5/4 A Story of ”Believing is Seeing” ; Entrepreneur Engineering 20 th Anniversary 起業工学20年の一コマ

Small Money Chips In   Home

Back to Contents of Issue: April 2000

Small Money Chips In:  Bob Johnstone (2000年4月に同誌で掲載された記事です)
The remarkable story of how an unlikely collaboration between a Colorado Springs entrepreneur and a maverick R&D manager from Matsushita led to ferroelectric random access memory for cell phones — and smaller handsets around the world.

by Bob Johnstone

Gota Kano

Dr. Carlos A. Paz de Araujo
Everyone assumes the hardware that will enable next-generation wireless netphones is ready to roll.

But it is not.What’s missing is memory. Current i-mode cell phones connect you to the Net at 9.6 kilobits per second. This is fine for data like email and stock quotes. By 2003, peak speeds could hit two megabits — fast enough for webcasts of TV shows, interactive virtual reality games, and all sorts of other groovy things.

Where you gonna put all that video (which by the way is also in color)? Not in the hard drive, because your phone ain’t got one. In the chips, then, which in the context of today’s cell phones means flash memory. But writing to flash is v-e-r-y slow, nowhere near fast enough to cope with what promises to be a veritable tsunami of bits.

Ergo, a bottleneck that threatens to be a showstopper. There is, however, a potential solution on the horizon — ferroelectric random access memory, aka feram, a technology that combines high speed with low power. Feram is the product of an unlikely alliance, one that brings together a tiny US startup and a giant Japanese chipmaker, one that features a brilliant Brazilian engineer and a feisty Japanese R&D manager who had the guts to stick his neck out.

Truly, the story of how Carlos Paz de Araujo and Gota Kano, respectively of Colorado Springs-based Symetrix and Osaka-based Matsushita, and how they developed feram is one of the most remarkable — and least known — in the annals of high-tech entrepreneurship.

The story begins, appropriately enough, with a phone call. It was a Saturday night in 1991, and Kano was enjoying dinner with his family. Then the phone rang and on the other end was Araujo, who at that point Kano had not met, raving about the merits of his ferroelectric materials. “He was so eager and enthusiastic,” Kano recalls, and he kept pleading “Believe me, believe me.”

Kano was very surprised, especially because he knew that it must be about three or four o’clock in the morning in Colorado Springs. But Kano was also excited by Araujo’s passionate outburst and was impressed, not logically or scientifically, but emotionally. So on Monday morning, he went to ask his boss at the electronics research laboratory of Matsushita Electronics Corporation (MEC) for permission to work with Araujo. And, for the third time, his boss refused. The first time had been two years earlier, when an unusual envoy had brought word of ferroelectric technology to Kano’s lab in Takatsuki, an industrial suburb of Osaka. Hank Perry was a relative of Commodore Matthew Perry, the man whose “black ships” had opened up Japan to trade with the West in 1854. Now here was his great-grand nephew back with some black chips. The following year, Perry returned to renew his pitch. Kano was still keen, but once again his boss turned him down.

It was easy to understand the reasons for the refusal.

Ferroelectrics were not exactly a new technology. In fact, they had been around since the 1950s. Despite the name, ferroelectric materials contain no iron. The materials that Araujo developed are in fact ceramics, exotic compounds of strontium, bismuth, and tantalum.

Ferroelectric materials seemed ideal for application to computer memory because they were bistable. Jolt them with a current, and they changed polarity; jolt them again and they changed back. The problem was fatigue — switch them on and off a few hundred times and they would wear out. The best brains in the biz — IBM, Bell Labs, you name it — tried to fix the fatigue problem, but to no avail. By the 1980s, the field was more or less moribund.

In the interim, however, techniques for laying down thin films of material had improved beyond recognition. Now here was Araujo claiming to have licked the fatigue problem and raising the tantalizing prospect of the holy grail — a fast memory that (unlike D-RAM) would retain the information stored in it even after the power was switched off.

No wonder Kano, who had been working on semiconductors since joining MEC in 1961, was beguiled. So he decided to go over his boss’s head. He appealed to his long-time mentor Hiroyuki Mizuno, who was then a senior vice president at MEC’s parent company, Matsushita Electrical Industries (MEI).

(MEC was established in 1952 as a joint venture with Philips of the Netherlands. The subsidiary’s role was to act as a conduit for the Dutch company’s technology, in particular semiconductors such as germanium transistors and diodes. MEI bought out Philips in 1993.)

Mizuno gave his protégé the go-ahead. Kano hopped on a plane and went to Symetrix to meet Araujo and sign a collaboration agreement. Despite the differences in age — Kano was then 53, Araujo 36 — and background, the two hit it off immediately.

In the first stage of the collaboration, Kano did a very clever thing, one which illustrates the highly pragmatic nature of his thinking. “My mission was to win people’s trust,” he explains, “especially the management and my boss.” So, instead of shooting for the ultimate target — a ferroelectric memory — he opted for a more modest goal, one that would not involve switching so fatigue would not even be an issue.

As it happens, ferroelectrics make excellent capacitors (basic components used to store charge in almost all electronic circuits). “When Kano arrived and saw what I had,” Araujo recalls, “he said, ‘That’s very good, but please don’t make it into a memory, make it into a high-density capacitor for cellular phones, because we have to have this phone within seven months.’ And I went ‘Aaaaaaaugh!'”

