All modern electronics rely on dedicated onboard memory to support their operation. The rapid advancement of technology requires memory or random access memory (RAM) to display characteristics such as very fast read and write speeds and high storage densities.
Frost & Sullivan believes that one of the most promising new technologies in this area is phase change memory (PCM or phase change random access memory [PRAM]), which is based on the distinctive behaviour of chalcogenide glass. By applying heat, chalcogenide glass can be switched between two states - crystalline and amorphous. It contains a chalcogenide element such as sulphur, selenium, or tellurium as a significant constituent. The crystalline and amorphous states of chalcogenide glass exhibit radically different electrical resistivity values, and this forms the basis of data storage in PRAMs.
PCM, although similar to flash, ferroelectric RAM (FRAM), and magnetoresistive RAM (MRAM), displays very different characteristics that distinguish its applications. By demonstrating smaller cell size, good data retention, and much better endurance to program and erase cycles, PCM looks promising enough to replace NOR flash.
Another factor in favour of PCM is its scalability compared to NOR flash, which is nearing its scalability limit. Like dynamic RAM (DRAM), PCM does not need to be 'flash' erased before programming, as the final storage state is determined by the programming condition alone; hence it is faster than NOR flash. This advantage of speed makes it an interesting technology among non-volatile memory technologies where performance is limited by memory access speed.
PCM does not pose many challenges to integrate into complementary metal oxide semiconductor (CMOS) logic processes. Although it cannot replace static RAM (SRAM) as L1 cache due to it being slower, PCM has the potential to be used as L2 and L3 cache for certain applications.
According to Rich Liu, vice president of Macronix International, there are two technical challenges for PCM to truly establish itself: the reduction of the current needed to switch the phase change storage element, and further improvement in the program/erase cycling endurance.
PCM technology has to meet the requirements of switching currents by ranging from less than 100 mA to approximately 700 mA. Trying to satisfy the upper switching current limit could mean larger memory cell size; hence larger die size and higher cost, and slower programming speed. The lower current can be dealt with using sub-lithography features and this translates into manufacturing cost and yield issues.
Overall, Liu feels that better thermal management and better materials (such as germanium antimony [GeSb]) will improve PCM technology and challenges, such as improving the endurance cycle, are both materials issues and technology maturity issues. As PCM is intrinsically a low-voltage technology, it promises scalability in low-voltage applications and cost reduction in production.
PCM-based RAM could be primarily positioned as a replacement for NOR flash and is not a direct competitor to FRAM or MRAM technology applications. It also promises to work well as embedded memory for both microcontrollers and L2 and L3 caches. In the long run, PRAM could challenge both DRAM and NAND Flash, as those technologies reach their scaling limits, and subsequently replace them.
For more information contact Patrick Cairns, Frost & Sullivan, +27 (0)21 680 3274, [email protected], www.frost.com
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version