This action might not be possible to undo. Are you sure you want to continue?
Storage systems are inevitable for modern day computing. All known computing platforms ranging from handheld devices to large super computers use storage systems for storing data temporarily or permanently. Beginning from punch card which stores a few bytes of data, storage systems have reached to multi Terabytes of capacities in comparatively less space and power consumption. Computer data storage often called storage or memory refer to computer components, devices and recording media that retain digital data used for computing for some interval of time.
There are a large number of storage systems emerging which have advantages such as faster storage of more data, powerful, high speed network and performance. These storage systems have a wide range of applications which include computers and laptops , embedded memory high speed networks ,security holograms and sensors.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES CHAPTER 1 INTRODUCTION TO STORAGE SYSTEMS Storage systems are inevitable for modern day computing. All known computing platforms ranging from handheld devices to large super computers use storage systems for storing data temporarily or permanently. Beginning from punch card which stores a few bytes of data, storage systems have reached to multi Terabytes of capacities in comparatively less space and power consumption. Storage Definition Here are a few definitions of storage when refers to computers. A device capable of storing data. The term usually refers to mass storage devices, such as disk and tape drives. In a computer, storage is the place where data is held in an electromagnetic or optical form for access by a computer processor. Computer data storage; often called storage or memory refer to computer components, devices and recording media that retain digital data used for computing for some interval of time.Of these, I like the definition coined out by wikipedia.com. Likes and dislikes apart, in basic terms, computer storage can be defined as "device or media stores data for later retrieval". From the definition, we can see that the storage device possess two features namely "storage" and "retrieval". A storage facility without retrieval options seems to be of no use. A storage device may store application programs, Databases, Media files etc.... As we see in modern day computers, storage devices can be found in many forms. Storage devices can be classified based on many criterions. Of them, the very basic is as we learned in schools ie; Primary storage and Secondary storage. Storage devices can be further classified based on the memory technology that they use, based on its data volatility etc...
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES
CHAPTER 2 CLASSIFICATION OF MEMORY
The following list gives a few classifications of memory devices. - Primary and Secondary and Tertiary Storage - Volatile and non-volatile storage - Read only and Writable storage - Random Access and Sequential Access storage - Magnetic storage - Optical storage - Semiconductor storage Primary and Secondary and Tertiary Storage In simple words, primary storage is the storage device that is directly connected to the CPU and store data temporarily during execution. i.e. CPU can directly access primary storage and stores instruction and data for execution/processing. The most popular example of this kind of memory is the RAM (Random Access Memory) that we use in modern day computers. CPU registers, Caches and other memories connected to the CPU local bus falls in this category. Primary storage devices are comparatively faster than all other kinds of memory types. Usually primary storage devices are considered to be directly connected to the processor. But in reality, modern computers employ components like Virtual Memory Manager, DRAM controllers etc.. in between processor and the memory but the notion of 'Direct connection' is still valid since these components are transparent to the processor . Volatile memories are usually used as primary storage. The picture below shows a RAM module.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES RAM Module, example for Primary Storage device On the contrary, Secondary storage may not be directly accessible by the processor. And is usually used for more permenent storage of data. This requires secondary storage devices to be non-volatile. Secondary storage devices are connected to storage controllers and the CPU is required to talk to the controllers in order to access information from secondary devices. The most popular example of secondary device is the Hard disk. CD ROM, DVD ROM, USB mass storage devices, Floppy etc.. are also falls in this category. Secondary storage devices are also called Mass Storage Devices since the capacity of these devices are comparatively large. In contrast to Primary and Secondar storage, Tertiary storage may not be directly connected to the CPU or the computer itself. Tertiary storage mechanisms usually used for storage of large volume of data such as backups etc.. A commonly used mechanism for Tertiary storage involves a large number of removable mass storage devices stacked as a library and a robotic arm picks up the required media and loads in to a media reader. Once the required information is read, the media is placed back in the library. The mechanical delays associated with Tertiary Storage makes it the slowest of all storage types but largest in capacity. Volatile and non-volatile storage As the name implies, volatile memory looses its contents when power supply is withdrawn. So usually Volatile memories are used for temporary storage of data. In some exceptional cases, volatile memory devices are used along with long life batteries to make semi-permenent storage devices. Compared with non-volatile storage, Volatile storage devices are faster while both reading and writing data. This makes these kinds of memories very suitable to be used as main memories of computers. In fact, the memories we use in computers (RAM) are volatile devices. Non-volatile storage devices retain the contents even in absence of active power source. This makes non-volatile devices suitable for long term permenent data storage. Non-volatile devices usually available in large capacities. Hard Disks, CD ROM, Floppy disks, Flash, ROM etc.. are examples of non-volatile memory devices. Non-volatile storage devices are slower when compared to volatile storage devices. But some non-volatile can faster during read operation and slower during write operation. Semiconductor non-volatile memory devices fall in this category.