spin torque transfer ram full report final
TRANSCRIPT
Bluetooth Low Energy
A SEMINAR REPORT
Submitted by
NITHIN THOMAS KANNANMANNIL
In partial fulfillment for the award of the degree
Of
BACHELOR OF TECHNOLOGY
IN
ELECTRONICS AND COMMUNICATION ENGINEERING
DEPARTMENT OF ELECTRONICS AND COMMUNICATION
ENGINEERING
MUSALIAR COLLEGE OF ENGINEERING AND TECHNOLOGY,
PATHANAMTHITTA, KERALA-689653
MG UNIVERSITY: KOTTAYAM
APRIL 2011
SPIN TORQUE TRANSFER RAM
A SEMINAR REPORT
Submitted by
NITHIN THOMAS KANNANMANNIL
In partial fulfillment for the award of the degree
Of
BACHELOR OF TECHNOLOGY
IN
ELECTRONICS AND COMMUNICATION ENGINEERING
DEPARTMENT OF ELECTRONICS AND COMMUNICATION
ENGINEERING
MUSALIAR COLLEGE OF ENGINEERING AND TECHNOLOGY,
PATHANAMTHITTA, KERALA-689653
MG UNIVERSITY: KOTTAYAM
APRIL 2011
MUSALIAR COLLEGE OF ENGINEERING AND
TECHNOLOGY, PATHANAMTHITTA
BONAFIDE CERTIFICATE
This is to certify that the seminar report entitled “ SPIN TORQUE
TRANSFER RAM” has been submitted by NITHIN THOMAS KANNANMANNIL in
partial fulfillment of the requirements for the award of degree of Bachelor of Technology in
Electronics and Communication Engineering is a bonafide record of the work carried out by
him under my guidance and supervision.
Prof. PAUL A J Asst.Prof.LIJESH L
Head of the Department SUPERVISOR
ECE, MCET ECE, MCET
ACKNOWLEDGEMENT
First of all I express my thanks to GOD the Almighty for showering infinite
blessing on me throughout the life and making me come up to this level. Also for the
immense grace, that strengthened me through the successful completion of this seminar
work.
I wish to convey my deep sense of gratitude to our beloved and respected
principal Dr. P G MATHEWS who helped me during the entire process of work.
I express my heartfelt gratitude to our Head of the Department of
Electronics and Communication Engineering, Prof .PAUL A J for the insight he has given for me
which has resulted in this prestigious seminar of my career.
I am grateful to my guide Mrs. DHANYA R, Assistant Professor, Department
of Electronics and Communication Engineering, for the valuable help and service she has
rendered to me and inspiring me during each stage of this work. Last but not the least I
express my love and gratitude for all my friends who showed great support and help.
ABSTRACT
Spin-Transfer Torque RAM (STT-RAM) is an emerging non-volatile memory
technology that is a potential universal memory that could replace SRAM in processor caches.
Magneto resistive RAM (MRAM) is a non-volatile memory technology in which a bit is
stored as the magnetic orientation of the free layer of a magnetic tunnel junction (MTJ). As
the free layer needs no current to maintain its state.The MTJ results in a high or low current
depending on whether the free layer is currently parallel or antiparallel to the magnetic
orientation of the hard layer. STT-RAM is a form of MRAM that uses spin transfer torque to
reorient the free layer by passing a large, directional write current through the MTJ. The
emerging Spin Torque Transfer memory (STT-RAM) is a promising candidate for future on-
chip caches due to STT-RAM’s high density, low leakage, long endurance and high access
speed. However, one of the major challenges of STT-RAM is its high write current, which is
disadvantageous when used as an on-chip cache.
STT-RAM cell uses magnetic tunnel junction (MTJ) to store binary data. An mtj
consists of two ferromagnetic layers reference layer and free layer and one tunnel barrier
layer. The magnetic direction of reference layer is fixed, while the magnetic direction of free
layer can be changed. The relative magnetization direction between the reference layer and
free layer results in different resistance of MTJ, which is used to represent the binary data
stored in the cell. When the magnetic field of the free layer and reference layer are parallel,
the MTJ resistance is low representing a logical “0”. When these two layers are in antiparallel
direction, the MTJ resistance is high which represents a logical “1”. STT-RAM combines the
advantages of all conventional memory. It combines the speed of SRAM. It is cost effective
and low-power memory solution like DRAM. There is no limit for write-read cycles and
radiation-resistant. It has greater performance, reliability and scalability.
TABLE OF CONTENTS
Introduction 4
1.1 Non volatile RAM................................................................................................ 41.2 Spin.............................................................................................................................. 61.3 Spin Transfer
Introduction
1.4 Spintronics.................................................................................................................. 71.4.1 Metal based spintronics devices.....................................................................71.4.2 Semiconductor based spintronics devices......................................................9
1.5 Tunnel Magneto resistance effect (TMR).................................................................. 101.6 STTRAM.....................................................................................................................10
1.7 History.........................................................................................................................131.8 Universal memory concept..........................................................................................14
2 Working of STTRAM
2.1 Spin torque transfer RAM...........................................................................................152.2 Writing '0' and '1'........................................................................................................ 162.3 Reading '0' and '1'........................................................................................................18
Fabrication of STTRAM…………………………………………………………... 22
3 Research undergoing on STTRAM 23
Scope of STTRAM………………………………………………………………….. 28
Comparison with conventional memory RAM…………………………………… 30
4 Applications STTRAM in Aviations and Military……………………………….. 32
IC designers benefits…………………………………………………………………34
Conclusions …………………….…………………………………………………
…35References………………………...……………………………………………………36
LIST OF FIGURES & TABLES
FIGURE 1.2 : SPIN 4
FIGURE 2.3 : MAGNETIC TUNNEL JUNCTION 11
FIGURE 4.1.1: NANOMAGNETS USED TO CONTROL THE SPIN 13
FIGURE 4.1.2: MATERIAL USED IN MTJ ......14
FIGURE 4.1.3: THE MTJ STATE CHANGES FROM PARALLEL TO
ANTIPARALLEL
FIGURE 4.1.4: SIMULATION MODEL SYMBOL IN THE SPECTRE
SIMULATOR
FIGURE 4.1.5: RESISTANCE EQUIVALENT CIRCUIT 15
FIGURE 4.1.6: STT-RAM ARCHITECTURE 15
FIGURE 4.1.7: WRITING LOGIC 0 AND LOGIC 1 IN STT-RAM 16
FIGURE 4.1.8: CONVENTIONAL MRAM AND STT-RAM 18
FIGURE 4.2.1: ION MILLING AND LIFT-OFF METHOD 19
FIGURE 4.2.2: STT-RAM OVERCOMING POWER TRADE OFF PROBLEM 20
FIGURE 4.2.3: DC AND TRANSIENT SIMULATION 21
FIGURE 4.2.4: DC SIMULATION OF MTJ (PARALLEL) 21
FIGURE 4.2.5: TRANSIENT SIMULATION OF MTJ FOR PARALLEL AND
ANTI-PARALLEL 22
FIGURE 6.4.1: INPUT SPIN-LUT ARCHITECTURE 31
FIGURE 6.4.2: THE FULL SCHEMATIC OF SPIN-MTJ BASED NON-VOLATILE FLIP-FLOP (SPIN-FF)………………………………………. 31
FIGURE 6.4.3: MTJ MEMORY CELLS ARE IMPLEMENTED ABOVE THECMOS CIRCUITS
FIGURE 6.4.4: FULL LAYOUT (5.65UMx10.15UM) OF SPIN-FF..........................................32
TABLE 6.2.1 : STT-RAM PROTOTYPE VERSUS EXISTING MEMORY CHIP...................28
LIST OF SYMBOLS & ABBREVIATIONS1. µ : INTRINSIC MAGNETIC MOMENT OF THE SPIN
POLARIZED ELECTRON.
