It is truly amazing how many companies are investing in new memory technologies. Some of these technologies were recently introduced, while others have been around for awhile, but may not be well known.
Magnetoresistive random access memory (MRAM) is a nonvolatile memory that's been in development for over 15 years, but only recently has come onto the market with Freescale Semiconductor's 4-Mbit offering.
The MRAM architecture (Fig. 1) is an application of spintronics that combines magnetic-tunnel-junction (MTJ) and CMOS technologies. The MTJ is composed of a fixed magnetic layer, a thin dielectric tunnel barrier, and a free magnetic layer. When applying a current bias to the MTJ, electrons start moving and become spin-polarized by the magnetic layer as they traverse the dielectric. If the magnetic vectors are parallel on both layers, a low resistance is detected and the result is a zero; otherwise, a high resistance is detected and the result is a one.
Freescale's 4-Mbit MRAM offers 35-ns symmetrical read/write with a data retention greater than 10 years, making it a solid alternative to battery-backed SRAM.
MirrorBit NOR flash technology, devised by Spansion, is more advanced than other flash memories, such as multi-level cell (two or more bits per cell). The MirrorBit cell doubles the intrinsic density of a flash memory array by storing two physically distinct bits on opposite sides of a memory cell. Each bit within a cell serves as a binary unit of data (zero or one) that's mapped directly to the memory array. MirrorBit also provides the capability to independently read or write each side of a memory cell.
Because of its symmetrical memory cell and non-conductive storage element, MirrorBit forms a simple, efficient memory array (Fig. 2). This array design simplifies device topography and manufacturing.
Zero capacitor RAM (ZRAM) by Innovative Silicon may soon replace standard six-transistor SRAM cells used in cache memory. Because ZRAM requires only a single transistor, much higher densities can be achieved. It's also denser than conventional DRAM, which uses one transistor and one capacitor per bit.
Z-RAM relies on what's commonly referred to as the ?floating body? effect, which was first encountered in CPU design based on the silicon-on-insulator (SOI) process. This effect causes capacitance to form between the transistor and the underlying insulating substrate. Though this posed a problem in its original CPU application, the same effect does allow a DRAM-like cell to be built using only a single transistor. The floating-body effect takes the place of the conventional capacitor required in DRAM. Thus, any application using SOI technology is ideally suited for ZRAM as an embedded drop-in technology.
Phase Change Memory (PCM or PRAM)
Phase-changing technology isn't new?it was patented around 40 years ago and has been used for the past decade in CD and DVD applications. Here, the optical properties of a chalcogenide material are modified using a laser. However, phase changing at the nano scale to create NVMs is relatively new and being heavily invested in by large companies such as Infineon and IBM.
Most PCM prototypes employ a chalcogenide glass material made up of germanium, antimony, and tellurium (Ge, Sb, Te) that's altered using heat generated by electricity to change the resistive (instead of optical) properties (Fig. 3). When the chalcogenide material is in a crystalline state, it represents a binary zero; in an amorphous high-resistance state, it represents a binary one.
Macronix found a way to manufacture PCM using only a self-aligned chalcogenide cell sandwiched between address lines. The cell doesn't require a diode or transistor to indicate the on/off state. In this manner, the chalcogenide serves as the rectifying element so that the low-resistance crystalline state is never used. Instead, the cell is manipulated between distinct amorphous states. This type of cell is very low cost, since it only requires two masking steps.
PCM technology should hit the market within two to four years. According to Dr. Stefan Lai, Intel's VP of Technology and Manufacturing Group, PCM is the ?leading candidate for flash replacement.? Yet we will not see a ?credible technology challenger to mainstream, volume Flash \[until\] 2010.?
Super Permanent Memory (XPM)XPM, offered by Kilopass Technology, is a relatively new one-time-programmable (OTP) NVM. It's built entirely on standard ?vanilla? CMOS using 90-nm process technology, with paths to 65 and 45 nm. No additional mask steps or process modifications are required with XPM. This electrically field-programmable technology offers high density, performance, and reliability.
The embedded XPM technology can be delivered in a wide range of configurations, from 16 bits to very-high bit-count blocks greater than 1 Mbyte, making it practical for a wide range of embedded NVM applications.
The XPM solution isn't hampered by the geometric limitations inherent in other NVM technologies, such as a minimum tunnel oxide thickness requirement for reliable operation. The anticipated data-retention life of an XPM IP block is on the order of hundreds of years and billions of read operations.This suits XPM for applications that require a cost-effective NVM using leading-edge process technologies. XPM is ideal for permanently storing security or encryption data, personal identification information, and other applications where reliable permanent data storage is essential.