STORAGE TECHNOLOGIES

							                                                                                   1


               STORAGE TECHNOLOGIES

Introduction:
               The rate of advancement in storage technology has been truly
amazing. How ever we are fast approaching the physical limit for storing
information on media such as magnetic platters of hard disk or the chemical
layers in optical layers such as CD’s and DVD’s. with the promise of
tomorrows operating systems in incorporating stunning graphical interfaces
that offer truly immersive virtual reality and next generation games that will
blur the line between fiction and reality, the demands of being able to
quickly store and retrieve enormous quantities of data are ever increasing.

              Research in these fields has created storage solutions such as
holographic storage, blue laser and Millipede offering digital storage
densities that are orders of magnitude.

              Even though devices that use these technologies are still on the
outskirts of implementation, they are slowly and surely advancing down the
road to commercial availability.

OAW Technology:
              In OAW technology, instead of using a regular head to read
and write data off a hard disk platter, a laser is used. This makes for extreme
precision. The polarization of a light beam is twisted. When it bounces off a
magnetized surface. When this is passed through a polarizing filter, the
intensity of the beam gives the magnetic alignment of the part at the surface
that it touched. This is how the disk is read. For writing, the same laser heats
a dot on the disk to a temperature at which the material’s magnetic
properties can be changed to effect either a one or a zero using a magnetic
coil.

             Essentially, since a laser is so precise, it can focus on areas
much smaller than are used to represent a bit on a traditional hard disk. This
makes for higher data densities.
                                                                                 2


AFC media:
              IBM demonstrated yet another way to get round the Super
Paramagnetic Effect (SPE): a three atom thick layer of ruthenium
sandwiched between two magnetic layers. Known as Anti-
Ferromagnetically-Coupled (AFC) media, the multi-layer coating permits
disks to store 100 gigabits per square inch, as of now. The ruthenium layer
makes the adjacent layers orient themselves magnetically in opposite
directions. When bits are written onto layers, they can be placed closer
together than if a single layer were used-meaning higher areal density.

Patterned Media Recording:
              One bit on a disk is stored as a group of several small grains.
Each of these grains is thermally unstable, and the Super Paramagnetic
Effect (SPE) comes into play. In patterned-media recording, one bit is stored
by one grain; a grain can, therefore, afford to be larger. Thus data integrity
can be maintained at very high densities, ranging from 100 gigabits per
square inch to tens of terabits per square inch. Patterned media refers to
ultra-high density magnetic storage media consisting of arrays of discrete,
lithographically patterned magnetic elements, each of which can store one
bit of data. Each discrete element is isolated from other elements, but inside
each one polycrystalline grains are strongly coupled, behaving more like a
larger, single magnetic grain. Because the Super Paramagnetic Effect (SPE)
limit applies to the entire bit, not to each of the grains, the single bit in
patterned media can be much larger than the one in conventional media. The
problem with patterned media is that there are no cheap mass-manufacturing
procedures.


Probe Storage: IBM’s Millipede:
              Punch cards were the first device that could be used to store
digital information. It was based on a very simple principle of indentations
on cards that were read by a machine and translated into 0s and 1s. IBM has
recently developed a very similar system but has scaled it down to extremely
small dimensions, almost bordering on the realm of atoms! Known as
Millipede, this technology is theoretically capable of storing information
equivalent to about 25 million books in an area the size of a postage stamp!

              One of the obvious advantages with this technology is the fact
that very large storage densities can be achieved in very small areas. Lower
power consumption makes it ideal for mobile applications such as handheld
                                                                                  3


computers and cellular phones. The next generation cellular phone would be
able to hold a gigabyte of multimedia content and contact information! The
main hurdle that lies in the path of commercialization of this technology is
the fabrication of the controllers that go into these chips.

The operational concept behind Millipede:
              The Millipede chip is created using an array of very tiny
cantilevers to create almost atomic-sized indentations in a plastic substrate
that is used as the recording material. This array works in a massively
parallel fashion where a bank of cantilevers access or create information.

