Flash memory: growing markets, shrinking technologies, accelerating pace

Electronic News, Feb 3, 1997 by Saul Zales

The flash memory market segment continues an impressive growth rate because it is designed into high-growth market segment products, like networking equipment, wireless communications and Pentium processor-based PCs.

According to market research firm Dataquest, flash is now the fastest growing non-volatile memory type and the third fastest growing memory overall. Furthermore, Dataquest, In-Stat and Integrated Circuits Engineering (ICE) show a flash memory market segment forecast at about $8 billion by 1999. This expansion is fueled by two basic application profiles: updatable firmware and non-volatile data storage.

Systems with flash-updatable firmware provide significant value. They don't become obsolete as fast as EPROM- or ROM-based market designs. In the fast-paced digital systems world, this capability has become a basic checklist item. Flash-based firmware gives an OEM just-in-time capability and reduces risk to customers receiving early shipments.

Digital systems run protocols, and if flash-based, can be modified to extend a system's useful life cycle. For example, any system tied to the networking boom must be upgradable to deal with the digital chaos and the rate of innovation. Flash protects their customers investments, allowing for future upgrades.

Systems using flash memory for non-volatile data storage provide consumer value simply because flash maintains its data even without power. Consider early electronic organizers based on costly battery-backed SRAM where external power loss or dead batteries caused data loss. Flash solves this potentially expensive problem.

Flash memory has evolved over the last five years. Products that are smarter, market-focused, complete, integrated and more economical all contribute to the accelerated pace of usage. Early bulk erase flash memory, requiring external 12V supplies and manual program/erase control, have been largely superseded by second, third and even fourth generation flash memory.

The advances of each generation have required significant improvements in technology. The ensuing market expansion has required flash manufacturers to step up their investments in factories, high volume process control, Q&R capability and manufacturing innovation. Throughout the last decade, over 10 different flash memory "innovations" have been announced, each with its own promise.

Industry Support

Beyond technology differences, market growth requires industry acceptance of basic specifications. For example, the disk drive industry has grown increasingly around the IDE and SCSI specifications. Likewise, the flash industry has grown with the adoption of various interface specifications.

The PC Card specification enabled system design around a thin card with a 68-pin interface. Flash cards compatible with the PC Card specification can be found in electronic organizers and as protocol expansion modules for enterprise routers.

More recently, smaller form factors have arrived. These will compete for the next generation handheld consumer devices, such as digital cameras, audio recorders, PDAs and Internet appliances. The Miniature Card specification provides PC connectivity and improved interoperability, an approach endorsed by leading flash memory suppliers, PC companies and consumer electronics manufacturers.

Emerging Requirements

Unlike the selection of DRAM, where true commoditization has occurred due to the unifying force of huge PC demand, flash memory comes in many varieties, based on fundamentally different base technologies. Users must be aware of the implication of choosing one technology and its architecture over another.

Technology differences have emerged due to the significant complexity of a storage transistor that can be updated quickly, yet retain data after the removal of power can shrink with lithography improvements for cost benefit, yet withstand the 5-20MeV fields required to move charge across an insulating oxide.

The flash selection question today must take into account more factors than what supplier has the lowest cost and the best specification promise. Also for consideration are the overall trend-drivers in the industry, the investment commitment of the vendors, the history of promises, and the realities of technology-driven capabilities. Finally, total cost of usage (COU) requires consideration.

The following manufacturing issues should also be taken into consideration by memory OEMs:

* Advanced lithography capability and capacity: Lithography capability enables more bits to be printed on the same wafer. Migration from 0.8- to 0.6- to 0.4-micron process enables worldwide bit-supply to approximately double at each advancement, as well as support traditional semiconductor cost learning curve expectations of 30 percent per year. This will require increasing levels of investment in both technology and factory sophistication.

* Depth of reliability understanding: Making 0.8-micron flash memory is significantly easier than making 0.4-micron in high volume. Shrinking transistor size and decreasing voltage tolerance, combined with the same charge storage requirement, increases the complexity of providing reliable flash memory. Vendors entering the flash memory market in the first half of the '90s had sufficient difficulties, with many vendors never quite achieving their promise. Moving forward to the year 2000 will only see continuing increases in complexity, and greater risks for success or failure.