2. Å : ANGSTROM=10-10 meters.
3. CMOS : COMPLEMENTARY METAL OXIDE SEMICONDUCTOR.
4. DRAM: DYNAMIC RANDOM ACCESS MEMORY.
5. ESPV : EXCHANGE-BIASED SPIN VALVES.
6. F: FERMI= 10 -15 meters.
7. FPGA : FIELD PROGRAMMABLE GATE ARRAY.
8. GMR : GAINT MAGNETO RESISTANCE.
9. MCP : MULTI-CHIP PACKAGES.
10 MRAM : MAGNETIC RANDOM ACCESS MEMORY.
11 MTJ : MAGNETIC TUNNEL JUNCTIONS.
12 NVRAM : NON- VOLATILE RANDOM ACCESS MEMORY.
13 RAP : RESISTANCE INTRODUCED IN TMR DUE TO ANTI
PARALLEL ACTION IN THE FREE FERROMAGNETIC
LAYER.
14 Rp : RESISTANCE INTRODUCED IN TMR DUE TO
PARALLEL ACTION IN THE FREE FERROMAGNETIC
LAYER.
15 SPIN-LUT : SPIN LOOK UP TABLE.
16 SPIN-FF : SPIN FLIP FLOP
17 SOC : SYSTEM ON CHIP
.
18. SRAM : STATIC RANDOM ACCESS MEMORY
19.STS: SPIN TRANSFER SWITCHING
20.STT-RAM: SPIN TORQUE TRANSFER RANDOM ACCESS MEMORY
21.TMR: TUNNEL MAGNETO RESISTANCE22.VLSI: VERY LARGE SCALE INTEGRATION
CHAPTER-1:BASIC TERMINOLOGIES
1.1 NON-VOLATILE RAM
Non-Volatile Random Access Memory (NVRAM) is the general name used to describe
any type of random access memory which does not lose its information when power is turned off.
This is in contrast to the most common forms of random access memory today, DRAM and
SRAM, which both require continual power in order to maintain their data. NVRAM is a
subgroup of the more general class of non-volatile memory types, the difference being that
NVRAM devices offer random access, like hard disks.
1.2 SPIN
In quantum mechanics, spin is a fundamental property of atomic nuclei, hadrons, and
elementary particles. For particles with non-zero spin, spin direction (also called spin for short) is
an important intrinsic degree of freedom.
As the name indicates, the spin has originally been thought of as a rotation of particles
around their own axis. This picture is correct insofar as spins obey the same mathematical laws as
do quantized angular momenta. On the other hand, spins have some peculiar properties that
distinguish them from orbital angular momenta: spins may have half-integer quantum numbers,
and the spin of charged particles is associated with a magnetic dipole moment in a way (g-factor
different from 1) that is incompatible with classical physics.
The electron spin is the key to the Pauli Exclusion Principle and to the understanding of
the periodic system of chemical elements. Spin-orbit coupling leads to the fine structure of atomic
spectra, which is used in atomic clocks and in the modern definition of the fundamental unit
second. Precise measurements of the g-factor of the electron have played an important role in the
development and verification of quantum electrodynamics. Electron spins play an important role
in magnetism, with applications for instance in computer memories. Manipulation of spins in
semiconductor devices is the subject of the developing field of spintronics. The manipulation of
nuclear spins by radiofrequency waves (nuclear magnetic resonance) is important in chemical
spectroscopy and medical imaging. The photon spin is associated with the polarization of light.
The head-on collision of a quark (red ball) from one proton (orange ball) with a
gluon (green ball) from another proton with opposite spin, spin is represented by the blue
arrows circling the protons and the quark. The blue question marks circling the gluon
represent the question: Are gluons polarized? Ejected from the collision are a shower of
quarks and a photon of light (purple ball).
Particles with spin can possess a magnetic dipole moment, just like a rotating electrically charged
body in classical electrodynamics. These magnetic moments can be experimentally observed in
several ways, e.g. by the deflection of particles by inhomogeneous magnetic fields in a Stern-
Gerlach experiment, or by measuring the magnetic fields generated by the particles themselves.
The intrinsic magnetic moment p of an elementary particle with charge q, mass m, and spin S, is
Where the dimensionless quantity g is called the g-factor. For exclusively orbital rotations it
would be 1.
In ordinary materials, the magnetic dipole moments of individual atoms produce magnetic fields
that cancel one another, because each dipole points in a random direction. Ferromagnetic
materials below their Curie temperature, however, exhibit magnetic domains in which the atomic
dipole moments are locally aligned, producing a macroscopic, non-zero magnetic field from the
domain. These are the ordinary "magnets" with which we are all familiar.
