisted Magnetic Recording (HAMR) (See Figure 3), a structure is added to the write head that focus the light from a laser into a very, very small spot of the disk. This light heats the medium, thus lowering its coercivity and making it easier to write, then as the head moves away the recorded bit cools down and the coercivity returns to normal, keeping the bit stable. In the second approach, called Microwave Assisted Magnetic Recording (MAMR) (See Figure 4), a different structure is built into the writer that generates a high frequency microwave signal in the disk. This signal destabilizes the local spins, making it easier for the applied magnetic field to write data. Again, as the head moves away from the recorded bit the microwave signal goes with it, so the recorded bit becomes stable. HAMR heads have actually been made and shown to work in principle. However, it adds expense to the hard drive for the laser and, possibly worse, raises the power requirement for the operating drive. The HAMR writer also requires a complex re-design of the materials and structure of the recording disk itself. It remains to be seen whether it’s a viable and economically feasible approach, but this is a main line of research for most or all of the hard drive companies. So the question is: will these technologies breathe new life into the hard drive industry’s push for areal density or will they be insufficient to prevent significant slow-down of the rate of improvement? New technologies have happened before and rejuvenated the industry. Examples of this would include thin film media, MR heads, GMR and TMR heads, and others. However, if magnetic data storage is actually within two or three orders of magnitude of the physical limit, it’s to be expected that progress will slow. This produces an opportunity for new modes of data storage that may be able to surpass rotating magnetic recording. Alternate Storage Technologies Other candidates include several that make use of the third dimension. That is, they achieve storage in a volume rather than on a surface. Holography is an early version of this, though it’s volume density is limited in practice to be uncompetitive with hard drives even in volume capacity. There are other candidates, too. These range from multiple layers of phase change recording cells to a kind of racetrack magnetic memory [10]. Others include polymer and molecular or DNA-based memories [11]. Some of these are obviously unlikely, but others show some interesting promise. At present, hard drive technology appears to be reaching a plateau that will slow down additional increases in areal density, the classic “top of the S-Curve.” Will these new technologies ignite a new period of rapid innovation and improvement or will the slow-down continue? This would open up an opportunity for a new storage technology to displace rotating magnetic storage. There are numerous candidates, all with pros and cons, and it will be interesting to see what happens over the next decade. REFERENCES Ed Murdock worked for Seagate Technology from 1992 to 2009 as an executive director of Engineering. Prior to Seagate, he worked at Hewlett-Packard Labs. He has a Ph.D. in Physics from the University of California at Davis. He is presently the principal of MurdockTech Solutions, consulting in areas of high tech product development, patent infringement litigation, management of technology and integration, and other areas. He can be reached at edmurdock1@comcast.net. Click here for reuse options!
|
||
|
![[Reuse options]](http://license.icopyright.net/images/icopy-w.gif){kind=small}
Revolutionary Developments In-Progress© 2011 Webcom Communications Corp.