Key Takeaways

• Alternative memory technologies are emerging as potential successors to Flash and DRAM, including MRAM, FRAM, and ReRAM.
• Current trends indicate that microcontrollers will adopt these alternative memories for non-volatile storage, with automotive applications leading the way.
• Despite initial challenges, these new memory types show promise for scalability and performance improvements in future chip designs.

Recent advancements in memory technology highlight the ongoing evolution from traditional semiconductor memories to innovative alternatives. When semiconductor memory technology began in the 1980s, options were limited to SRAM, DRAM, and early forms of EEPROM. Over the decades, Flash memory gained popularity, especially from the 1990s onward, leading to extensive use in modern devices. However, the limits of scaling in traditional memories have prompted researchers and manufacturers to explore alternative memory types like FRAM, MRAM, ReRAM, and PCM.

At a recent webinar organized by the Storage Networking Industry Association (SNIA), experts Tom Coughlin and Jim Handy provided insights into the current status and future potential of these alternative memory technologies. Coughlin emphasized the challenges faced by microcontrollers, which commonly use NOR Flash memory for code and data storage. Due to limitations in the physical scaling of NOR Flash memory at the 28nm node, there is a growing need for alternative nonvolatile memory solutions in these applications.

Coughlin outlined three possible paths for upcoming microcontrollers facing this memory transition. The first option involves coupling FinFET logic circuits with planar NOR Flash on a single chip; however, this approach is not economically viable for lower-cost microcontroller applications. The second option, using external Flash memory for code, introduces additional costs and packaging complexities. Additionally, this method impacts performance since it necessitates moving data from external memory to on-chip SRAM to execute applications efficiently—an expensive and complicated solution.

The third and more promising alternative involves implementing different types of on-chip nonvolatile memory such as FRAM, MRAM, and ReRAM. Notable developments already showcase this trend; for instance, NXP and TSMC are collaborating on MRAM-based microcontrollers for automotive use, while Texas Instruments offers MSP430FR57xx microcontrollers that utilize FRAM. Furthermore, Nuvoton announced a new microcontroller integrating ReRAM for non-volatile storage, indicating a shift towards these advanced memory technologies.

Handy focused on the struggles faced by traditional NAND Flash memory, which, after reaching physical limits at approximately 15nm, has pushed manufacturers toward 3D NAND solutions. While companies like SK hynix have announced chips with as many as 321 layers, these technologies still encounter scaling limitations. In light of these challenges, alternative memories offer promising attributes like faster write speeds and improved radiation tolerance, making them suitable for demanding applications like aerospace and military uses.

Despite their potential, Handy acknowledged that early attempts to integrate these memory solutions, such as Intel’s PCM technology, have not succeeded on a commercial scale. The economic feasibility of production compared to NAND Flash and DRAM remains a significant barrier. However, as semiconductor manufacturing continues to evolve and requirements shift, alternative memories may gain traction by employing newer materials and architectures that align better with compact chip designs.

A timeline presented by Handy forecasts that the integration of alternative memory technologies into mainstream applications may take another decade. Embedded applications will likely see faster adoption due to ongoing advancements in microcontroller technology. In contrast, standalone memory chips will experience more delay because of the current stronghold of traditional memory types.

Coughlin and Handy’s findings indicate a significant future shift toward alternative memory technologies, though economic and technical challenges will determine the pace of this transition. Their comprehensive report, “A Deep Look at New Memories,” goes into further detail and is available for those interested in exploring this critical subject matter further.

The content above is a summary. For more details, see the source article.