Figure 1.2 SPIN
The study of the behavior of such "spin models" is a thriving area of research in
condensed matter physics. For instance, the Ising model describes spins (dipoles) that have only
two possible states, up and down, whereas in the Heisenberg model the spin vector is allowed to
point in any direction. These models have many interesting properties, which have led to
interesting results in the theory of phase transitions.
1.3 SPIN TRANSFER
Spin transfer is the phenomenon in which the spin angular momentum of the charge
carriers (usually electrons) gets transferred from one location to another. This phenomenon is
responsible for several important and observable physical effects.
Most famously, spin polarized current passing into a nanoscale magnet tends to deposit
some of its spin angular momentum into the magnet, thereby applying a large torque to the
magnetization. This enables magnetic manipulations far more efficiently than can be achieved
with magnetic fields alone, especially as device applications shrink in scale. In the hard disk
industry, where a series of nanoscale magnetic layers called a spin valve is often used to measure
the small local magnetic fields above the disk surface, this is an undesirable effect, as it hinders
the ability to measure the state of the valve without disturbing it. In the MRAM industry,
however, this effect may prove incredibly useful in reducing power consumption.
CHAPTER 2:
INTRODUCTION
2.1. SPINTRONICS:
Spintronics (a neologism meaning "spin transport electronics"), also known as magneto
electronics, is an emerging technology which exploits the intrinsic spin of electrons and its
associated magnetic moment, in addition to its fundamental electronic charge, in solid-state
devices.
Electrons are spin-1/2 fermions and therefore constitute a two-state system with spin "up"
and spin "down". To make a spintronic device, the primary requirements are to have a system that
can generate a current of spin polarized electrons comprising more of one spin species - up or
down - than the other (called a spin injector), and a separate system that is sensitive to the spin
polarization of the electrons (spin detector). Manipulation of the electron spin during transport
between injector and detector (especially in semiconductors) via spin precession can be
accomplished using real external magnetic fields or effective fields caused by spin-orbit
interaction.
Spin polarization in non-magnetic materials can be achieved either through the Zeeman
Effect in large magnetic fields and low temperatures, or by non-equilibrium methods. In the latter
case, the non-equilibrium polarization will decay over a timescale called the "spin lifetime". Spin
lifetimes of conduction electrons in metals are relatively short (typically less than 1 nanosecond)
but in semiconductors the lifetimes can be very long (microseconds at low temperatures),
especially when the electrons are isolated in local trapping potentials (for instance, at impurities,
where lifetimes can be milliseconds).
9.1 2.1.1 Metals-based spintronic devices
The simplest method of generating a spin-polarized current in a metal is to pass the
current through a ferromagnetic material. The most common application of this effect is a Giant
Magneto Resistance (GMR) device. A typical GMR device consists of at least two layers of
ferromagnetic materials separated by a spacer layer. When the two magnetization vectors of the
ferromagnetic layers are aligned, the electrical resistance will be lower (so a higher current flows
at constant voltage) than if the ferromagnetic layers are anti-aligned. This constitutes a magnetic
field sensor. Two variants of GMR have been applied in devices:
• Current-In-Plane (CIP), where the electric current flows parallel to the layers.
• Current-Perpendicular-to-Plane (CPP), where the electric current flows in a direction
perpendicular to the layers.
Other metals-based spintronics devices:
• Tunnel Magneto Resistance (TMR), where CPP transport is achieved by using quantum-
mechanical tunneling of electrons through a thin insulator separating ferromagnetic layers.
• Spin Torque Transfer , where a current of spin-polarized electrons is used to control the
magnetization direction of ferromagnetic electrodes in the device.
9.1.1 The storage density of hard drives is rapidly increasing along an exponential
growth curve, in part because spintronics-enabled devices like GMR and
TMR sensors have increased the sensitivity of the read head which measures
the magnetic state of small magnetic domains (bits) on the spinning platter.
The doubling period for the areal density of information storage is twelve
months, much shorter than Moore's Law, which observes that the
number of transistors that can cheaply be incorporated in an integrated
circuit doubles every two years.
MRAM, or magnetic random access memory, uses arrays of TMR or Spin torque transfer
devices. MRAM is nonvolatile (unlike charge-based DRAM in today's computers) so information
is stored even when power is turned off, potentially providing instant-on computing. Motorola
has developed a 256 kb MRAM based on a single magnetic tunnel junction and a single
transistor. This MRAM has a read/write cycle of fewer than 50 nanoseconds. Another design in
development, called Racetrack memory, encodes information in the direction of magnetization
between domain walls of a ferromagnetic metal wire.
9.2 2.1.2 Semiconductor-based spintronic devices
In early efforts, spin-polarized electrons are generated via optical orientation using
circularly-polarized photons at the band gap energy incident on semiconductors with appreciable
spin-orbit interaction (like GaAs and ZnSe). Although electrical spin injection can be achieved in
metallic systems by simply passing a current through a ferromagnet, the large impedance
mismatch between ferromagnetic metals and semiconductors prevented efficient injection across
metal-semiconductor interfaces. A solution to this problem is to use ferromagnetic semiconductor
sources (like manganese-doped gallium arsenide GaMnAs), increasing the interface resistance
with a tunnel barrier, or using hot-electron injection.
Spin detection in semiconductors is another challenge, which has been met with the
following techniques:
• Faraday/Kerr rotation of transmitted/reflected photons
• Circular polarization analysis of electroluminescence
• Nonlocal spin valve (adapted from Johnson and Silsbee's work with metals)
• Ballistic spin filtering
The latter technique was used to overcome the lack of spin-orbit interaction and materials
issues to achieve spin transport in Silicon, the most important semiconductor for electronics.
Because external magnetic fields (and stray fields from magnetic contacts) can cause large
Hall effects and magneto resistance in semiconductors (which mimic spin-valve effects), the only
conclusive evidence of spin transport in semiconductors is demonstration of spin precession and
de-phasing in a magnetic field non-collinear to the injected spin orientation. This is called the
Hanle effect.
Advantages of semiconductor-based spintronics applications are potentially lower power
use and a smaller footprint than electrical devices used for information processing. Also,
applications such as semiconductor lasers using spin-polarized electrical injection have shown
threshold current reduction and controllable circularly polarized coherent light output. Future
applications may include a spin-based transistor having advantages over MOSFET devices such
as steeper sub-threshold slope.