              Before a read or write operation, the polymer-based medium,
which is just about 50 nanometers thick, is positioned beneath the cantilever
array. This medium is mounted on a magnetically driven scanner that can
move in three dimensions. During read-write, the medium is moved by he
cantilevers along the X-Y axis while the cantilevers actuate and create
indentations on the recording surface. Using this process and with a single
cantilever design, researchers have managed to achieve a storage density of
an astounding 60 to 80 GB per square centimeter. Also, this substrate can be
‘erased’ and data can be re-written onto it repeatedly. This is achieved by
momentarily heating the polymer to a temperature of 150 0C so that the
surface is effectively smoothened and ready for rewrite. However,
individual bits of information cannot be erased; only larger sections of the
polymer surface can be cleared.


              The image above shows an actual electron microscope image
of one of the Millipede cantilevers. The tip of the cantilever head is about 50
angstroms wide-that’s just a few atoms clustered together!

             The tip at the end of a cantilever makes for the ultimate in local
confinement of interaction. This makes tip-based storage technologies
natural candidates for extending the physical limits being approached by
conventional magnetic storage. Probe storage technology makes
indentations 10 nm (nano meters) in diameter on a plastic film. Each
indentation is a bit, which can be heated back to its original shape.
Thousands of tiny probes-tips at the end of a cantilever move back and forth
ever the film. Because of the many probes IBM’s codename for this
technology is ‘Millipede’. There are problems inherent in this idea, and
IBM’s solution was to use micro-electro-mechanical systems (MEMS)
based arrays of cantilevers operating in parallel, with each cantilever
performing write, read and erase operations in an individual storage field.
                                                                                    4


Probe storage is expected to use just about as much power as a flash card,
but store up to 100 times as many bits as traditional disks.

Cantilever:
               Cantilever is a horizontal fixture, such as a mechanical arm,
that projects beyond a vertical support, and is unsupported at one end.

Holographic Data Storage (HDS):
               Holographic data storage is a breakthrough technology that
literally goes beneath the surface of the media and stores information within
its volume. Current generation storage media store digital information on
layers of platters either magnetically or optically. Holographic storage
utilizes the space within special types of crystals with optical characteristics,
which are very similar to those used to create 3D holograms. Even though
holographic storage has been around for over a decade now, giants in the
field of technology such as IBM, Lucent and Rockwell are just beginning to
make serious breakthroughs with this technology. Holographic storage
incorporates the use of lasers, optics, image sensors and special crystals that
respond to laser light.

              Holography allows a million bits of data to be written and read
in parallel with a single flash of light. This enables transfer rates
significantly higher than current optical storage devices.

              Combining high storage densities, fast transfer rates, with
durable, reliable, low cost media, make holography poised to become a
compelling choice for next-generation storage and content distribution
needs.

              Holographic data storage (HDS), which makes use of full
volume of the recording medium, possesses high potential by promising fast
transfer rates of hundreds of Megabytes/sec and storage densities greater
than 200 Gbytes per 120mm disk. The restrictions that are placed on the
holographic media, however, are stringent. Described hear is a high
performance photopolymer based medium that has the properties necessary
to enable this technology. Through the use of several different holographic
techniques, the material characteristics that are necessary for holographic
storage products may be determined. The two different systems that are
discussed here include plane Wave and Digital Holographic Data Storage.
These measured characteristics include high dynamic range, sensitivity, and
small recording-induced Bragg detuning. In addition, results of archival and
shelf-life environmental testing of the media will be discussed.
                                                                                 5


              Besides the astounding amount of data that can be stored using
this technology, there are other inherent advantages that holographic storage
offers. The main advantage is the speed of retrieving the data. Such systems
will be able to retrieve data in tens of microseconds as compared to a data
access time of almost 10 milliseconds offered by the fastest hard disks
today. By the time they are commercially available, scientists say
holographic systems could transfer an entire DVD movie in under 30
seconds! Another very important advantage is that of information search and
retrieval. Consider the case of large databases that are stored on hard disks
today. To retrieve a piece of information, you first provide some reference
date (a keyword), the data is then searched for by its address, track, sector
and so on, after which it is compared with the reference data. In holographic
storage, entire pages can be retrieved, where the contents of two or more
pages can be compared optically without having to retrieve the information
contained in them. Imagine what this would imply in the field of data
mining and warehousing, massive sets of data values can be compared in
fractions of a second!