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Read only and Writable storage Read only storage devices only allows contents to be read from and doesn't allow the contents to be modified. Meanwhile, Writable storage devices allow both content retrieval as well as content modification. Read only devices are usually used for long term permenent storage where modification of data is not necessary. CD ROM, DVD etc are examples of Read Only Storage devices. Some Read Only Storage devices comes with factory programmed data which you can only read but not modify. There is another class of devices called Write Once Read Multiple (WORM) devices which allows us to write data to it one and only one time and allows any number of subsequent reads. CD-R and DVD-R are technically comes under this category. Random Access and Sequential Access storage Random Access storage devices allow retrieval of content from any location in the same amount of time. i.e. Latency (The time taken to access a particular location in storage) is independent of content's location. RAM used in computers is an example of Random Access Memory. In sequential access storage devices, data can be accessed in sequential manner only. And the time taken to read from a particular location will be dependant on the location last accessed. Example of a sequential access device is the Tape storage device. Sequential access devices are usually used for backup purpose only, where frequent access of information is not required. Magnetic storage Magnetic storage devices store information in the form of magnetic field on magnetically coated surface. Magnetic storage devices fall in the category of non-volatile devices. This makes magnetic storage devices to be useful for long term data storage. Hard disks, Floppy disks and tape devices are examples of magnetic storage devices. Data is written to magnetic media with the help of electromagnetic heads. The same head is usually used also for retrieval of data. Though magnetic storage medias can hold large amount of data, these are considered to be bulky and not usable for mobile applications. But technological advancements made it possible to create large capacity magnetic storage devices with small size. An example for this is Apple's ipod classic which is available with a built in hard disk with a capacity as large as 160 GB for music/video storage.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Optical storage Optical storage devices store data on reflective polycarbonate discs in the form of pits and bumps. Data is recorded on the disc by pointing modulated laser beam on to the rotating disc. This makes a series of tiny pits which doesn't reflect light and bumps that reflect light. For reading the data, a low power laser beam is focused to the track and the reflected beam is directed to a photo diode. The photo diode detects the presence of pits and bumps from the reflected laser beam and convert it in to bits and bytes of information. Semiconductor storage Semiconductor storage devices store data in tiny memory cells made of very small transistors and capacitors made of semiconductor materials such as silicon. Each cell can hold one bit of information and an array of cells stores large chunk of information. Semiconductor storage devices are volatile.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES CHAPTER 3 CHALLENGES IN MEMORY TECHNOLOGY PC technologies continue to evolve as performance demands grow, and system memory technology is no exception. Throughout the history of the PC, memory technology has steadily progressed in capacity and performance to meet the increasing requirements of other PC hardware subsystems and software. In the past, there have been relatively clear industry transitions from one memory technology to its successor. However, today there are multiple choices² PC133, RDRAM, and Double Data Rate (DDR)² and more choices may exist in the future as providers of DRAM system memory accommodate a growing variety of platforms and form factors. These range from small handheld devices to high-end servers, each with different power, space, speed, and capacity requirements for system memory. This seminar focuses on PC system memory issues and trends. It begins by reviewing the role of memory in the system and the key memory parameters that affect system performance. This report then presents the basics of memory technology and today's alternatives. Finally says about the key upcoming transitions in the DDR and Rambus memory technologies. Role of memory in a system The primary role of memory is to store code and data for the processor. Although caching and other processor architecture features have reduced its dependency on memory performance, the processor still requires most of the memory bandwidth. Figure 1 shows the major consumers of memory bandwidth: the processor, graphics subsystem, PCI devices (such as high-speed communications devices), and hard drives. Other lower bandwidth interfaces such as the USB and parallel ports must also be accommodated. The memory hub provides an interface to system memory for all of the high bandwidth devices. The I/O hub schedules requests from other devices into the memory hub. Memory plays a key role in the efficient operation of I/O devices such as graphics adapters and disk drives. In a typical system, most data transfers move through system memory. For example, when transferring a file from the network to a local disk, the PCI host adapter transfers data from the network to memory. This is commonly referred to as direct memory access (DMA), as opposed to programmed I/O (PIO), in which the processor is directly involved in all data transfers. The processor, after performing any required formatting
2011 DR.AIT, ECE Page 7
NEXT GENERATION MEMORY TECHNOLOGIES operations, initiates a transfer from memory to local disk storage. Once initiated, the data is transferred directly from memory to disk without any further processor involvement. In summary, the system memory functions as the primary storage component for processor code and data, and as a centralized transfer point for most data movement in today's system. Performance Factors Memory parameters that impact system performance are capacity, bandwidth, and latency. Capacity How does capacity impact system performance? The first step in answering this question is to describe the memory hierarchy. Table 1 shows the capacities and speeds of various storage mechanisms found in a typical mainstream desktop computer in service today. These storage mechanisms range from the very fast, but low-capacity, Level 1 (L1) cache memory to the much slower, but higher-capacity, disk drive. Ideally, a computer would use the fastest available storage mechanisms²in this case L1 cache²for all data. However, the laws of physics (which dictate that higher capacity storage mechanisms are slower) and cost considerations prevent this. Instead, PCs use a mechanism called ³virtual memory,´ which makes use of the L1 and L2 cache, main system memory, and the hard drive. The virtual memory mechanism allows a programmer to use more memory than is physically available in the system, and to keep the most frequently and recently used data in the fastest storage. When more memory is needed than is available in system memory, some data or code must be stored on disk. When the processor accesses data not available in memory, information that has not been accessed recently is saved to the hard drive. The system then uses the vacated memory space to complete the processor's request. However, disk access is comparatively slow and system performance is significantly impacted if the processor must frequently wait for disk access. Adding system memory reduces this probability. The amount of capacity required to reduce disk activity to an acceptable level depends on the operating system and the type and number of active applications, including background tasks. Semiconductor technology has provided capacity improvements consistent with Moore's law²greater than a 1.4x compound annual growth rate²and the outlook is for similar increases over the next few years. These increases have exceeded the requirements of mainstream desktop PCs and, as a result, the number of
2011 DR.AIT, ECE Page 8
NEXT GENERATION MEMORY TECHNOLOGIES memory slots is being reduced from three to two in many of today's client platforms. However, servers and high-end workstations continue to take advantage of the capacity increases.
