Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > EDA/IP

ZyCube duo has big plans for 3D circuits

Posted: 02 May 2005 ?? ?Print Version ?Bookmark and Share

Keywords:2d? 3d? chip technology? zycube? vias?

Bonkohara: ZyCube may well be the first venture company in Japan in decades that has the capability to go from design to production.

ZyCube Co. Ltd is run in tandem by its co-founders: Manabu Bonkohara, president and CEO, and CTO Mitsumasa Koyanagi, a Tohoku University professor who developed the 3D chip technology upon which the venture is built. The 3D architecture presents the possibility of folding all kinds of functional blocksprocessor, memory, sensor, logic, analog and power ICsinto one IC, interconnected by embedded vias with a vast bandwidth. In chipmakers' packaging road maps, 3D always appears in the 2008-2010 time frame. But ZyCube has recently done a product launch, and Koyanagi built an operating artificial-retina chip in three layers. The duo sat down recently with EE Times' Yoshiko Hara, who has translated their conversation from the Japanese.

EET: Tell us about your technology.

Koyanagi (K): The base concept is to make vias in a 2D large-scale IC (lsi), thin it to expose the ends of the vias and then stack it atop another LSI. I have the basic, important patents for the 3D LSI, but they do not necessarily cover all of the technologies involved.

Are other companies working on the concept?

K: There are various research projects concerning 3D LSI fabrication. In Japan, the Association of Super-Advanced Electronics Technologies (ASET) is investigating a stacked chip-on-chip approach. Research institutes in the United States, such as MIT, are developing the concept by stacking wafer-on-wafer. They are also trying different materials, but I believe that the base for most of them are my own Tohoku University method.

Bonkohara (B): When I was preparing to launch a 3D packaging project at ASET about seven years ago, I found out about Professor Koyanagi's 3D LSI technology. It was outstanding, but it was impossible for the idea to come to fruition at that time.

What makes it possible now? Was alignment of the different wafers a big challenge?

K: Yes, it was challenging. But now it's not, thanks to a newly developed technology that I cannot detail yet. It is not simply a 3D technology, but a whole new integration scheme that will drastically change the process of manufacturing LSIs.

When many different functions are fabricated on a single LSI chip, they naturally must all use the same process technologysay, a 0.1?m process. But that is not always the best technology for each of the individual functions in the design, from the standpoints of performance, power consumption and cost. If there were some ways to fabricate each of the blocks in the process best suited for it and then glue them together, the result would be a higher-performance LSI at a lower cost. We are developing a method for doing that, using a process technology similar to those now used for standard chips.

Isn't that the same concept as system-in-package?

K: SiP does not satisfy all the requirements of high performance, low power consumption and low cost. Wire bonding is the limitation. Even when microbumps are used, only two LSIs can be stacked. Our target is the ultimate LSI.

The via should be the core of the 3D LSI. How can you make small-diameter vias?

K: We can make vias of about 0.5m in diameter that go all the way through a silicon wafer, which is 20 to 30?m thick. That is a diameter-to-depth ratio of about 1:40, or in our shorthand, an aspect ratio of 40. Normal contact openings in LSI processes are also about 0.5m across, but the holes only go through about 0.8?m of oxide in a normal 2D LSIthat's an aspect ratio of 1:6. So we are talking about a hole more than 20 times deeper than a contact opening. A dedicated etching system is required to cut such a deep trench.

Does ZyCube own dedicated production equipment?

B: Some production systems have become available on the market, though they are not fully satisfactory for our needs. But we are building some quite specialized equipment for 3D LSI production in-house.

K: Since we are stacking wafers one on top of the other and the vias have to line up, we need special equipment to place the wafers in vertical alignment before we bond them together. And we must also have special equipment to inject the adhesive that glues the wafers together. That step is, of course, not present in a 2D LSI process where you have only one wafer.

ZyCube installed a pilot line recently. Does it basically consist of used equipment?

B: The line is an 8-inch wafer line consisting of 60 percent used equipment and 40 percent new systems on the investment-value basis. New systems are expensive and they need some modification for 3D LSI production. We are going to do the modification in-house to protect the technology.

K: The forefront of semiconductor processes is 90nm, but 3D LSI production does not require such an advanced process initially. Rather, we are working at the 1-3?m scale. For a quick launch, we'll employ used, low-cost production facilities. Even such old facilities can build products with higher performance than the output from the most advanced lines.

For ultimate 3D LSIs, however, we need special, dedicated systems. We are going to have lines with such special systems eventually. At present, we have connections with production system makers, but we have not yet entered into business partnerships with any of them.

You describe your business model as "semifabless." ZyCube will do about 30 of the 500 processes needed to build its 3D chips, leaving wafer fabrication and packaging to third parties. How did you come up with this concept?

K: IC companies make their investment plans based on the International Technology Roadmap for Semiconductors. As a process scales down, the investment that's required swells. At a certain point, there will come a breaking point where the investment needed will surpass the size of the business. That's why we are going to establish a different business model.

The front-end production will eventually fall into the hands of several large foundries. Using such foundries to fabricate LSIs, we have devised a way to add value. Only by combining the design and production of differentiating processes have we invented a new business model.

B: ZyCube may well be the first venture company in Japan in decades that has the capability to go from design to production.

What are your plans for the near future?

B: We are going to start offering samples around the end of January. It will be an image sensor built as a one-layer LSI. Only a part of the 3D LSI technology is employed for the sensor, but it will be a highly value-added part. Based on customers' reactions, we will nail down plans for volume production. And we want to be listed on the Japanese stock market in 2007.

K: We have many customers who are asking us to make simple 3D LSIs as soon as we have the capacity. This is the first step. We want to offer ZyCube's unique products, such as the artificial retina, as the second or the third step.

You opened a branch office in Silicon Valley last June. What is its mission?

B: We wanted to start business in the United States, as the conditions for a venture company in Japan were not favorable at all. Japan's way of fostering venture businesses is inept. Thus, although quite a few venture businesses have succeeded in Japan, many more new ventures emerge in the United States.

But our technology base was Tohoku University and we ourselves live in Japan. Our legal adviser told us to first establish a company here. But as the United States was an important market, we founded a branch office. Moving the headquarters to the United States is a possibility one day.

- Yoshiko Hara
EE Times

Article Comments - ZyCube duo has big plans for 3D circ...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top