TORONTO — Ferroelectric memory (FRAM) was supposed to dominate by the mid-90s — at least according to an old Dataquest research forecast — but instead FRAM, also known as FeRAM, remains one of those many memory technologies that haven’t quite taken off as hoped.
FRAM works internally like DRAM, has the high speed of SRAM and the non-volatility of flash memory. When the first successful circuits were built in the 1980s, it was thought their universal function would allow them to replace DRAM, SRAM and EEPROM in many applications.
“It just never happened," semiconductor analyst Jim Handy (Objective Analysis) told EE Times in the wake of Cypress acquiring Ramtron. "It ended up that it was always a more expensive technology to build than conventional memory technologies.”
Since the late 1980s, FRAM developers ran into all kinds of problems due to a limited understanding of the physics of ferroelectric materials, many of which were similar to those faced today by other promising non-volatile memory technologies under development, such as magnetoresistive RAM (MRAM), resistive RAM (RRAM), and phase-change memory (PCM). But despite a bumpy road, FRAM is in commercial production with product lines and use cases.
In a more recent chat with EE Times, Handy recalls someone observing that “'FRAM is the technology of the future and destined to stay that way’. FRAM has been something that has held promise for some time but hasn't really main inroads.”
As with all memory technologies, cost is a barrier to adoption. FRAM relies on a layer of Perovskite crystals being deposited on top of a standard silicon substrate. Perovskite crystals contain elements that can interfere with silicon transistors, so a barrier layer is generally used to isolate this layer from the silicon below. Handy said this adds to the device’s cost by increasing the cost of processing the wafer.
But there is some demand: A report released by Research and Markets just over a year ago predicted the global FRAM market would grow 16.4% compound annual growth rate (CAGR) between 2013 and 2018, with low power consumption being one of key contributors to market growth. Vendors, including the now Cypress-owned Ramtron, as well as Texas Instruments and Fujitsu, were also increasing research and development in both hardware and software. This increased spending had improved the performance and efficiency of FRAMs, the report said, while increasing the application areas for FRAM.
“Chicken and egg” challenge
Fujitsu Semiconductor’s most recent FRAM product was announced at the end of 2014 with the release of MB85RDP16LX, an ultra-low-power device with an integrated binary counter function, which contributes to energy savings. The company is targeting MB85RDP16LX at industrial automation applications involving energy harvesting for rotary encoders, motor control and sensors. To meet the temperature ranges required by the industrial automation market, it can operating at a temperature range of -40 to 105 °C without risk of data loss in a 10-year timeframe.
Cypress, which bought FRAM pioneer Ramtron, has committed to the technology just as it has to SRAM. Last month, it introduced a family of 4Mb serial FRAMs. The new product line is aimed at applications requiring "continuous and frequent high-speed reading and writing of data where security is essential," said Rainer Hoehler, VP the nonvolatile products business unit at Cypress, in an interview with EE Times. Applications include industrial controls and automation, industrial metering, multifunction printers, test and measurement equipment and medical wearables.
These systems require high-density, high-reliability, high-endurance and energy-efficient non-volatile (NV) RAMs, Hoehler said. Alternative nonvolatile memories, such as EEPROM and MRAM, cannot match the performance of FRAM, which consumes 30% of the power of the most advanced EEPROM and offers 100 million times the write endurance, he told us.
Handy said FRAM is similar to many memories as it faces the quintessential “chicken and egg” challenge. Demand for more volume is required make wafers for cost-efficiently, and the only to obtain that demand is a lower cost for the FRAM.
Despite the challenges, he said there is a viable market for vendors focusing on FRAM applications.
—Gary Hilson covers memory and flash technologies for EE Times and is the editor of Memory Designline.
Join over 2,000 technical professionals and embedded systems hardware, software, and firmware developers at ESC Silicon Valley July 20-22, 2015 and learn about the latest techniques and tips for reducing time, cost, and complexity in the embedded development process.
Passes for the ESC Silicon Valley 2015 Technical Conference are available at the conference’s official site with discounted advance pricing until July 17, 2015. The Embedded Systems Conference and EE Times are owned by UBM Canon.