              While this technology points to an almost utopian era for
storage, the success of holographic storage lies in the ability to accurately
focus the reference laser on the exact position within the crystal to retrieve
that page of information. You would be unable to locate the data if there’s
an error of even a thousandth of an inch. Also, the crystals used in the
fabrication of the storage element need to exhibit very exacting optical
characteristics to store the data correctly and there are very few substances
that adequately and economically meet these needs. Holographic storage
should be available in the next five years and they would debut with
capacities of about 125 GB with a data transfer rate of 40 MBps.

              The flexibility of the technology allows for the development of
a wide variety of holographic storage products that range from handheld
devices for consumers to storage products for the enterprise. Imagine 2GB
of data on a postage stamp, 20 GB on a credit card, or 200 GB on a disk.

How is Data Recorded?
             Light from a single laser beam is split into two beams, the
signal beam (which carries the data) and the reference beam. The hologram
is formed where these two beams intersect in the recording medium.

              The process for encoding data onto the signal beam is
accomplished by a device called a spatial light modulator (SLM). The SLM
translates the electronic data of 0’s and 1’s into an optical "checkerboard"
pattern of light and dark pixels. The data is arranged in an array or page of
                                                                                 6


around a million bits. The exact number of bits is determined by the pixel
count of the SLM.

             At the point of intersection of the reference beam and the data
carrying signal beam, the hologram is recorded in the light sensitive storage
medium. A chemical reaction occurs in the medium when the bright
elements of the signal beam intersect the reference beam, causing the
hologram stored. By varying the reference beam angle, wavelength, or
media position many different holograms can be recorded in the same
volume of material.




How is Data Read?
              The interference pattern induces modulations in the refractive
index of the recording material yielding diffractive volume gratings. The
reference beam is used during readout to diffract off of the recorded
gratings, reconstructing the stored array of bits. The reconstructed array is
projected onto a pixelated detector that reads the data in parallel. This
parallel readout of data provides holography with its fast data transfer rates
(10's to 100's of MBytes/second).

              The readout of data depends sensitively upon the
characteristics of the reference beam. By varying the reference beam, for
example by changing its angle of incidence or wavelength, many different
data pages can be recorded in the same volume of material and read out by
                                                                                 7


applying a reference beam identical to that used during writing. This process
of multiplexing data yields the enormous storage capacity of holography.




             In the past, the realization of holographic data storage has been
frustrated by the lack of availability of suitable system components, the
complexity of holographic multiplexing strategies, and perhaps most
importantly, the absence of recording materials that satisfied the stringent
requirements of holographic data storage.
               Recently the development of practical components for
holographic systems has rekindled interest in this technology. While the
development of the needed components has been accomplished for non
holographic markets, the volume of these markets is expected to lead to low-
cost, reliable components for holographic data storage. DVD-R (red 680nm)
and DVD-B (blue 405-407nm) have been developed for the optical storage
market place. These recording sources have the desired characteristics for
holographic storage and are attractive due to their small size, ruggedness,
and low cost. Digital micromirror devices appearing in new types of
displays are ideal spatial light modulators with their large numbers of pixels
(~ 1 million), fast frame rates (2000 Hz) and high optical contrast. The
                                                                                 8


CMOS active pixel detector arrays emerging in digital photography exhibit
the rapid access and data transfer properties required for holography.

             Recently several multiplexing techniques that yielded a simple,
easily implementable architecture for holographic storage systems. Spurred
by this development, focused on the long-standing problem of the lack of
suitable storage materials and invented new high-performance recording
media with demonstrated high density data storage capabilities. Our work
serves as the foundation for a practically realizable, high capacity storage
system with fast transfer rates and low-cost, removable recording media.