Bandwidth Memory bandwidth is a measure of the rate at which data can be transferred to and from memory, typically expressed in megabytes per second (MB/sec). Peak bandwidth is the theoretical maximum transfer rate between any device and memory. In practice, peak bandwidth is reduced by interference from other devices and by the ³lead-off´ time required for a device to receive the first bit of data after initiating a memory request. There should be adequate memory bandwidth to support the actual data rates of the highest-speed devices and to provide enough headroom to prevent significant interference between devices. In many systems, memory and I/O hubs are designed to accommodate peak requirements by buffering transfers and scheduling conflicting memory requests. Table 2 shows the data rates of various system components over the last 4 years. Although the need for memory bandwidth is not directly proportional to these data rates, the upward trend is obvious. Memory systems have done a fairly good job of keep with system requirements over this period of time, moving from 533 MB/sec to 2133 MB/sec. Dual-memory interfaces using Rambus or DDR memory boost bandwidth to 3200 MB/sec.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Latency Latency is a measure of the delay from the data request until the data is returned. It is a function of peak bandwidth, lead-off time, and interference between devices. In general, processors are more sensitive to latency than bandwidth because they work with smaller blocks of data and can waste a significant number of clocks waiting for critical data. In contrast, I/O data transfers are relatively long, and bandwidth is a more important consideration than latency. Data transfers moving to and from system memory must pass through the memory hub and, in many cases, the I/O hub. These components are collectively referred to as the chip set or core logic, and are major contributors to the latency from a device source.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES CHAPTER 4 MEMORY TECHNOLOGIES Holography Holographic data storage refers specifically to the use of holography to store and retrieve digital data. To do this, digital data must be imposed onto an optical wave front, stored holographically with high volumetric density, and then extracted from the retrieved optical wav front with excellent data fidelity. A hologram preserves both the phase and amplitude of an optical wave front of interest called the object beam by recording the optical interference pattern between it and a second coherent optical beam the reference beam . The reference beam is designed to be simple to reproduce at a later stage (A common reference beam is a plane wave a light beam that propagates without converging or diverging). These interference fringes are recorded if the two beams have been overlapped within a suitable photosensitive media, such as a photopolymer or inorganic crystal or photographic film. The bright and dark variations of the interference pattern create chemical and/or physical changes in the media, preserving a replica of the interference pattern as a change in absorption, refractive index or thickness. When the recording is illuminated by a readout beam similar to the original reference beam, some of the light is diffracted to ³reconstruct´ a copy of the object beam. If the object beam originally came from a 3-D object, then the reconstructed hologram makes the 3-D object reappear. Collinear Holography HVD uses a technology called µcollinear holography¶, in which two laser rays, one blue-green and one red, are collimated into a single beam. The role of the blue-greenlaser is to read the data encoded in the form of laser interference fringes from the holographic layer on the top, while the red laser serves the purpose of a reference beam and also to read the servo info from the aluminum layer ± like in normal CDs ± near the bottom of the disk. The servo info is meant to monitor the coordinates of theread head above the disk (this is similar to the track, head and sector information on a normal hard disk drive).
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES
Recording Data A simplified HVD system consists of the following main components: Blue or green laser (532-nm wavelength in the test system) Beam splitter/merger Mirrors Spatial light modulator (SLM) CMOS sensor Polymer recording medium. The process of writing information onto an HVD begins with encoding the information into binary data to be stored in the SLM. These data are turned into ones and zeroes represented as opaque or translucent areas on a "page" -- this page is the image that the information beam is going to pass through. When the blue-green argon laser is fired, a beam splitter creates two beams. One beam, called the object or signal beam, will go straight, bounce off one mirror andtravel through a spatial-light modulator (SLM). An SLM is a liquid crystal display(LCD) that shows pages of raw binary data as clear and dark boxes. The information from the page of binary
2011 DR.AIT, ECE Page 12
NEXT GENERATION MEMORY TECHNOLOGIES code is carried by the signal beam around to the light-sensitive lithium-niobate crystal. Some systems use a photopolymer in place of the crystal. A second beam, called the reference beam, shoots out the side of the beam splitter and takes a separate path to the crystal
fig 4.1 RECORDING DATA Reading Data To read the data from an HVD, you need to retrieve the light pattern stored in the hologram. In the HVD read system, the laser projects a light beam onto the hologram ± a light beam -- a light beam that is identical to the reference beam. An advantage of a holographic memory system is that an entire page of data can be retrieved quickly and at one time. In order to retrieve and reconstruct the holographic page of data stored in the crystal, the reference beam is shined into the crystal at exactly the same angle at which it entered to store that page of data. Each page of
2011 DR.AIT, ECE Page 13
