The need for less expensive non-volatile, high-speed, low-power, small sized ultra-high density storage devices for system-on-chip and application-specific-integrated circuits drive the growth of the embedded memory industry. The increasing criticality of the above mentioned parameters has led to the evolution of new memory technologies such as magnetic memories, phase change memories, and ferroelectric memories. Additionally, current research also addresses carbon nanotube, molecular and biomolecular memories.
Embedding memory into a logical processor/microprocessor is gaining importance in order to meet the growing performance demands of system-on-chip solutions or application-specific-integrated-circuits for various applications such as consumer electronics, telecommunications, wireless networks, automotive and space applications, notes Frost and Sullivan Research Analyst Kasthuri Jagadeesan. Magnetic, ferroelectric, and phase change memories have made their entry into the memory market with certain enhanced performance characteristics compared to embedded static, embedded dynamic, and embedded flash memories.
However, mass production and penetration into the existing memory market seems to be the greatest challenge for memory types such as ferroelectric RAM (FRAM), phase change RAM (PRAM) and magnetoresistive RAM (MRAM). Currently, magnetic and phase memories consume relatively more power than competing memory types and ferroelectric memories have yet to achieve relatively high densities.
Embedded static random access memory (eSRAM) and embedded dynamic RAM (eDRAM) have problems with non-volatility and soft error rates. Additional masking layers/processing steps during manufacture, high-voltage required for programming, endurance, and reliability represent the major issues associated with embedded flash memories, notes Frost and Sullivan Research Analyst Vishnu Sivadevan. Scalability beyond 45 nm node also remains a challenge for eSRAM, eDRAM, and eflash devices.
Given these challenges, memory manufacturers must undertake cutting edge research to achieve scalability competitive to other memory devices. Additionally, easy integration of the memory device with chips in embedded applications will contribute to greater acceptance of the memory technology.
Clearly, the race is on to develop a 'universal' memory technology which, while being non-volatile, also has fast access times, high density, low-power consumption, and the capability to be scaled below 45nm process node, says Frost & Sullivan Research Analyst Kasthuri Jagadeesan. An alternative memory technology that can achieve densities as high as Flash, and at the same time overcome the limitations of Flash memory will find considerable acceptance in the memory industry.
