Life is Good

pms laflame at mindspring.com
Sat Jun 12 10:35:24 PDT 1999


Oh gee Chris, I'm more on the level of: Ugh. Non-volatile memory, good.

Do me a favor, go over to Ramtron homepage and tell me if what they got is good enough to make a bunch of cool, new consumer gadgets.

Alls I know is, Japanese companies keep giving them money and staff (and they're sueing NEC for patent infringement of their EDRAM)

And I've lurked on the message board at Tom's Hardware and apparently the Ramtron folks are admired by the techno-rebel crowd.

Say, isn't this Rambus thing a hoot. Is this the middle of the beginning of the end of the Intel Empire? I read that pc 133 tech is on the streets of Japan. Whatever that means. (just kidding, I sorta know what it means).

As far as the Electrical Union goes, I may be assuming incorrectly that they have something to do with the Electrical Beneificial Fund, which is carrying a big marker on Ramtron, which in turn is looking to restructure. What does this all mean? Not sure. They sure do sound like a cute lil ole company though. Been watching since about 25 cents a share, 67 now.

It sounds to me like Asia is making a tech comeback that's not the tech comeback US companies may have envisioned. The Chinese could poo-poo MSFT CE OS and the Japanese seem to have left US companies in the dust in the retreat from Rambus world. Palm's will probably be passe by next Christmas(the next fersure, give me a Panasonic, please) And the Europeans are really moving along on the next generation satilite thing. Oh my, my, the worm is restless.

Say Chris, what about this DDR RAM stuff? Tom's Hardware thinks that's the ticket.

Gotta get away from the magic box now.

toodles paula

PARTNERS IN CAPITAL

Matsushita Hits Pay Dirt in the Rockies

In semiconductors, the time lag from initial invention to commercial fruition is notoriously long. In the case of Symetrix, a closely held startup nestled next to the Rocky Mountains in Colorado Springs, Colorado, the wait has lasted more than a decade. But as Symetrix’ long and still-growing list of licensees attests, the company’s so-called ferroelectric technology is red-hot. The giants of the semiconductor world have enlisted Symetrix’ skills, among them NEC, Mitsubishi, Matsushita Electrics Corp. of Japan, Motorola of the U.S., Siemens AG of Germany and Hyundai of South Korea.

Matsushita, however, enjoys an enviable lead over its rivals. Several products that rely on the Colorado firm’s technologies, such as mobile phones and smart cards, are already in the marketplace courtesy of the Japanese electronic giant’s early reliance on Symetrix’ technologies and its investment in time and money in the firm. Other potentially revolutionary applications for semiconductors may be just around the corner.

Matsushita put its faith in Symetrix in the early 1990s and took a 10 percent stake in the firm in 1996. The reason: Symetrix’ ferroelectric materials and integrated capacitor technology promises to clear away the complexity in manufacturing next-generation gigabit-class dynamic random access memory chips and revolutionize the market for computer memory chips in other intriguing ways.

This promise is not just theoretical. Ferroelectric capacitors are already a factor in the global marketplace. Matsushita Electronics is currently cranking out millions of phone chips per month with integrated ferroelectric capacitors.

Symetrix’ claim to fame is this hitherto obscure family of materials called ferroelectrics. In the form of capacitors, they are simple plate-like components that are used for storing electrical charge and are used as such in mobile phones. Ferroelectrics as insulators, however, polarize when exposed to an electrical field — and stay switched on even when the field is removed. This has enormous consequences for the future storage capacity of computer memory chips.

The three men who discovered the usefulness of ferroelectrics and went on to found the firm are a diverse international set of scientists from Brazil, Australia and the U.S. The Brazilian-born Carlos Araujo, 44, the company’s chairman and chief executive officer, directs the microelectronics center at the University of Colorado at Colorado Springs. Larry McMillan, 61, a former executive with Motorola, NCR, Storage Technology and Honeywell, is the president and chief executive officer of Symetrix. And the 55 year-old James Scott, an ex-Bell Labs researcher, is currently a professor of physics at the University of New South Wales and is a member of Symetrix’ board of directors. A fourth key figure is Stephen Philby, the company’s chief operations officer, who joined shortly after the founding of the company in 1996.

