Cliff Hirsch, Publisher, Semiconductor Times

An inside look at innovative semiconductor start-ups

The dearth of capital during the past two quarters is clearly noticeable, with few new semiconductor companies forming, venture capitalists protecting their strongest investments and weak companies dropping like flies. In this helter-skelter environment, it's hard to spot the obvious home runs. Yet, some trends are still evident.

For some time, there has been little room for traditional fabless startups in the 40-nanometer digital domain. Design costs are simply too high to make a reasonable return on investment. These days, most action appears at the fringe of the industry rather than in the mainstream.

After eliminating the $5 billion sub 65-nanometer foundries and their major league customers, what is left? Actually, quite a lot. There are still many emerging innovations in the compound semiconductor and microelectromechanical systems (MEMS) arenas. Materials such as Indium Phosphide and Gallium Nitride still have lots of runway, and there are many companies pursing optical, power, lighting and radio frequency (RF) products based on these materials.

From personal area network (PAN) to local area network (LAN) and wide area network (WAN), there is always innovation in the RF arena (both baseband and mixed-signal); however, most recent efforts are focused on the front end as the cost of baseband development has become prohibitive.

There have always been companies pursuing alternative memory technologies to replace both dynamic random access memory (DRAM) and Flash. Today, more than a handful of companies are making advances and one was recently unveiled.

Unity Semiconductor was formed in 2002 to develop storage-class non-volatile memory (NVM) products utilizing multi-layer memory array architectures and a new memory cell technology called CMOx. The company has raised more than $70 million to date and anticipates seeking another $25 million in 2010.

Why do so many companies chase the memory market? Because it's huge! Based on market research forecasts, Unity believes the total available market (TAM) for its storage-class NVM products will be approximately $15 billion in 2010 and grow to more than $25 billion in 2013.

Furthermore, Unity argues that the ability of NAND Flash technology to scale is reaching its limits, creating an opening for new technologies such as resistive random access memory (RRAM), ferroelectric RAM (FeRAM), phase-change memory (PCM) and magnetoresistive random access memory (MRAM).

In contrast to other emerging memory technologies, Unity argues that only its CMOx technology has the small cell size to beat NAND Flash in cost and density. Unity's objective is to have the smallest die size and the lowest manufacturing cost-per-bit storage-class NVM products. The company has filed an intellectual property (IP) portfolio that presently includes 60 granted patents, with another 90 patent applications in various stages of being granted.

Unity's technology is based on a passive cross-point, multi-layer memory array and its CMOx memory cell, a technology that is based on conductive metal oxide materials and the motion of ions. Unity's CMOx-based design uses four physical layers of multi-level cell (MLC) memory and is the key to increasing the density of its storage-class memory products. CMOx will yield products with 4x the density and 5–10x the write speed of today's NAND Flash.

Beyond technology innovation, Unity has also developed several groundbreaking business concepts. Key among these is the separation of front-end-of-line (FEOL) CMOS base wafer processing from back-end-of-line (BEOL) memory layer processing.

No new process technology is used in the CMOS base wafer, which can be fabricated on a trailing-edge 90-nanometer CMOS process. The CMOS FEOL strategy allows Unity to be a moderate follower in CMOS transistor technology. Its shrink path is unconventional in that a higher density memory core doesn't require base CMOS technology migration. Instead, Unity can use the same 90-nanometer base CMOS process for multiple generations, as well as use proven design IP to reduce risk and time-to-market.

The finer pitch BEOL memory strategy calls for Unity to form a joint venture partnership for volume manufacturing with a top-tier integrated device manufacturer (IDM) already in the memory business.

Unity has been processing 64-kilobit devices for two years, a 64-megabit device for one year, and is in design of a 64-gigabit device that is now close to tape-out and slated for pilot production in 2H 2010, with volume production in 2Q 2011.

Unity's planned second-generation products are high-capacity, high-performance 128-gigabit and 256-gigabit storage-class memories with an interface, command set and features tuned for state-of-the-art solid state devices (SSDs). The third generation will achieve the historic milestone of single-chip 1-terabit storage-class memory. Unity expects to build a 1-terabit storage-class memory product within five years using a 20-nanometer CMOx process technology.

Many alternative memory technologies have come and gone over the years. Why has it been so hard for any to achieve success? Certainly, the incumbent technologies are like steam rollers, flattening anything in their path. The inability of today's memory technologies to scale into the future, if this is truly insurmountable, may create an opening for new emerging technologies, and Unity is well poised.

The other issue is that few emerging technologies ultimately deliver. At the end of the day, they all have problems, whether it's cost, reliability, performance or manufacturability. Whereas many companies appear to have a singular focus on one technology, regardless of whether it's flawed or not, Unity appears to have taken a pragmatic view of the market, focusing on business and market issues as strongly as technology issues as demonstrated by its separation of front-end and back-end wafer processing. Based on this, I believe Unity's approach has much higher odds of success than many other competing alternatives.

Darrell Rinerson, Founder, Chairman & CEO L. David Sikes, President Robin Cheung, VP, Process Engineering Christophe Chevallier, VP, Design Engineering David Eggleston, VP, Product Engineering & Strategic Marketing Steve Hsia, VP, CMOS Technology Wayne Kinney, VP, Device Technology William Skolout, CFO Morgan Malino, Chief Legal Counsel

255 Santa Ana Court Sunnyvale, California 94085 USA (T) 408-737-7200 (F) 408-737-8067 (W) www.unitysemi.com

Cliff Hirsch (cliff@pinestream.com) is the publisher of Semiconductor Times, an industry newsletter focusing on semiconductor start-ups and their latest technology. For information on this publication, visit www.pinestream.com.