Example drive for Holographic Data Storage




Blue Laser:
               Lasers have long been used in the optical storage of digital
data. We see them in everything from CD-ROM to DVD and magneto-
optical drives. Information is stored in these devices by encoding data into a
stream of 0s and 1s in a spiral on the surface of the media. The amount of
digital information that can be stored on a single layer of optical media
depends on two factors: the size of the encoded bits on the media surface
and how close these bits are packed onto the surface. The density of these
bits is directly dictated by the wavelength of the laser beam that reads them.
The shorter the wavelength of the light, the narrower is the laser beam, and
                                                                                    9


consequently the smaller can be the size of the bits of information that are
encoded on the disk.

              As of now, the lasers used in CD and DVD-ROM drives have a
wavelength between 630 nm and 650 nm (1 nm = 1 millionth of a
millimeter), which puts them in the red band of light. The reason why blue
lasers are so exciting is because their wavelength lies in the 400 to 450 nm
range. This shorter wavelength light translates into disks written with blue
laser being capable of storing between 25 - 30 GB of data per layer per side!
That gives it 5 to 6 times the storage density of existing DVD’s. Imaging
having all the star wars movies with full blown surround audio and crystal
clear video quality on a single disk with room to spare!

               If it is a simple a question of using laser light with a smaller
wavelength, you might ask why this was not done before. The reason for
this is that the materials used to generate blue lasers have a relatively shorter
lifespan compared to those used for red lasers. While blue lasers are in the
research phase, there are three types of methods they are used to generate
these lasers.

Zinc Selenide (ZnSe):
              The initial method for implementing blue lasers involved the
use of Zinc Selenide to fabricate the diodes that generate blue lasers.
However, this material has a relatively short lifespan and its power
requirements make it economically unsuitable for commercial
implementation. Also, these lasers have wavelengths ranging from 460 to
520 nm, putting them at the end of the blue and closer to the green light
band of the spectrum.

Gallium Nitride (GaN):

                  This material has proved to be very successful in the
creation of blue lasers and has generated wavelengths as low as 370 nm with
relatively high reliability. Most of the work in blue lasers today is based on
this material.

Second Harmonic Generation Lasers:
                   These lasers are relatively new on the blue laser scene
but have exhibited very high levels of reliability. Through this intelligent
method, the frequency of a given laser is doubled (that is, the wave length is
halved) and laser light within the blue spectrum is generated. This is done
through an apparatus called a Distributed Bragg Reflector (DBR) where, for
                                                                                      10


example, the frequency of an infrared laser with a wavelength of 850 nm is
doubled, resulting in a blue laser with a wavelength of 850 nm is doubled,
resulting in a blue laser with a wavelength of 425 nm.

              It will be some time before blue laser technology becomes
commercially viable. The major hindrance to this technology is the cost of
implementation. A Blue ray device available today would cost about Rs 2
lakh! The second hurdle is that of reliability: red lasers that are used in all
CD-ROM and DVD-ROM drives today have a life cycle of about 10,000
hours. Now compare that to the meager hundreds of hours that the Gallium
Nitride based blue lasers last for using today’s technology! However, it is
just a matter of time before these issues are addressed. Scientists predict that
in just a couple of years nearly all new optical storage devices would be
based on blue lasers.

Flash Memory:
              Companies need to keep shrinking Flash memory as portable
devices get smaller cell phones and digital cameras, for example. Flash
sticks transfer data fast and, like hard disks, don’t need to be powered on to
retain data. These sticks don’t have moving parts, so they’re much more
rugged.
              It’s becoming increasingly difficult for flash memory to shrink,
and therefore, to obey the exponential law. The reason hinges upon the way
Flash memory works: memory cells in Flash sticks are transistors whose
gates are wrapped in a layer of silicon dioxide that prevents electrons from
escaping. The cell holds a one or a zero, depending on the charge in the
transistor. The problem is that the silicon dioxide insulating layers are about
90 angstroms thick; and at 80 angstroms, the layer will no longer be as
effective.