NEXT GENERATION MEMORY TECHNOLOGIES data is stored in a different area of the crystal, based on the angle at which the reference beam strikes it. The key component of any holographic data storage system is the angle at which the reference beam is fired at the crystal to retrieve a page of data. It must match the original reference beam angle exactly. A difference of just a thousandth of a millimeter will result in failure to retrieve that page of data. During reconstruction, the beam will be diffracted by the crystal to allow there creation of the original page that was stored. This reconstructed page is then projected onto the CMOS, which interprets and forwards the digital information to a computer. MAGNETIC MEMORY All materials have an inherent magnetic character arising from the movements of their electrons. Since dynamic electric fields induce a magnetic field, the orbit of electrons, which creates atomic current loops, results in magnetic fields. An external magnetic field will cause these atomic magnetic fields to align so that they oppose the external field. This is the only magnetic effect that arises from electron pairs. If a material exhibits only this effect in an applied field it is known as a diamagnetic material. Magnetic properties other than diamagnetism, which is present in all substances, arise from the interactions of unpaired electrons. These properties are traditionally found in transition metals, lanthanides, and their compounds due to the unpaired d and f electrons on the metal. There are three general types of magnetic behaviors: paramagnetism, in which the unpaired electrons are randomly arranged, ferromagnetism, in which the unpaired electrons are all aligned, and antiferromagnetism, in which the unpaired electrons line up opposite of one another. Ferromagnetic materials have an overall magnetic moment, whereas anti ferromagnetic materials have a magnetic moment of zero. A compound is defined as being ferrimagnetic if the electron spins are orientated antiparrallel to one another but, due to an inequality in the number of spins in each orientation, there exists an overall magnetic moment. There are also enforced ferromagnetic substances (called spin-glass-like) in which antiferromagnetic materials have pockets of aligned spins. Semiconductor spintronics devices combine advantages of semiconductor with the concept of magnetoelectronics. This category of devices includes spin diodes, spin filter, and spin FET. To make semiconductor based spintronic devices, researchers need to address several following
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES different problems. A first problem is creation of inhomogeneous spin distribution. It is called spin-polarisation or spin injection. Spin-polarised current is the primary requirement to make semiconductor spintronics based devices. It is also very fragile state. Therefore, the second problem is achieving transport of spin-polarised electrons maintaining their spin-orientation . Final problem, related to application, is relaxation time. This problem is even more important for the last category devices . Spin comes to equilibrium by the phenomenon called spin relaxation. It is important to create long relaxation time for effective spin manipulation, which will allow additional spin degree of freedom to spintronics devices with the electron charge . Utilizing spin degree of freedom alone or add it to mainstream electronics will significantly improve the performance with higher capabilities. The third category devices are being considered for building quantum computers. Quantum information processing and quantum computation is the most ambitious goal of spintronics research. The spins of electrons and nuclei are the perfect candidates for quantum bits or qubits. Therefore, electron spin and nuclear based hardwares are some of the main candidates being considered for quantum computers. Spintronics based devices offers several advantages over conventional charge based devices. Since magnetized materials maintain their spin even without power, spintronics based devices could be the basis of non-volatile memory device. Energy efficiency is another virtue of these devices as spin can be manipulated by low-power external magnetic field. Miniaturization is also another advantage because spintronics can be coupled with conventional semiconductor and optoelectronic devices. However, temperature is still a major bottleneck. Practical application of spintronics needs room-temperature ferromagnet in semiconductors. Making such materials represents a substantial challenge for materials scientists.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Spin based Devices The present status of spintronics devices at the commercial level is limited to giant magnetoresistance (GMR) based devices. In GMR based memory devises electron spin play passive role . It is limited to detect the change of magnitude of resistance depending on direction of the spin . The change in resistance is controlled by a local or an external magnetic field . But, it is predicted that spintronics can go beyond this passive spin device by integrating electron spin into conventional semiconductors. Thus, the technology based on spintronics may replace conventional semi-conducting devices by introducing active control of electron spin. Spin transistors The spin-transistors exploit electron spin either by spin-valve effect or by active control of electron spin . The design of transistor is similar to that of GMR devices. It consists of three layers, out of which the non-magnetic layer is sandwiched between the two ferromagnetic layers Johnson was the first to propose about spin-valve transistor. As per him, the first magnetic layer acts as a spin injector or emitter while the second acts as a spin detector or collector . The nonmagnetic layer acts as a base . The magnetization direction of the collector can be changed by the application of an external magnetic field . When the voltage is applied across the emitter-base, it generate electrons with either spin-up or spin-down . When the magnetization direction of emitter and collector is parallel, the current can flow throw the base to the collector . The electrons face high resistance when the relative magnetization direction is opposite. Thus, device acts as one-way switch . Electron spin plays passive role in Johnson¶s spin-valve transistor.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES
Figure 3 Dutta-Das field effect transistor; at zero gate voltage, electron preserves spin state in transport channel (a) it enables current flow from source to drain. With applied gate voltage, electrons change their spin state from parallel to anti parallel to the direction of magnetization of ferromagnetic layer (b) this offers high resistance to flow of current. Therefore, electron scattering occurs at drain and no current flow from source to drain . The first model of transistor using active control of electron spin was proposed by Datta and Das. In the Datta-Das field effect transistor, the non-magnetic layer acts as a gate while two ferromagnetic layers act as source and drain respectively (fig 2(a)) . The gate plays an important role in Datta-Das field effect transistor. The gate modifies electron spin by generating effective magnetic field and thereby in switching the transistor . When voltage is applied to the gate, it generates effective magnetic field (fig 2(b)). Thus, by modifying gate voltage one can modify electron spin . The electrons ballistically transport in transport channel, if its spin is parallel to the magnetization direction of drain (spin detector) . Otherwise, it will scattered away . The control of charge current in spin transistor is similar to the conventional transistors [2, 4]. But, the spin transistors possess advantage over conventional transistors. They are smaller in size, and consume less power . Still, the spin-transistors are exist is in prototype models because of theoretical limitation related to spin behavior in different materials.