08-10-2014, 10:37 AM
An Outlook of MRAM Technology Potential
An Outlook.pdf (Size: 388.02 KB / Downloads: 72)
INTRODUCTION
AGNETORESISTIVE Random Access Memory, or
MRAM, is showing promising potential for commercial
competition among universal memory [1]. Existing
semiconductor memories all have shortcomings and
limitations, and cannot fulfill all the important attributes
needed for memory with one solution. MRAM, however,
possesses all of these key features. It is the only non-volatile,
ARCHITECTURE & OPERATION
MRAM data is stored by changing the magnetic polarity of the
memory element. The storage element, called the Magnetic
Tunneling Junction (MJT), is composed of two magnetic
plates that are separated by a thin insulating layer. The thin
insulating layer allows tunneling when a current is applied
through the cell, allowing current to pass through the junction
for reading purposes. The MJT operates similar to a capacitor
such that data is stored between two metal plates. One plate is
made of soft ferromagnetic material, and the other is
permanent magnetic material. Essentially, we store data bits
by controlling the plates’ magnetic polarities. A memory
device consists of an array of these cells.
INDUSTRY
In order to gain a better understanding of the future trends of
MRAM we will now analyze the past and present MRAM
industry. The origins of the MRAM industry began in 1989
when several important discoveries were made by IBM on the
Giant Magnetoresistive (GMR) effect. It was soon realized
that this effect could be used in a memory circuit and several
other companies join the research and development of
MRAM. Some of these companies were: IBM, Freescale,
Infineon, NVE, and Cypress. The first chip containing MRAM
was manufacture in 2003, it had 128Kb of memory and was
built using 0.18um technology this chip was not made
. LIMITATIONS
MRAM has several limitations that need to be considered
when comparing it to other substitutes. A few of the major
issues that affect the viability of MRAM is cell density,
temperature compatibility during manufacturing with existing
processes, reliability and economical feasibility.
Density
Most basic MRAM cell sizes are limited to 180 nm or more
due to half-select problem. For toggle mode MRAM, this
becomes a problem near 90 nm. For feasible production,
MRAM cell size needs to be in the region of 65 nm, which
will require the spin-torque-transfer mode. This mode requires
switching more current through the control transistor than
conventional MRAM, requiring larger MRAM.
. CONCLUSION
The functionality, industry achievements, performance
parameters, and limitations of MRAM have been thoroughly
analyzed, followed by a comparison of MRAM against several
commercial memories. We conclude that a company’s
investment should be largely dependent upon the MRAM
application. For small memory requirements in extreme
environments where devices will be inaccessible for long
periods of time, such as satellites and industrial applications,
MRAM is currently a great investment. If a company needed
large amounts of memory but does not require a large number
of writes, flash is still the best available technology. From a
foundry’s point of view, investing in equipment to fabricate
MRAM would make them leaders in the MRAM industry.
Being the first major foundry with the ability to place MRAM
as the primary memory on a processor would be very
beneficial. As stated in [6], MRAM has been proven to
outperform SDRAM, the leading memory in processors.
Perfecting this MRAM integration could lead to a new wave
of products that were never realizable before.