              Although companies such as Intel agree that the issue is
currently manageable, making Flash chips with a component size of 45 nm
will be difficult. And the 45-nm manufacturing process is set to begin in
2007. Companies, therefore, are looking at alternatives-nanocrystals,
Ferroelectric RAM (FeRAM), Magnetic Random Access Memory
(MRAM), Ovonics, and polymers.

                The use of nanocrystals promises to halve chip sizes by using a
lattice of silicon crystals as the insulator, instead of the silicon dioxide layer.
At the nanoscale, silicon behaves as an insulator. Silicon-Oxide-Nitride-
Oxide-Silicon (SONOS) is another similar alternative that’s being
researched.
                                                                                11


              FeRAM is non-volatile, like Flash. However, it’s much faster,
and consumes less power. The disadvantages include lower density and
higher cost. FeRAM stores data by using an electric field to shift the
position of individual atoms in crystals.

               Ovonics is very different from traditional electron-based
storage; it includes a charge by a chemical process. After rapidly heating an
alloy substrate similar to the material used in CD-RWs, dots representing
ones and zeroes are created in the alloy: a one is encoded by creating a
crystalline dot, and zero is encoded by creating an amorphous dot. The bits
are retrieved by measuring the change in electrical resistance between the
alloy’s crystalline and amorphous states. Ovonics promises to be much
faster than Flash, and can be re-written many more times. The problem,
currently, is that its fabrication process is expensive.

              Intel and other companies are developing polymeric
ferroelectric RAM (PFRAM), also known as plastic RAM. The technology
involves sandwiching a thin polymer sheet between two perpendicular
layers of metal strands. A memory cell is formed at the intersection of these
strands, and data is stored by changing the polarization of the polymer
between the strands. PFRAM is a lower power technology. The
manufacturing process is simple. And there are cost advantages too:
PFRAM requires only a single layer of CMOS circuitry, and can stack up
eight layer of inexpensive polymer material. The drawback is speed. Read
and write speeds don’t compare with that of today’s Flash chips. Hence,
application may be limited to memory sticks.


Non Volatile Devices (RAM):
DDR2:
        DDR is the current memory standard for most computing systems;
and DDR2 DIMMs have already been manufactured. DDR2 starts where
DDR left off; it’s faster, and consumes less power the first DDR2 modules
to hit the market were faster than the fastest DDR modules available now.
DDR2 is not without its drawbacks, tough, at least initially: it’s not
backward compatible with DDR, and motherboards need to be specifically
designed to support the standard. This is not a small problem; not many
motherboards manufacturers have shown interest in the new standard,
because of the relatively large initial price difference between DDR and
DDR2. However, as an indicator of the importance of the emergence of the
standard, AMD will build support for DDR2 into the Opteron’s on-board
memory controller, later this year.
                                                                              12


GDDR3:
              GDDR3 is RAM optimized for graphics processing. It is the
fastest memory available today. Speed and power are the important factors
when it comes to graphics designs; GDDR3, therefore, delivers four data
bits every two cycles, and consumes half the power, of graphics DDR2. The
first products you can expect to see using GDDR3 chips will be high-end
graphics cards.

RLDRAM:
              Reduced Latency DRAM (RLDRAM) features a new design
that minimizes the time between the beginning of the access cycle and the
first appearance of the data. RLDRAM is thus all about performance. It
combines fast access due to low latency with high density and high
bandwidth. The technology optimizes performance for cache and
networking applications. Although lower latency would certainly mean
better performance for personal computers, RLDRAM is being targeted
mostly at the networking equipment segment, such as routers and switches.

FCRAM:
               Fast Cycle RAM (FCRAM) was developed by Fujitsu in 1999.
It’s radical in its approach; it changes the core of the DRAM architecture,
whereas other technologies have obtained speed improvements by
enhancing the components that accessed the memory core. Changes to the
DRAM core include core segmentation and pipelined operation. This is the
first time the core of the DRAM technology has been changed, in more than
twenty years. And the changes make for amazing possibilities: row and
columns information can be sent at the same time in FCRAM, and because
of pipelined operation, a secondary command can be issued without waiting
for the primary command to be fully executed. The interface to FCRAM can
be both SRAM like, which will be used in mobile phone applications, and
Single Data Rate (SDR) or Double Data Rate (DDR2), which can be used in
PC systems.