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Manipulation of Electron Spin Spintronic devices are based on careful manipulation of the electron spin. The spin can be easily manipulated by applying external magnetic field or by shining polarized light . In general, the scheme of spin manipulation works fundamentally on: (1) generation of spin-polarized electron, (2) injection and transportation of the spin-polarized electron, and (3) detection of the spinpolarized carriers with information. Generation of spin polarization The generations of spin-polarized electron spins mean generation of spin polarized current. This spin polarized current carries non-equilibrium spin population. The Spintronics devices detect the distribution of spin-up and spin-down electrons in spin polarized current to control the current . This phenomenon of controlling current in spintronic device makes it suitable to act as electronic switch of transistor. Thus, the control of current is then either a control of phase of electron spin or spin-population. It can be generated by transport, optical, and resonance methods or by their combination . Figure 2 shows the schematic representation of generation of spinpolarized current by transport method. Spin injection and spin-polarized transport The spintronic device requires efficient transport of generated non-equilibrium spin (spinpolarized current) across the electrode/sample interface. The transport of non-equilibrium spin across interface is called spin injection. The non-equilibrium spin can be injected by driving ordinary current through ferromagnetic electrode to sample. The current can be driven in plane plan of interface called 'current in plane (CIP) geometry' (fig 4(a)) or perpendicular to the interface called 'current perpendicular to plane (CIP) geometry' (fig 4(b)). The spin can be also injected by optical method. The efficiency of spin injection is determined by rate of accumulation of non-equilibrium spin in sample. There are several proposed ways to transport spin-polarized current across interface. These are: (1) formation of Ohmic contact between electrode-sample interface, (2) Ballistic electron injection, (3) electron tunneling from space charge region and, (4) Hot spin injection.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Ohmic injection The most basic approach to spin injection is through the perfect Ohmic contact between ferromagnetic/non-magnetic (F/N) interfaces (fig 4 (a)) . The interface can be produce by taking metals or semiconductors or superconductors as non-magnetic region with ferromeget. The degree of spin injection in non-magnetic region depends on the ratio of the conductivities of ferromagnetic region (F) and non-magnetic region (N) . For typical conductivity mismatch, when conductivity of F region N region, higher the spin injection efficiency (fig 4(b) and (c)). When conductivity of F region non-magnetic region, smaller the spin injection efficiency. This phenomenon is called ³conductivity mismatch´ . In the case of ferromagnet/semiconductor interface, Ohmic contacts resulted from the doping of semiconductor surface. However, doping leads to loss of spin polarization by spin-flip scattering . The electrochemical potential of N region increases with spin injection. The difference of spin dependent electrochemical potentials generates effective resistance R on either side of F/N interface. In superconductor/F interface, increase in total resistance with spin injection results in switching superconducting state to normal state of much higher resistance. Ballistic electron injection The ballistic spin injection works on principle of GMR effect and electrons are dynamically transported. The ballistic transport is favorable in ferromagnet/non-magnet/ferromagnet (F/N/F) interfaces. The F/N/F interface is formed by sandwiching a non- ferromagnetic layer of finite thickness between two finite ferromagnetic layers . Fully ballistic transport takes place when ferromagnetic layers aligned in the same direction . This condition provides low resistance path to the spin polarize current. The probability of spin polarized electron back flow or reflection is less in ballistic transport, once it enters in the non-magnetic region. The transmission probability of ballistic transport depends on difference of two spin conduction sub bands of the ferromagnet and the conduction band of the semiconductor . Spin detection Spin detection typically depends on the sensing the changes in the signal due to spin injection . The injection of non-equilibrium spin either induces voltage or changes resistance corresponding to buildup of the non-equilibrium spin . This voltage can be measured in terms of change in
2011 DR.AIT, ECE Page 19
NEXT GENERATION MEMORY TECHNOLOGIES resistance by potentiometric method; while change in resistance can be measured in terms of voltage by balancing Wheastone Bridge . The transport and optical methods of spin detection are most widely adopted to detect spin. The efficiency of spin detection in transport method is depends on interface properties. Therefore, spin detection is low and also suffered from difficulty discuss above . The optical spin detection technique is well established. The spin can be detected by determining the helicity of emitted light from LEDs connected with interface. Spin Relaxation Non-equilibrium spin accumulates in non-magnetic region due to process of spin injection. It comes to equilibrium by the phenomenon called spin relaxation . The rate of accumulation of non-equilibrium spin depends on the spin relaxation . Electrons can remember their spin state for finite period of time before relaxing. That finite time period is called µSpin lifetime¶ . Longer lifetime is more desirable for data communication application while shorter for fast switching . The distance traveled by the electron without loosing spin state is called µSpin diffusion length¶ . It is most important variable in spintronic devices, which determines maximum allowable thickness of the non-magnetic region in device. It is also depend on spin lifetime . There are four proposed ways by which conduction electrons of metals and semiconductors relax: (A) The Elliott-Yafet mechanism, (B) The D¶yakonov-Perel¶ mechanism, (C) The Bir-Aronov-Pikus mechanism, and (D) hyperfine-interaction . Spin Relaxation Non-equilibrium spin accumulates in non-magnetic region due to process of spin injection. It comes to equilibrium by the phenomenon called spin relaxation . The rate of accumulation of non-equilibrium spin depends on the spin relaxation . Electrons can remember their spin state for finite period of time before relaxing. That finite time period is called µSpin lifetime¶ . Longer lifetime is more desirable for data communication application while shorter for fast switching . The distance traveled by the electron without losing spin state is called µSpin diffusion length¶ . It is most important variable in spintronic devices, which determines maximum allowable thickness of the non-magnetic region in device. It is also depend on spin lifetime . There are four proposed ways by which conduction electrons of metals and semiconductors relax: (A) The