Symetrix’ founders made their key breakthrough in ferroelectics in 1991 but decided to keep the details of the discovery under wraps until they were ready to file a comprehensive set of patents. Such secrecy drew criticism from peers but the firm’s founders remain unfazed. “The tradition in science is that if you publish, you have to give all the details so that other people can reproduce your results,” Scott explains. “But that’s not compatible with commercial interests.”

In April 1995, just one month before the disclosure of their key patent, the Symetrix researchers published a paper in Nature magazine. Soon there was a rush of companies seeking to license the new technology.

The reason: Ferroelectric memories have many advantages over conventional memory chip technologies such as electrically erasable programmable read-only memory and flash memory. Ferroelectric memories are more durable, operate on lower power, and because of their simple structure — just three process steps on top of standard silicon devices — are cheaper to make. But the biggest plus of all is that information can be written to this new type of chip at least four times faster than conventional chips. In mobile applications such as networked smart cards, such a high access speed is a key advantage.

Last October, Matsushita announced it had developed such a memory chip based on Symetrix’ ferroelectric technology, and applied it to a “contactless” smart card. The company hopes that contactless ferroelectric cards will replace conventional, magnetic-stripe smart cards in a myriad of applications, including electronic commerce, drivers’ licenses and commuter travel. This market is currently worth about US$5 billion and growing fast.

Beyond smart cards, two even bigger markets beckon for ferroelectric memories. One is ferroelectric memory as an adjunct to embedded processors. Chipmakers ship about 50 times more of these devices — which are used to control everything from windshield wipers to washing machines — than they do microprocessors for personal computers. The other is computer memory itself. Because of their superior capabilities — non-volatility, high speed switching, and durability — ferroelectric chips could conceivably replace both magnetic disks and DRAMs in future portable computers. They are, it appears, the perfect memory device.

This is good news indeed for bulk producers of memory chips like NEC. A dynamic random access memory remembers by storing charge in capacitors. As the density of integration — the number of bits each chip can store — quadruples every three years, chipmakers have been forced to resort to ever more complex strategies to provide enough storage space. For several generations now, they have employed three-dimensional geometries, either etching tiny trenches into the silicon wafer or stacking capacitor “skyscrapers” above it. But even such elaborate methods are now beginning to run out of steam.

By the time they hit 4 gigabits, sometime around the year 2004, chipmakers still using a low-dielectric-constant capacitor like silicon dioxide in their DRAMs face an absolute nightmare. “Whatever objections you have to using exotic materials like ferroelectrics, you have no choice,” Scott claims. “They’re the only game in town.”

Matsushita was the first to realize the potential of Symetrix’ ferroelectric experiments. In 1991 the two firms joined forces to integrate a ferroelectric material, barium strontium titanate (BST), as the capacitor on a set of high-frequency gallium arsenide amplifier chips for use in digital mobile phones. By 1993 Matushita had fully integrated BST into its production.

In tying up with Symetrix, Matsushita unwittingly gained access to a second, and potentially far more significant, ferroelectric technology.

In August 1991, six months after the initial agreement between the two firms, Symetrix co-founder and chairman Carlos Araujo discovered that another ferroelectric compound, strontium bismuth tantalate (SBT), was the solution to a daunting problem that had long prevented the commercial application of ferroelectrics as full-fledged memory chips in their own right.

This was the problem of fatigue. Ferroelectric memories work using a technique called destructive read-out. That is, every time information is read out of the memory, the information is destroyed and must be written back into the memory again. Using conventional ferroelectrics, notably lead zirconium titanate (PZT), which was the focus of research at most companies, for each rewrite, the amount of charge that is switched decreases until ultimately the information becomes unreadable.

With ordinary barium titanate, this point is reached around 1 million cycles; with PZT, at around 1 billion cycles, well below the level required for a commercial device (at least 1 trillion). With SBT, however, there was no evidence of any degradation in switched charge. Symetrix had demonstrated for the first time that it was possible to eliminate totally the fatigue problem.

Turning fragile prototype ferroelectrics into robust commercial devices took three years of intense engineering effort. In learning how to integrate ferroelectric films into commercial devices, Symetrix relied heavily upon help from Matsushita. The Japanese firm assigned 40 engineers to the project. “We are winning only because the folks at Matsushita made devices that show the advantages of our technology,” says Araujo. “Without them, there would be no Symetrix.”