IRAM:
             One approach to reduce the problem of processor-memory
communication delay is Intelligent RAM (IRAM) - Integrating processor
and memory on a single die. Of course, for the solution to be complete,
memory latency needs to be reduced, too. Currently IRAM products are not
commercially available; however, several university and company research
labs are working on IRAM architectures.
                                                                                  13


MRAM:
             This type of RAM was born out of the same cast as the hard
disk drive. Using a phenomenon called Giant Magneto resistance (GMR),
which was first developed by IBM, very small magnetic fields can be stored
and sensed reliably and quickly.

              MRAM uses the same principle where similar types of GMR
materials are used to fabricate arrays of microscopically small magnetic
elements. The orientation of the magnetic field in these elements (North or
South) determines whether it is carrying a digital zero or one. Each GMR
cell is a sandwich made of special metallic materials (like iron and
chromium). In an MRAM cell, there are parallel rows of microscopic wires
above and below the magnetic sandwiches. The ones above run parallel to
the sandwich while the ones below run perpendicular to it, looking much
like the weft and wrap of a fabric room. In its normal state, each magnetic
element will have an inherent charge in it. If a current is passed through the
neighboring wires, this magnetic orientation can be changed and thus the
data can be modified and also written into the element.

              While reading data from these elements, a current is passed
through the sandwich and the magnetic resistance is measured a low
resistance signifies a zero while high resistance signifies a one. This process
of reading does not disturb the magnetic condition of the cell.

              The inherent advantage with MRAM is that since it is based on
the principle of magnetism, it retains whatever information is contained in it
even after power is switched off. Additionally, unlike conventional DRAM,
these elements do not need an electric charge to retain the information. It is
only needed during read or write processes. Hence, power consumption of
these devices is fractional compared to that of DRAM, which needs to be
continually refreshed in order to retained data. Finally, with the absence of
the delays associated with transferring electricity between the storage
elements for retaining information for MRAM is expected to be 30 times
faster than DRAM. With these characteristics, MRAM is poised to given
rise to a new breed of instant-on computing devices.

              Magnetic Random Access Memory (MRAM) is non-volatile,
like Flash. It can combine the high speeds of SRAM, the capacities of
DRAM, and the non-volatile nature of Flash. In an MRAM chip, the
memory cells are the intersections of the rows and columns. Electric signals
polarize the cells into ones and zeroes. MRAM is like a Flash stick-there is
no moving parts; besides, power consumption is lower. It’s therefore likely
that MRAM and Flash will compete in the portable-device arena. MRAM
                                                                             14


has the edge because it will prolong battery life. As mainstream storage,
MRAM chips are unlikely to hit the market very soon. IBM, however, is
working on a combination of MRAM and holographic memory.

Applications:
                    Blue Laser Technology used in Enhanced DVD (Digital
Versatile Disk) applications. The data transfer rate is medium.

                  Holographic Data Storage is used for Data Mining and
Data Warehousing. The data transfer rate is very high.

                    Millipede is used to store for mobile and portable
devices. The date transfer rate is high.

Conclusion:
      With so many materials and techniques being explored, one can’t
assume that storage will always be magnetic disks, or that memory will
always be dynamic RAM. Systems of the future may build upon the
concepts that exist today, or they may employee entirely new technologies
such as holographic memory.

      But we can be sure of few things – for example perpendicular
recording will happen very soon, MRAM will be researched more
aggressively, because of its promise as a replacement for Flash as well as
becoming a PC RAM solution etc.,


Bibliography:
Digit Magazine (March 2004)
                  - Jasubhai Digital Media

Digit Magazine (Aug 2002)
                  - Jasubhai Digital Media

www.inphase-technologies.com
                     Contents



Introduction

Patterned media recording

Millipede

Blue Laser

Holographic Data Storage

Flash Memory

Non Volatile Device Technologies

               DDR2
               GDDR3
               RLDRAM
               FCRAM
               IRAM
               MRAM

Applications

Conclusion

Bibliography