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES Elliott-Yafet mechanism, (B) The D¶yakonov-Perel¶ mechanism, (C) The Bir-Aronov-Pikus mechanism, and (D) hyperfine-interaction. SPINTRONIC TECHNOLOGY is already in your computer, at least in a primordial incarnation. Modern hard disk drives have a read head that relies on an effect known as giant magnetoresistance, or GMR, which was discovered by French and German researchers in the late 1980¶s. Basically, when the spins of electrons in the read head point in the same direction as those creating the small magnetic domains on the disk, the head¶s electrical resistance decreases. When the spins are in opposite directions, the resistance increases slightly. More recently. engineers have developed even better read heads that rely on tunnel magneto- resistance, a kind of enhanced GMR. It is this ability to sense very feeble magnetic fields that has allowed hard-disk makers to keep doubling the capacities of hard- disk drives on a schedule that¶s even out- paced Moore¶s Law. Many advances in spintronics resulted from two big research programs that the U.S. Defense Advanced Research Projects Agency, or DARPA, funded in the 1990s. The first one produced the earliest MRAM prototypes. These devices used memory cells consisting of magnetic tunnel junctions: two layers of a ferromagnetic material like iron separated by an extremely thin, nonconductive barrier of magnesium oxide. When the spins of the electrons in the two ferromagnetic layers point in the same direction ,in other words, when their magnetizations are aligned,the electrical resistance across the junction decreases; when the spins point in different directions, the junction becomes more resistant to current. The prototypes used this change in resistance to sense whether a 1 or a 0 was stored. Some MRAM chips built at the time contained millions of memory cells, each with dimensions of about 150 nanometers, an impressive achievement back then. But the researchers soon discovered that going below too nm was not going to be easy. The problem had to do with the method they used to change bits, which was to drive currents through electrodes connected to each memory cell, creating a magnetic field that oriented the spin state of the cell. This method required currents that proved quite high, draining lots of power. Worse still, the magnetic fields affected not only the desired bit but also others nearby, resulting in errors. Researchers are now trying to improve on this scheme. The most promising alternative is called spin-torque- transfer, or STT. The idea is to send electrons through a magnetic layer of cobalt, which tends to orient their spins in the same direction. The resulting spin- polarized current then flows into another layer of cobalt material. There, by virtue of one of the many mysteries of quantum mechanics, the incoming spin2011 DR.AIT, ECE Page 21
NEXT GENERATION MEMORY TECHNOLOGIES polarized electrons transfer their spin orientation to the electrons on this second layer, thus magnetizing it. So instead of writing a bit by applying a magnetic field, as early MRAM designs do, STT uses a spin-polarized current of electrons. To be commercially viable, the magnetic region where the bit is stored has to be quite small, of course. Researchers believe STT should work down to at least 6 nm and possibly even smaller dimensions. Last year, engineers at Hitachi and Tohoku University demonstrated a prototype capable of storing 32 megabits this way. But that¶s not all that much. For comparison, a modern DRAM chip can hold ia8 times that amount. And though in theory such memories should require very small currents to change a bit, in practice the currents are still too high for most commercial applications. For such reasons, our group and several others are betting on a different approach entirely. Forget about currentinduced magnetic fields and spin-polarized currents, Instead, find a storage medium with a permanent magnetism that you can control by applying small voltages. These materials exist They are called dilute magnetic semiconductors. As their name suggests, they are semiconductors that are also somewhat magnetic. Their magnetism stems from certain metal atoms added in a process similar to doping. What¶s interesting about these materials is that the presence of charge carriers electrons and holes (vacancies left when electrons are missing in places where they¶d normally be found)²can alter their magnetic properties. As part of DARPA. second MRAM research program. initiated in 1999, researchers investigated several dilute magnetic semiconductors, in particular gallium manganese arsenide and indium manganese arsenide. Both proved to be good candidates, There was just one problem: A material is magnetic only up to a given temperature in this case about 200 Kelvin, or ± 73 oC. That¶s colder than nighttime in most parts of Mars! Go above that level known as the Curie temperature and atomic vibrations cause the spins to lose the orderly arrangement that makes the material a permanent magnet. If this was a memory chip, you¶d lose your data.The resulting gallium manganese nitride turned out to be very promising. When a magnetic field to this substance is applied, it becomes permanently magnetized. That is to say, when we remove the field, the magnetization doesn¶t go away, so it can be used to store data. The next major step, was the ability to manipulate the magnetic properties of this semiconductor electrically. It ws started with ordinary gallium nitride, then applied a thin layer of gallium nitride that contained a little added silicon, a dopant that donates electrons, thereby creating an n-type semiconductor. (The ³n´ stands for ³negative.´ reflecting the addition of
2011 DR.AIT, ECE Page 22