2.2 TUNNEL MAGNETORESISTANCE EFFECT (TMR)
In physics, the Tunnel Magneto Resistance effect (TMR), occurs when a current flows
between two ferromagnets separated by a thin (about 1 nm) insulator. Then the total resistance of
the device, in which tunneling is responsible for current flowing, changes with the relative
orientation of the two magnetic layers. The resistance is normally higher in the anti-parallel case.
The effect is similar to Giant Magneto Resistance except that the metallic layer is replaced by an
insulating tunnel barrier.
2.3 STT-RAM
STT-RAM is underway to fine tune a digital-data-recording technology, which will lead
to durable, high density memory chips impervious to radiation and capable of virtually unlimited
read/write cycles called SPIN TRANSFER TORQUE RANDOM ACCESS MEMORY (STT-
RAM) chips, the devices are expected to be cost-competitive with conventional magnetic memory
chips and provide a "revolution in military and space electronics".
STT-RAM chips use SPINTRONICS, a technology that controls the spin of an electron to
record binary data- the zeroes and ones of digital language. (Spintronics is short for "spin
electronics").
An STT-RAM chip is somewhat similar, in manufacture at least, to the conventional
magnetic memory chips used in computers known as SRAM (STATIC RANDOM ACCESS
MEMORY), DRAM (DYNAMIC RANDOM ACCESS MEMORY) and MRAM
(MAGNETORESISTIVE RANDOM ACCESS MEMORY-a relatively recent nonvolatile chip
technology), and to NOR and NAND flash memory chips.
Toward the end of the manufacturing process, however, a magnet and thin film structure
are added to the chip. Rather than using a magnetic field to write zeroes and ones as with
conventional chips, an STT-RAM chip records digital data by passing a current through the
magnet and then over the film structure, which is about 700A thick. As the current moves through
the magnet, it becomes polarized. Transferring the current through a pre-layer of film 20A thick
creates torque. The torque changes the direction of the polarized current and the orientation of the
pre-layer to the film below. By adjusting the current that passes through the magnet, it can be
made to move the pre-layer in parallel or low- resistance direction to the film, which is recorded
zero. If the current is adjusted so the pre-layer moves counter parallel or in a high-resistance
direction to the film, a one recorded.
STT-RAM is a novel nonvolatile memory which utilizes spin torque transfer
magnetization switching, the spin-RAM is programmed by magnetization reversal through an
interaction of a spin momentum-torque-transferred current and a magnetic moment of memory
layers in Magnetic Tunnel Junctions (MTJs), and therefore an external magnetic field is
unnecessary as that for a conventional MRAM.
In this spin transfer torque switching technique, data is written by re-orienting the
magnetization of a thin magnetic layer in a Tunnel Magneto Resistance (TMR) element using a
spin-polarized current. An electrical current is generally unpolarized (consisting of 50% spin-up
and 50% spin-down electrons), a spin polarized current is one with more electrons of either spin.
By passing a current through a thick magnetic layer one can produce a spin polarized current. It
uses currents of spin-aligned electrons (spin-polarized currents) rather than fixed magnetic fields
for processing and storing information. Spin-polarized currents can be generated by driving a
current through a ferromagnetic layer. If the ferromagnetic electrode has a magnetization aligned
in one direction, the magnetic moments of electrons passing through it become aligned in the
same direction. It was theoretically predicted that such a spin-polarized current could transfer its
angular momentum to a second magnetic layer and therefore switch the magnetization of this
second layer into an alignment parallel with the reference layer, provided the dimensions of the
device are about 100 nm or less. Further research into the STT phenomena has led to new
materials and the replacement of metallic films with MTJs. MTJ resistance levels can be adjusted
through material selection and MTJs also provide greater difference between the resistance states
of the cell. Operation at room temperature has been reported using Aluminum Oxide as the
tunneling barrier, and more recent efforts with Magnesium Oxide have shown promise for a
reduction in power. Based on innovations in the MTJ, it is quite possible to develop a very
compact and scalable memory that exploits spin and avoids the circuit issues surrounding prior
art. With the optimal choice of materials, this STT memory would have fast read/write and be
non-volatile, low power, and scalable. It would also be uncomplicated to fabricate and could
conceivably be done in a silicon foundry if the process and materials were designed for
compatibility as well. This seeks innovative research proposals for achieving a STT technology
for dense solid-state memories for defense applications.
CHAPTER 3:
HISTOR Y
Prior research in micro-magnetic and spintronics has led to the exploitation of Giant
Magneto Resistance (GMR) effects for rotating magnetic disk drive memories, as well as
Magnetic Tunneling Junctions (MTJ) for Magnetic Random Access Memory (MRAM). The Spin-
Torque Transfer (STT) switching effect is a new physics phenomenon that exploits magnetic spin
states to electrically change the magnetic orientation of a material that was theoretically predicted
in 1996 and first demonstrated in metallic thin films as recently as 2000. Although STT switching
Reference Layer----------
Pinning layer —■—
Figure 2.3: Magnetic tunnel junction. Resistance is low when the magnetization of the
reference and storage layer is aligned in same direction and high when the layers aligned in
opposite direction.
Storage Layer_______Tunnel Barrier--------
'0 "state Magnetic Field (Oe)
currents were initially orders of magnitude too high for application in practical devices, Grandis
was founded in 2002 with the goal of developing a novel non-volatile
memory technology that applies the many benefits of STT switching
to Magnetic Tunnel Junctions (MTJs). Researchers at Grandis undertook
pioneering research in spintronics and pursued new magnetic materials
and innovative MTJ structures to lower STT switching currents. Through
these advances in materials research, coupled with its extensive modeling,
simulation, integration, cell architecture, circuit and system design
capabilities, Grandis has developed a package that enables its licensees to
incorporate stand-alone or embedded STT-RAM nonvolatile memory into
their products.
The prototype chips are available for testing. The different companies are working hard to reduce
Time-To-Market. The STT-RAM will be available in the market by this year or by 2010.
CHAPTER 4:
WORKING OF STT-RAM
4.1 SPIN-TRANSFER TORQUE RAM as universal memory
Nanomagnet are used to control the spin. By passing electrons through
the nanomagnet, the spin of the electrons can be
alignedsame
as the
nanomagnet.