The capacitors were to be integrated on a set of six gallium arsenide chips. Gallium arsenide was Kano’s main claim to fame. A compound semiconductor that outperforms silicon in some respects, particularly at high frequencies, gallium arsenide is notoriously bulky. Nonetheless, Kano and his crew at the electronic research lab managed to tame the material and apply it commercially as amplifiers in mobile (car) phones.

Now the idea was to take advantage of ferroelectrics to reduce the size of the amplifier chips. And that is exactly what happened. Symetrix completed the project on schedule, and in 1993 Matsushita put the chips into production. The resultant devices were 50 times smaller than their predecessors, and used only a fraction of the power. If you’ve ever wondered why cell phones have shrunk so much, here’s a large part of your answer.

In 1994, Matsushita’s chipset won the Nikkei Shimbun‘s Product of the Year Award. The following year, it was ready to meet the huge increase in demand for components that accompanied the launch of the personal handyphone. In 1999, Matsushita cranked out 120 million of the devices, which Araujo claims are used in 75 percent of all cell phones. Today, Matsushita is the largest producer of gallium arsenide chips in the world, with annual sales, according to Kano, approaching half a billion US dollars.

Having won his management’s trust by demonstrating that ferroelectrics were a successful technology that could be manufactured in high volume, Kano decided it was time to move on to memory. Once again, however, he was careful not to bite off more than he could chew.

The application Kano chose was contactless smart cards. The size of an ordinary credit card, contactless smart cards contain an antenna built into the edge of the card that enables them to be read without physical contact with the reader. The smart part is the chip they contain. Conventional contactless cards are used for applications like railway passes, toll roads, ID, drivers licenses, even telephone cards. Regular memory such as flash or EEPROM is adequate so long as all you need to do is read the information in the card. But not when it comes to updating (such as boosting the balance on a cash card) or downloading (such as authorizing transactions).

As e-commerce becomes ubiquitous, the problem is obvious. “When you’re authorizing a transaction, you don’t want everybody on the Net waiting for three minutes for authorization,” says Araujo. Conventional smart cards take about a thousandth of a second per bit to write, compared with 60 billionths of a second for ferroelectric memory. So instead of three minutes, users only have to wait a fraction of a second. “You go pi-pi-pi, transaction done,” Araujo claims.

On 30 June 1998 Matsushita began shipping the first of 1 million ferroelectric contactless smart cards for use by customers of Japanese gas stations and 7-Eleven convenience stores. Capable of storing 256 kilobits and costing less than ¥1,000, this was the first commercial feram product. It was the result of three years of intense joint R&D by Symetrix and Matsushita, who assigned around 40 people to the project.

Now, having proven the viability of the technology, Matsushita (pursued — at some distance — by most of the rest of the semiconductor industry, Symetrix having licensed its patents to all and sundry) is racing to increase memory density to cope with the requirements of mobile connected devices such as wireless handhelds and third-generation cell phones.

“You’re going to have all these phones, and they’re all going to have some interoperability with the Net,” explains Gavin Bourne, Silicon Valley-based director of business development for handheld OS specialist Symbian.

One potential scenario Bourne envisages is “somebody wants to trade a stock, and because the resources of the phone are going to be so low-level, you actually download the application and request simultaneously. If you follow the logic of that, sooner or later you’re going to come down to a memory layer, and that memory layer must execute extremely quickly.”

Worldwide sales of contactless smart card sales are predicted to reach $7 billion by 2003. The same year, setmakers expect to ship some 40 million mobile Net-connected devices to US stores. The demand for feram will thus be tremendous — but will the new memory be ready?

Only time will tell.

It is of course highly ironic that it should be Matsushita, of all companies, that is showing the rest of the world the way in ferroelectrics. This is, after all, a company commonly known in Japan as mane shita, “they copied,” in reference to the fact that Matsushita’s policy has long been to let other firms make the running, then simply reproduce the technology.

But such a policy is not well-suited to the fast-moving Internet Age, and Matsushita knows it. The company has recently been scrambling to remake itself US-style, offering employees stock options, linking management incentives to share price, and encouraging executives to start new, intrapreneurial ventures.

In this context, the success of the collaboration with Symetrix could be highly significant. (In October 1996, Matsushita paid $5 million for a 10 percent stake in the US company.) “Their culture is changing,” Araujo believes, “and this project is helping that culture to change.”

But is it realistic to think that such a highly contingent set of circumstances can be replicated? As Kano points out, “this kind of synergy has never been seen in Matsushita, nor in any other Japanese company.”

And in Japan risk-takers such as Kano are extremely rare. “He’s a very unusual individual,” says Jim Harris, a professor of electrical engineering at Stanford University who has known Kano for almost 20 years.

“He has grasped that when you take a chance, sometimes this thing works, and sometimes it doesn’t, and if it doesn’t — well, you just go on,” Harris says. “Kano is one of the few people who actually understands that and accepts that that’s a risk of doing something off the beaten path, and I think he has learned that lesson probably better than any other Japanese.”

Kano himself is eager to put his experience to good use. In October 1998, having reached the age of 60, he retired. He is now a professor at Kochi University of Technology in Shikoku.

In this beautiful but unlikely setting (Kochi has little or no industry to speak of), together with his old friend and mentor Dr. Mizuno, he is helping to start Japan’s first graduate school for entrepreneurs. Its main purpose? “To encourage and increase the spirit of entrepreneurship.”