NEXT GENERATION MEMORY TECHNOLOGIES negative charges, electrons ). Next there was an addition of another gallium nitride layer, this time using magnesium as a dopant to remove electrons from the lattice of atoms, creating a ptype layer (p stands for positive). Finally, a very thin veneer of gallium manganese nitride was deposited on top of all this. The junction between n- and p-type layers was key. That¶s because we can control the concentration of electrons and holes around a p-n junction by applying a voltage across it. And that¶s exactly what should have been done. When 5 volts is applied across the p-n junction. the magnetization of that upper layer approached 0. When voltage was removed , the magnetization shot up.lt is a faint magnetization to be sure, but enough for storing bits. To know why the voltage on a p-n junction change the magnetization nearby? To understand that. you have to first think about what goes on at a p.n junction when no voltage is applied across It .First, recall that the n-type material has an abundance of negative charge carriers,electrons,which are free to move around. In the p-type material, the charge carriers are holes, spots in the atomic lattice that are lacking in electrons. When we put one of these materials against the other, electrons move from the n-type material into the p-type material, filling what were vacancies, or holes. So we end up depleting both types of charge carriers in the vicinity of the p-n junction. which is called, naturally enough, the depletion zone. This process is self limiting ,though. The loss of electrons from the n-type material leaves it with a positive charge, while the gain of electrons in the p-type material makes it negatively charged. This sets up an electric field that opposes the migration of any more electrons across the junction. As with an ordinary diode. If the p-type material is made positive with respect to the n-type material, the applied voltage can overcome this electric field, sending holes and electrons racing toward the junction, reducing the thickness of the depletion zone. A voltage of the opposite sense boosts the internal electric field and makes the depletion zone wider. What makes this device different is that the p-type material is very thin and is positioned right next to the magnetic layer of gallium manganese nitride. So by adjusting the voltage across the p-n junction, we can control the concentration of holes in the p-type layer at the interface with this magnetic material. That¶s important because the pervasive quantum-mechanical weirdness that arises at these scales allows these holes to interact with the manganese atoms sitting a few hundred angstroms away. Though there is a debate in our community, we believe the quantum
2011 DR.AIT, ECE Page 23
NEXT GENERATION MEMORY TECHNOLOGIES phenomenon at work here is what is known as carrier-mediated ferromagnetism. It¶s as though the holes told some of the electrons around these manganese atoms to align their spins and start acting like a refrigerator magnet. By the same token, a negative voltage is applied across the p-n junction, there is an increase in the width of the depletion zone enough to diminish the number of holes at the interface with the magnetic material. That then allows the spins of the electrons in these manganese atoms to revert to random directions. The device¶s magnetization vanishes. This was the first demonstration that ferromagnetism can be controlled by applying voltages to a p-n junction without relying on ultracold temperatures. Hope this discovery will help turn spintronics into a hot topic again, so to speak. The initial prototype built can¶t be readily used as a memory cell. First, we need a major improvement on our design. The problem is that although we can control the magnetization of our device using voltages, when we remove the voltages the magnetization returns to a baseline level. For a device to work as a memory, we need to be able to switch back and forth between two stable states. One idea currently considered is making the device¶s layers even thinner and adding a barrier of non magnetic material, also very thin, between the p-type and magnetic layers. There is a hope that by applying a voltage across these two layers, we can change the concentration of holes in the p-type region and also force some of the holes to cross the newly added barrier and migrate into the magnetic section of the device. The barrier would then play a key role after the voltage is removed, it would prevent the holes from migrating back to the p-type region. Now, if we take the device in this magnetized state and apply a voltage in the reverse direction, the holes would cross the barrier back into the p-type region. The holes would remain trapped there, and the magnetization would disappear. This approach would provide the two stable states we need to use the device as a memory. If this design is successful, the next step would be miniaturization fact, our initial prototype is rather big each memory cell is about the size of a fingernail. To build smaller memory cells, investigation is done in two ways: One is using conventional photolithography, which we believe could lead to cells about so nm in size. Another idea is to grow the cell structures as nanowires, which we speculate might shrink them as small as 20 nm. Such reduced dimensions would lead to another challenge: reading the bits in these tiny cells. As
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES we proceed to nanoscale dimensions, the strengths of the magnetic fields will become even smaller, How to detect them remains an open question. We might have to equip each memory cell with a tiny magnetic sensor, similar to a read head of a hard drive but etched as a series of layers in the semiconductor. It¶s a possibility, but we don¶t know how it will perform and whether the resulting device would be economically viable. NANO TECHNOLOGY Carbon Nano Tube Carbon nanotubes (CNT) are allotropes of carbon with a cylindrical nanostructure. Nanotubes have been constructed with length-to-diameter ratio of up to 132,000,000:1, which is significantly larger than any other material.CNT developed by Sumio Iijima. These cylindrical carbon molecules have novel properties. This property is useful in many applications in nanotechnology, electronics, optics and other fields of science.Nanotubes are members of the fullerene structural family. It has ability to conduct electricity aswell as copper.it is stronger than steel and as hard as diamond. The wall of a single-walled carbonnanotube is only one carbon atom thick and the tube diameter is approximately 100,000 timessmaller than a human hair. Their name is derived from their size, since the diameter of a nanotubeis on the order of a few nanometers (approximately 1/50,000th of the width of a human hair),while they can be up to several millimeters in length (as of 2008). Nanotubes are categorized as single-walled nanotubes (SWNTs) and multi-walled nanotubes (MWNTs). Technology Nantero's technology is based on a well-known effect in carbon nanotubes where crossed nanotubes on a flat surface can either be touching or slightly separated in the vertical direction(normal to the substrate) due to Van der Waal's interactions. In Nantero's technology, eachNRAM "cell" consists of a number of nanotubes suspended on insulating "lands" over a metalelectrode. At rest the nanotubes lie above the electrode "in the air", about 13 nm above it in thecurrent versions, stretched between the two lands. A small dot of gold is deposited on top of thenanotubes on one of the lands, providing an electrical connection, or terminal. A second