4.1.1, below a
magnetization
direction As shown
in Figure Magnetic
TunnelPage 14
in the
Junction (MTJ) is at the heart of a STT-RAM bit cell. The MTJ consists of two ferromagnetic
electrodes with a thin insulating layer in-between. The top nanomagnet is the storage layer or
"free" layer; the middle is the barrier; and the bottom nanomagnet is the reference or "pinned"
layer or spin filter.
Figure 4.1.1: Nanomagnets used to control the spin.
Spin Transfer Switching (STS) changes the MTJ's state from antiparallel or "1" to parallel
or "0" and vice versa. This is performed by running current from the top to the bottom of the MTJ
and vice versa. An STT-RAM chip addresses each bit individually by flowing current directly
through the bit.
Consequently, unintended writing errors are completely eliminated. A conventional
CMOS transistor below the MTJ produces the current. In this instance, switching is performed via
spin polarized currents. By polarizing the current, data is passed from the fixed MTJ layer that is
the polarizer to the free MTJ layer.
The materials used as ferromagnet in MTJ is CoFe and the layer in between the
ferromagnet i.e., tunnel barrier layer is made of AlxO or MgO, as shown in figure4.1.2 below.
Figure 4.1.2: Material used in MTJ.
TMR= (Rap-Rp) / R a p , Tunnel Magneto Resistance Ratio
CoFe
CoFe
CoFe
CoFe
Figure 4.1.3: The MTJ state changes from Parallel
(P) to Antiparallel (AP) if the positive direction
current density I>Ic, on the contrast, its state will
return if the negative direction current density I >
Ic. Where Ic is critical current density.
This process is called spin-transfer torque
switching. Current running through the fixed layer
polarizes the electrons. Those polarized electrons then
affect the switching of the free layer, hence the parallel
and anti-parallel configurations.
Unlike the STT-RAM cell that exploits electron
spins for writing, the conventional MRAM cell, shown
CoFe
CoFe
CoFe
CoFe
in Figure 4.1.8 at 20 to 30 F2 cell size uses a magnetic
field to perform switching. When the cell is activated,
the bit line and write word line become active with
current pulses flowing through them, thus creating a
magnetic field around the bit.
Figure. 4.1.4 Simulation model symbol in the specter simulator.
Figure 4.1.5: Resistance equivalent circuit
Figure-4.1.6! STT-RAM architecture.
Figure 4.1.7: STT-RAM writing logic 0 and logic 1- it addresses each bit individually by
flowing current directly through the bit. Unintended writing errors are completely eliminated.
It is this magnetic field that changes the state of the bit from "1" to "0" and vice versa.
The scalability issue is exacerbated as a conventional MRAM chip is subjected to increasingly
smaller geometries. As feature size shrinks, more current is required to create the magnetic field.
In a conventional MRAM cell, conductors or wires above and below the MTJ are used to
generate the fields necessary to switch the state of the free magnetic layer. The additional write word
line and bypass line required in this cell geometry translate into a more complicated architecture.
Hence, more lithographic steps and a more costly manufacturing process are required.
Each conductor or wire requires a minimum of 5 to 10 mill amperes (mA) of current to
perform a switch. Conversely, STT-RAM technology incurs a considerably lower switching current
on the order of 100 microamperes, as a result of its more efficient spin-transfer torque techniques.
A cell's thermal instability can erroneously trigger random switching to the opposite state.
This means the cell stability in conventional MRAM demands an increasingly higher current
flowing in the conductors or wires to efficiently switch states. Due to this bit or write disturbance,
the number of bad bits in conventional MRAM is very high. Consequently MRAM vendors must
resort to redundancy architectures with as much as 25 percent redundancy.
Shown at a cell size of 6 F2, the STT-RAM cell, Figure 4.1.8 on the other hand, does not
require the metal wire below the MTJ. The write word line, bypass line, and cladding associated
with the conventional MRAM cell are eliminated. Instead, current flows perpendicular through the
MTJ memory cell. As STT-RAM technology is scaled and its cells become increasingly smaller, the
cross-sectional area becomes smaller and smaller.
Figure-4.1.8: Conventional MRAM and STT-RAM cell - Magnetic field, generated by the bit
line and write word line, is used to switch between the '0' and '1' states. STT-RAM cell by
eliminating the write word line and bypass line, it is considerably smaller than the first
generation MRAM cell.
Because STT-RAM uses a current running through the cell, the required writing current that
flows through the smaller MTJs decreases. As a result, STT-RAM has superior scaling properties.
20-30 P
Wrti Wyd Liu(« Pint-genenttoo WRAM cdl it* STT-RAM c*l
That is the reason total required current in STT-RAM continues to be considerably less with
increasingly smaller geometries, whereas in conventional MRAM, the required switching current
increases with scalability, as shown in Figure 4.2.2.
Attempts have been made to correct the half-select bit write disturbance issue with
conventional MRAM by modifying its bit cell. One such technique is the toggle MRAM bit cell. Its
intent is to increase the operating window of the write operation. The objective is to create a
significant margin between the level of fields required for switching all bits and the level at which
write disturbs occur.
The toggle MRAM bit cell adds a second free magnetic layer and an anti-ferro-magnetic
coupling layer above the first free magnetic layer. Studies performed by the conventional MRAM
vendor reveal an increasing percentage of bits properly switch as a result of applying the new toggle
MRAM bit cell. However, it incurs higher currents (nearly 2X higher) and tighter thickness control
is required (control within one atomic layer).
DC and transient simulations have been performed to verify the bias-voltage dependent
resistance, TMR effect, the switching current and performance from parallel configuration (P) to
antiparallel configuration (AP) and AP to P (see Figure 4.2.3 to 4.2.5).
4.2 FABRICATION OF STT-RAM:
The fabrication of STT-RAM is done using Ion-milling and Lift-off Method.
Figure 4.2.1: ION MILLING AND LIFT-OFF METHOD
Figure 4.2.2: STT-RAM overcoming power trade off problem - "Total required current in
STT-RAM continues to scale lower with increasingly smaller geometries. Conversely,
conventional MRAM switching current increases with smaller geometries.
Figure 4.2.3: DC and Transient simulation.