By the same token, thanks to Symetrix, Matsushita currently has a lead of several years over its competitors. Matsushita is launching ferroelectric memories aimed at three markets.

The first is a 32 kilobit memory embedded in an 8-bit silicon microprocessor, which is scheduled to be marketed later this year. Chipmakers ship about 3.5 billion such devices a year for use as microcontrollers in such everyday products as cars and household appliances.

The second product is the contactless smart card launched last October. Capable of wireless communications, it has many potential applications. An obvious initial one is commuter travel passes. When Japanese train stations switched to automated ticket gates a few years back, commuters were annoyed because instead of flashing their passes at the ticket collector, they had to physically remove the passes from their wallets in order to stick them in the machines. The new system allows commuters to wave their tickets over a reader at a distance of up to 30 cm.

Japanese commuters are not the only ones who can look forward to an easier life. Motorola, which is marketing smart cards supplied by Matsushita, reckons that some 50 cities around the world have put out requests for proposals for automatic fare collection systems based on smart cards. For example, the Malaysian government is seeking a supplier of smart cards that combine identity, health insurance details and a driver’s license.

Smart cards that can store cash are yet another market. Consumers may well prefer the convenience of smart cards that do not have to be swiped through a machine. But ferroelectric cash cards will also score over conventional cards in that they can be topped up much faster.

And as Araujo points out, in any kind of electronic commerce on the World Wide Web, “if the card has a very slow download, the network will be tremendously slow, and the typical cards made with EEPROMs (electrically erasable programmable read-only memory chips) have download speeds that are at least 20 times slower.”

An offshoot of smart cards will be radio frequency ID tags. These will substitute for bar codes in many applications, most notably as baggage tags used at airport check-ins. But the tags will have to be extremely cheap, 50 cents or less, which in turn means that the chips will have to be very small. In the short term, there is not enough demand for tags to fill the production demands of a busy semiconductor plant. Thus the tag market will probably have wait until it can piggyback on mainstream ferroelectric products like embedded controllers and smart cards.

Finally, there is the stand-alone memory chip. Here, Matsushita has developed a 256-kilobit chip that is aimed as a much cheaper substitute for battery backed-up static random access memories. In addition to Matsushita’s efforts in this area, companies like Symetrix licensees’ Hyundai and Siemens are racing to develop four-megabit ferroelectric memories. Araujo explains the attraction. “If you’re going to do a hand-held network computer, all you need is about 8 megabytes,” he observes. “With that you can download anything. And so this substitutes for having a hard disk drive.”

What’s more, says Araujo, “you cannot have DRAM and non-volatile RAM and make that computer inexpensive, so that the batteries don’t die on you.” What is required is a main memory that is non-volatile, he explains. “Currently there is no main memory that can be non-volatile, because all memories that are non-volatile are too slow in the write cycle, and/or they suffer from fatigue.” The only way to clear the impasse, he says “is to use ferroelectric memory.” Araujo predicts that the minute a ferroelectric memory chip becomes available, you’ll see a stampede in this direction.” He thinks Siemens will probably have such a chip by early 1999.

Not bad timing for Symetrix, which is contemplating the benefits of an initial public offer over a private placement toward the end of the year. The founders are well vested in the company’s stock and are in no rush to release equity. The company had originally planned to do an initial public offer this year, according to Scott, but instead may end up selling a block of stock to one or two private investors or other corporations. “Once you go public you basically have a lot of people to answer to,” Scott explains.

— Bob Johnstone

At 06:59 AM 6/12/99 -0400, you wrote:
>pms wrote:
>>
>> ps. Memories.......FRAM is definitely the future. Whether Ramtron and the
>> Electrical Union(30%) have got the goods, is the $64,000 question.
>
>How *is* that working out? BaTiO2, SrTiO2, PbZrTiO2 or something
>else for the ferroelectric? What's the deposition technique? CVD,
>sol-gel?
>How about fatigue after many R/W cycles?
>
>Who's the Electrical Union? I used to know more about this...
>
>Chris
>



More information about the lbo-talk mailing list