2011 DR.AIT, ECE Page 25
NEXT GENERATION MEMORY TECHNOLOGIES electrodelies below the surface, about 100 nm away.Normally, with the nanotubes suspended above the electrode, a small voltage applied between the terminal and upper electrode will result in no current flowing. This represents a "0" state.However if a larger voltage is applied between the two electrodes, the nanotubes will be pulled towards the upper electrode until they touch it. At this point a small voltage applied between theterminal and upper electrode will allow current to flow (nanotubes are conductors), representing a"1" state. The state can be changed by reversing the polarity of the charge applied to the two electrodes. What causes this to act as a memory is that the two positions of the nanotubes are both stable. In the off position the mechanical strain on the tubes is low, so they will naturally remain in this position and continue to read "0". When the tubes are pulled into contact with the upper electrode a new force, the tiny Van der Waals force, comes into play and attracts the tubes enough to overcome the mechanical strain. Once in this position the tubes will again happily remain there and continue to read "1". These positions are fairly resistant to outside interference like radiation that can erase or flip memory in a conventional DRAM. NRAMs are built by depositing masses of nanotubes on a pre-fabricated chip containing rows of bar-shaped electrodes with the slightly taller insulating layers between them. Tubes in the "wrong" location are then removed, and the gold terminals deposited on top. Any number of methods can be used to select a single cell for writing, for instance the second set of electrodes can be run in the opposite direction, forming a grid, or they can be selected by adding voltage to the terminals as well, meaning that only those selected cells have a total voltage high enough to cause the flip. Currently the method of removing the unwanted nanotubes makes the system impractical. The accuracy and size of the epitaxy machinery (Epitaxy refers to the method of depositing a
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES monocrystalline film on a monocrystalline substrate. The deposited film is denoted as epitaxial film or epitaxial layer. The term epitaxy comes from the Greek roots epi, meaning "above", and taxis, meaning "in ordered manner". It can be translated "to arrange upon".) is considerably "larger" that the cell size otherwise possible. Existing experimental cells have very low densities compared to existing systems; some new method of construction will have to be introduced in order to make the system practical. ROBOTIC STORAGE Large quantities of individual magnetic tapes, and optical or magneto-optical discs may be stored in robotic tertiary storage devices. In tape storage field they are known as tape libraries, and in optical storage field optical jukeboxes, or optical disk libraries per analogy. Smallest forms of either technology containing just one drive device are referred to as autoloaders or autochangers. Robotic-access storage devices may have a number of slots, each holding individual media, and usually one or more picking robots that traverse the slots and load media to built-in drives. The arrangement of the slots and picking devices affects performance. Important characteristics of such storage are possible expansion options: adding slots, modules, drives, robots. Tape libraries may have from 10 to more than 100,000 slots, and provide terabytes or petabytes of near-line information. Optical jukeboxes are somewhat smaller solutions, up to 1,000 slots. Robotic storage is used for backups, and for high-capacity archives in imaging, medical, and video industries. Hierarchical storage management is a most known archiving strategy of automatically migrating long-unused files from fast hard disk storage to libraries or jukeboxes. If the files are needed, they are retrieved back to disk.
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES ADVANTAGES AND APPLICATIONS Computer and Laptops (Enabling instant ±on performance, with no for boot up) Mobile devices (Faster storage of more data for PDA¶s and handhelds) Embedded memory (More powerful microprocessor, microcontroller, other logic device) High speed network Security hologram Holographic interferometry Holographic sensors An HVD (holographic Versatile Disc), a holographic storage media, is an advanced optical disc that¶s presently in the development stage. Polaroid scientist J. van Heerden was the first to come up with the idea for holographic three-dimensional storage media in 1960. An HVD would be a successor to today¶s Blu-ray and HD-DVD technologies. It can transfer data at the rate of 1 Gigabit per second. The technology permits over 10 kilobits of data to be written and read in parallel with a single flash. The disc will store upto 3.9 terabyte (TB) of data on a single optical disk. Holographic data storage, a potential next generation storage technology, offers both high storage density and fast readout rate. In this article, I discuss the physical origin of these attractive technology features and the components and engineering required to realize them. I conclude by describing the current state of holographic and development efforts in the context of ongoing improvement to established storage technologies. The new technology based on spintronics utilizes electron spin and charge in conventional electronics .. The potential advantage is considerable increase in capacity of conventional electronic devices . But, it suffers from fundamental limitations. The spin dynamics is not clearly understood in transport across interface . This uncertainty imposes limits on design of devices . However, in recent years, understanding of spin dynamics in metallic multilayer gives partial success in utilizing electron spin as GMR read head and data storage devices . But, the projection of spintronics will go beyond this and may end regime of charge based electronic .
2011 DR.AIT, ECE
NEXT GENERATION MEMORY TECHNOLOGIES BIBILOGRAPHY y y y y y y y y y y y y y y y Special issue on spintronics IEEE, volume 91 , no. 5, May 2003 Spintronics info.com A Discovery Company, How Stuff Works Wikipedia Whatis.techtarget.com(MRAM) Scribd.com(MRAM) Physics.umd.edu IEEE Spectrum, Nov, 2010 Psaltis, D. Mok, F. Holographic memories. Scientific American. Encyclopedia of Optical Engineering. www.ibm.com - IBM Research Press Resources Holographic Storage www.howstuffworks.com www.hvd-forum.org http://www.tech-faq.com/hvd.shtml . http://www.eweek.com/article2/0,1759,1759907,00.asp
2011 DR.AIT, ECE