{sum i:if.os:iip>
Parallel Anti-Parallel
Figure 4.2.5 Transient simulation of MTJ, the critical current is about 153.206uA for parallel
and -221.202uA for anti-parallel
CHAPTER5:
RESEARCH UNDERGOING ON STT-RAM IN DIFFERENT
ORGANISTION
• A NOVEL NONVOLATILE MEMORY WITH SPIN TORQUE TRANSFER
MAGNETIZATION SWITCHING: SPIN-RAM:
Hosomi, M. Yamagishi, H.
Yamamoto, T. Bessho, K. Higo, Y. Yamane, K. Yamada, H. Shoji,
M. Hachino, H. Fukumoto, C. Nagao, H. Kano, H. Sony Corp.,
Kanagawa;This new programming mode has been accomplished
owing to tailored MTJ, which has an oval shape of 100 times 150
nm. The memory cell is based on a 1-transistor and a 1-MTJ
structure. The 4kbit spin-RAM was fabricated on a 4 level metal,
0.18 mum CMOS process. In this work, writing speed as high as 2
ns, and a write current as low as 200 muA were successfully
demonstrated.
• NANOELECTRONIC S RESEARCH INSTITUTE- JAPAN:
To realize a large capacity Magnetic Random Access Memory (MRAM) that uses spin-
transfer switching for writing, it is essential to evaluate thermal durability and intrinsic critical
currents correctly. Here, we examined the theoretically predicted logarithmic relationship between
critical currents of spin-transfer switching and duration of injected pulsed currents using Giant
Magneto Resistive (GMR) samples with different magnetic materials, e.g., Co, Co-Fe25, and
CoFeB. This relationship was verified for the samples by giving reasonable thermal- durability
coefficients and intrinsic critical currents as fitting parameters. We found that thermal durability
was underestimated when an effective magnetic field acted on magnetic memory cells antiparallel to
their magnetization. We then experimentally demonstrated that thermal assistance in spin-transfer
switching decreased with increasing thermal durability.
• GRANDIS INC., 1123 CADILLAC COURT, MILPITAS, CALIFORNIA:
Dual Magnetic Tunnel Junction (MTJ) structures consisting of two MgO insulating barriers
of different resistances, two pinned reference layers aligned antiparallel to one another, and a free
layer embedded between the two insulating barriers have been developed. The electron transport
and spin dependent resistances in the dual MTJ structures are accounted for by sequential tunneling
with some spin-flip relaxation in the central electrode (the free layer). With a tunneling magneto
resistance ratio of 70%, a switching current density Jc (at 30 ms) of 0.52 MA/cm2 is obtained,
corresponding to an intrinsic value of Jc0 (at 1ns) of 1.0 MA/cm2. This
value of Jc0 is 2-3 times smaller than that of a single MgO insulating barrier MTJ structure and
results from improvements in the spin-transfer torque efficiency. The asymmetry between
JcAPand Jc P mP are significantly improved, which widens the read-write margin for
memory device design. In addition, the experimental results show that the switching current density
can be further reduced when an external field is applied along the hard axis of the free layer.
Grandis is looking for STT-RAM to replace existing memory technologies at 45nm and beyond.
Since most processing at such advanced technology nodes is done on 300mm MTJ wafers.
• JAPAN SCIENCE AND TECHNOLOGY AGENCY, SENDAI 980-8579, JAPAN:
SUBSTANTIAL REDUCTION OF CRITICAL CURRENT FOR MAGNETIZATION
SWITCHING IN AN EXCHANGE-BIASED SPIN VALVE-
Great interest in current-induced magnetic excitation and switching in a magnetic nanopillar
has been caused by the theoretical predictions11, 12 of these phenomena. The concept of using a
spin-polarized current to switch the magnetization orientation of a magnetic layer provides a
possible way to realize future 'current-driven' devices13: in such devices, direct switching of the
magnetic memory bits would be produced by a local current application, instead of by a magnetic
field generated by attached wires. Until now, all the reported work on current- induced
magnetization switching has been concentrated on a simple ferromagnet/Cu/ferromagnet trilayer.
Here we report the observation of current-induced magnetization switching in Exchange-Biased
Spin Valves (ESPVs) at room temperature. The ESPVs clearly show current- induced magnetization
switching behavior under a sweeping direct current with a very high density. We show that insertion
of a ruthenium layer between an ESPV nanopillar and the top electrode effectively decreases the
critical current density from about 108 to 107 Acm -2. In a well-designed 'antisymmetric' ESPV
structure, this critical current density can be further reduced to 106 Acm -2. We believe that the
substantial reduction of critical current could make it possible for current-induced magnetization
switching to be directly applied in spintronic devices, such as magnetic random-access memory.
• VORTEX POLARITY SWITCHING BY A SPIN--POLARIZED CURRENT Jean-
Guy Caputo, Yuri Gaididei, Franz G. Mertens, Denis D. Sheka
The spin-transfer effect is investigated for the vortex state of a magnetic nanodot. A spin
current is shown to act similarly to an effective magnetic field perpendicular to the nanodot. Then a
vortex with magnetization (polarity) parallel to the current polarization is energetically favorable.
Following a simple energy analysis and using direct spin—lattice simulations, we predict the
polarity switching of a vortex. For magnetic storage devices, an electric current is more effective to
switch the polarity of a vortex in a nanodot than the magnetic field.
• DISTRIBUTION OF THE MAGNETIZATION REVERSAL DURATION IN
SUBNANOSECOND SPIN-TRANSFER SWITCHING
T. Devolder, C. Chappert, J. A. Katine, M. J. Carey, and K. Ito Institute d'Electronic
Fundamental, CNRS UMR 8622, University Paris Sud, Bat, 220, 91405 Orsay, France.
Hitachi Cambridge Laboratory, Hitachi Europe, Ltd., Cavendish Laboratory, Madingley Road,
Cambridge CB3 0HE, United Kingdom. Hitachi GST, San Jose Research Center, 650 Harry Road,
San Jose, California 95120, USA.
The experimental distribution of switching times in spin-transfer switching induced by sub-
ns current pulses in a pillar-shaped spin valve, whose free layer easy axis is parallel to the spin
polarization of the current. The pulse durations leading to successful switching events follow a
multiply stepped distribution. The step positions reflect the precessional nature of the switching.
Modeling indicates that the switching proceeds through an integer number of gradually amplified
precession cycles. This number is determined by the initial magnetization state. The switching
probability distribution can be modeled considering the thermal variance of the initial magnetization
orientation and by analyzing the occurrence of a vanishing total torque condition in the set possible
magnetization trajectories. Modeling helps us to understand why switching cannot happen with a
reproducible sub-ns duration when the free layer easy axis is parallel to the spin polarization of the
current. To circumvent that problem, we propose to bias the spin valve with a hard axis field, which
could provide an increased reproducibility of the switching duration.
• A NOVEL SPRAM (SPIN-TRANSFER TORQUE RAM) WITH A SYNTHETIC
FERRIMAGNETIC FREE LAYER FOR HIGHER IMMUNITY TO READ DISTURBANCE
AND REDUCING WRITE-CURRENT DISPERSION:
Miura, K. Kawahara, T. Takemura, R. Hayakawa, J. Ikeda, S. Sasaki, R. Takahashi, H. Matsuoka,
H. Ohno, H.Hitachi Ltd., Tokyo;
A novel SPRAM (spin-transfer torque RAM) consisting of MgO-barrier-based Magnetic
Tunnel Junctions (MTJs) with a Synthetic Ferromagnetic (SyF) structure in a free layer was
demonstrated for both higher immunity to read disturbance and a sufficient margin between the read
and write currents. Since magnetization of the free layer becomes stable against thermal fluctuation
with increasing thermal-stability factor E/kBT, the SyF free layer of the MTJs
realized magnetic information retention of over 10 years due to its high E/kBT of 67.
Furthermore, it was found that the SyF free layer has an advantage of reducing dispersion of
write-current density Jc, which is necessary for securing an adequate margin between the read
and write currents.
CHAPTER 6:
SCOPE OF STT-RAM
6.1. ADVANTAGES OF STT-RAM CHIPS OVER
CONVENTIONAL CHIPS
• The magnetic film on which digital data are recorded in conventional chips is prone to
"write disturbance" under some conditions, which affects accuracy and retention. But this is
not in case of STT-RAM.
• Conventional magnetic films have a scalability problem, this not in case of STT-RAM.
• STT-RAM chip use less current as the technology becomes smaller in size, so their cells can
shrink with no tradeoffs in power. This is all dependent on the concept "more current is
needed as the chip sizes shrink to create magnetic film, which makes the cell size bigger
because a higher current requirement affects the size of transistors".
• STT-RAM technology is the front runner because it offers embedded designers the best of
all worlds.
• An external magnetic field is unnecessary as that for a conventional MRAM.
• The STT-RAM chips comparatively require less power than conventional chips.
• 1.2V is enough to run STT-RAM chip, where as DRAM chips require 2V, a NOR flash
memory chip needs 6-8V and a NAND chip uses 16-20V.
• In terms of speed STT-RAM write as fast as 2 nanosecond, which is comparable to MRAM
chips.
• The read/ write cycle for all practical purposes infinite (in excess of 10 15- 1,000 trillion-
cycles).
• The user won't need to move data from one format to another when using STT-RAM,
because it as the ability to work with virtually all systems equally well.
6.2. STT-RAM versus other Conventional Memory Chip
As shown in Table 6.2.1 below, STT-RAM technology will provide embedded designers
better features compared with conventional memory technologies and future "universal memory"
candidates.
With a two nanosecond (ns) write time, STT-RAM is as fast as SRAM, which currently has
a write time ranging from 1 to 100 ns, depending on the technology used. As far as cell size, STT-
RAM fares much better than SRAM cell size. When STT-RAM reaches the 32 nm technology node,
the cell will be equal to or smaller than DRAM or NOR flash.
Nan-volatility
Silt
ftHAH Flash(N0A)
Y« i Y« Y«
1640 6 1? 16 34
ID
Write/ E f a t c Time (nj)
Write power
Other power MOS Sub-
Refresh
consumption tfrHltttt tea it curat
3-20 20-50 2080 3-1S
3 20 JO/30 SO/SO lnVlOrns
Mid-Htgh Low Hid High
Red Time
>10« 10U I011 10*Endurance
None None Hone None
Comparing STT-RAM with recent universal memory candidates, read and write/erase time
are each at 2 to 20 nanoseconds (ns) for STT-RAM, while phase-change RAM (PRAM) takes 20
to 50 ns for a read, 30 ns for write. Further, STT-RAM endurance is unlimited at >1015 while
PRAM is less than 1012.
Although it is not a new technology, PRAM has recently received considerable attention.
A number of leading semiconductor companies are investing in PRAM and developing their own
sets of IP. While it is an admirable technology, PRAM has limited endurance and slow speed
compared with STT-RAM.
Chalcogenide material, usually Ge2Sb2Te5 also known as GST, is used in a PRAM for
data storage. The PRAM uses the reversible phase change between the crystalline and amorphous
states of Chalcogenide GST by applying heat. Crystalline GST has low resistivity and amorphous
GST has high resistivity. The data "0" corresponds to the crystalline state, while data "1" is
associated with the amorphous state.
Switching time between states is larger than 20 ns, which means PRAM cannot replace
SRAM since it is considerably slower. The time required to reset the state of the bit must be made
long. If it is not, the phase change material cools too fast to achieve the crystalline state. Further,
due to the constantly changing of the phases and the heat applied to them, there is material
degradation and therefore limited endurance.
RRAM, or resistive memory, relies on a resistance change caused by an electric field. It is
in an earlier state of development compared with PRAM, but suffers similar problems in terms of
reliability and endurance. In addition, the switching mechanisms involved in the many proposed
RRAM materials are not well understood.
FRAM, or ferroelectric RAM, uses a ferroelectric material to store a polarization. It
suffers from limited read endurance and a destructive read process. Also, its endurance of around
1012 cycles, while suitable for a flash replacement, cannot be used as universal memory.
6.3. STT-RAM Application Outlook:
STT-RAM is a disruptive technology that can revolutionize the performance of products in many
areas, from consumer electronics and personal computers to automotive, medical, military and
space.
It also has the potential to create new sectors in the semiconductor industry and enable entirely
new products not yet envisaged.
• STT-RAM has key initial markets replacing embedded technologies such as SRAM, Flash
and DRAM, and providing new functionality at 65 nm and beyond. In automotive
applications, it has higher speed and lower power than Flash and is denser than SRAM.
• In portable and handset applications, it can eliminate Multi-Chip Packages (MCPs),
provide a unified memory subsystem, and reduce system power consumption for extended
battery life. In personal computers, it can replace SRAM for high-speed cache, Flash for
non-volatile cache, and PSRAM and DRAM for high-speed program execution.
6.4. STT-RAM in AVIONICS and MILITARY Application:
We have investigated two non-volatile logic circuits based on Spin-MTJ for Field
Programmable Gate Array (FPGA) and System On Chip (SOC). The first one is a Spin-MTJ based
non-volatile Look Up Table (Spin-LUT) (Figure 6.4.1). Working as a programmable and
reconfigurable logic function generator, it memorizes all the configuration data in the MTJ arrays;
and thus allows the FPGA logic circuit to reduce significantly its start-up latency from some
microseconds down to some nanoseconds. It also allows realizing a sub-ns dynamical
reconfiguration of the LUT during the signal processing. Spin-LUT non-volatile logic circuit has
great potential in the field of complex logic digital system such as high performance game console
and radar signal processing.
The second one is Spin-MTJ based non-volatile Flip-Flop (Spin-FF) (Figure 6.4.2), data
register and synchronizer, it stores permanently all the intermediate data processed in the FPGA or
SOC circuit, thereby improves the data security and enable the complex logic system to restart
immediately. Spin-FF could be advantageously used in the field of aviation and space where the
security of data is one of the most important considerations. The lower critical current of STT
writing approach makes these logic circuits work in less power and occupy smaller chip surface
than with other writing techniques. Therefore the reduction of critical current has a strong impact
on the performance of these Spin-MTJ based non-volatile logic circuits. Another advantage of
Spin-MTJ technology is that the storage element does not take much die area, because it is
processed over the chip surface (see Figure 6.4.3). By using STMicroelectronics 90nm CMOS
technology and a behavior Spin-MTJ simulation Model, Spin-LUT and Spin-FF have been
demonstrated that they could work with high speed performance and small layout surface.
Figure 6.4.1: Input Spin-LUT architecture
Figure 6.4.2: The full schematic of Spin-MTJ based Non-Volatile Flip-Flop (Spin
-FF)
6.5. IC DESIGNERS BENEFITS:
STT-RAM is a more appropriate technology for future MRAM produced using ultra-fine
processes and can be efficiently embedded in subsequent generations of such semiconductor
devices as FPGAs, microprocessors, microcontrollers and Systems-On-Chip (SOC). A special
bonus for embedded designers is the fact that the internal voltage STT-RAM requires is only 1.2
Volts.Hence, it can operate with a single 1.5 Volt battery, whereas DRAM and flash require
charge pumps to supply higher voltages. Existing NAND flash technology requires the internal
voltage to be raised to 10 to 12 volts for write operations. That voltage is boosted with the help of
a charge pump, which requires considerable power and presents adverse design conditions for
embedded designers.
Another major benefit STT-RAM technology hands embedded designers is low writing
current on the order of 100 to 200 microamperes at the 90 nm node, thanks to its efficient spin-
transfer torque techniques. At the 45 nm semiconductor node and beyond, writing current
continues to scale down significantly below 100 micro amps. This lower current translates to a
denser, less expensive memory.
Figure 6.4.4: Full layout (5.65umx10.15um) of Spin-FF, MTJs are placed above the two
points ML and MR, see also Figure 14
SUMMARY OF WORK
Finally, STT-RAM combines the capacity and cost benefits of DRAM, the fast read and
write performance of SRAM, and the non-volatility of Flash, coupled with essentially unlimited
endurance. Its performance exceeds that of other prospective non-volatile memory technologies,
and it solves the key drawbacks of first-generation, field-switched MRAM. It has excellent write
selectivity, excellent scalability beyond the 45 nm technology node, low power consumption, and
a simpler architecture and manufacturing process than first-generation MRAM.
FUTURE SCOPE OF WORK• Continuous exploration of VOx-MTJ (Vortex Magnetic Tunnel Junction); focusing on
barrier growth and interface quality.
• Re-visit of AlOx-MTJs to better understand the dependence of interface and barrier
quality on TMR.• New barrier materials exploration: Oxides like TiOx, TaOx, Nitrides like BN, etc.
• Further optimization of lithographic patterning process.
REFERENCE
[1] . http://en.wikipedia.org/wiki/Spin_(physics)
[2] . http://en.wikipedia.org/wiki/File:MRAM-Cell-Simplified.svg
[3] . http://en.wikipedia.org/wiki/Spin_Torque_Transfer
[4] . http://en.wikipedia.org/wiki/File:MRAM-Cell-Simplified.svg
[5] . http://www.bama.ua.edu/~tmewes/Java/Reversal/reversal.shtml
[6] . http://www.iop.org/EJ/abstract/0953-8984/19/16/165209.
[7] . http://arxiv.org/abs/cond-mat/0009034.
[8] . http://arxiv.org/abs/cond-mat/0607362.
[9] . http://www.nature.com/nmat/journal/v3/n6/abs/nmat1120.html.
[10] .
http://www.automotivedesignline.com/showArticle.jhtml;jsessionid=CEHFFSV3FFZX4
QSNDLOSKHSCJUNN2JVN?articleID=197008084&queryText=stt-ram.
[11] . http://www.future-fab.com/documents.asp?d_ID=4210.
http://www.freescale.com/webapp/sps/site/overview.jsp?nodeId=015424[12] . http://www.gigascale.org/pubs/1479.html.
[13] .
w w w . g r a n d i s i n c . c o m / t e c h n o l o g y / i n d e x . h t m l h t t p s : / / w w w . f b o . g o v / i n d e x ?
s=opportunity&mode=form&id=ddf20ec1bb1cf832bbfc7266cf50cf2e&tab=core&_cview=1&cc
k=1&au=&ck=
[14] .http://www.dgmarket.com/eproc/np-notice.do~2311168
[15] .http://www.darpa.mil/MTO/Programs/sttram/index.html
[16] AVIATION WEEK JANUARY 2009, TOPIC: SPIN STARTS NOW, Pg 78, (Defense
Technology International magazine).