When you name the pioneers of semiconductor tool and fab automation, Mihir Parikh is on the short list.
As the head of Asyst Technologies from its early days as a struggling company evangelizing the revolutionary idea of wafer-handling minienvironments to the early success stories of SMIF (standard mechanical interface) pods at TSMC and other fab companies to the development of SMIF's big brother, FOUP (front-opening unified pod), and the harrowing epoch of the false start/restart of the 300-mm wafer size evolution, Mihir and his team were there, turning vision into action.
After Mihir left Asyst in 2003, he took some time off, but he didn't "retire." No, not this driven soul. He was back in the saddle in 2004, working on a mysterious new venture that was so deep in stealth mode that he wouldn't even tell me its name until last year! Aquest Systems came out into the light earlier this year with news of a second round of funding: To no one's surprise, the company is trying to make a serious play in the semiconductor-manufacturing automated material handling systems (AMHS) space. Also to no one's surprise, Aquest's leadership team is rife with ex-Asyst folks, a seasoned group that Mihir knows and trusts.
Mihir Parikh, fearless leader of Aquest Systems.
After the company debuted its first product offerings in May, I knew I had to interview Mihir and get his side of this breaking corporate story. The following conversation actually took place via email, even if the familiar intonation of Mihir's voice seems to resound when he answers my queries. Here are the Q's and A's....
You had a good run with Asyst, and it’s not like you really had to work again in terms of making a salary. Why did you get back in the game, especially the automation game?
Why back into the automation game? Because there is a real opportunity to create a truly successful automation company, with innovative technology and a highly profitable business model. While I had the opportunity at Asyst to revolutionize the fab environment with minienvironments and SMIF/FOUP technology, I believe Aquest has a similar opportunity to revolutionize fab productivity by eliminating "waits" throughout the manufacturing process. That’s why I’m back. And yes, I don’t need and take a "salary"…in fact, I am investing in Aquest and automation.
What lessons did you learn from your Asyst experience that you have applied or can apply to the Aquest venture?
Given my 18-year tenure in building a $500 million company like Asyst, I have learned a great deal about both automation and meeting emerging manufacturing requirements. Now I get the chance to do it even better with Aquest. It starts with vision and innovation, which forms the basis for creating great products. Next, you need to complement great products with a customer-focused, solutions-oriented organization to achieve market acceptance and leadership. And finally, it needs to be both cost-effective and global if the company is to generate industry-leading profits and shareholder value. This is not only what I have learned over the years but it is exactly what is being done at Aquest.
AMHS and automation in general are very competitive spaces, with several entrenched players that have significant market positions and installed bases. Why does the industry---and the customer---need another player? What does Aquest bring to the table that’s not there already: Why should a chipmaker switch to your systems/approach, and if they do switch, how well do your systems/software “interoperate†with other vendors’ gear?
Existing automation, especially AMHS, products have minimal performance differentiation---each vehicle carries a single FOUP. But it’s also important to note that while they have a capacity of one FOUP, typically about 60% of them are running around the fab empty, going to pick up or return after dropping off FOUPs. So their effective capacity is about 0.4 FOUPs. All of these empty vehicles add to the traffic congestion, which causes delays in critical wafer deliveries. This is a highly inefficient way to move FOUPs.
Aquest brings a completely new technology---FabEX AMHS---which is not limited by any vehicles and therefore has capacity determined only by the packing density of FOUPs. And it is fast, [with a] higher speed and higher acceleration than other OHT [overhead transport] vehicles. This will be particularly critical to enable high capacity and short delivery times when the industry goes to smaller lot sizes, which will be imperative for high-mix, high-flexibility manufacturing environments. The industry needs a better solution than current AMHS and automation.
Based on our discussions with chipmakers, it is clear that there is a pressing industry need for a next-generation AMHS that can deliver both high speed and high capacity. We call it No-Wait-Manufacturing, with No-Wait-AMHS as the first step in our vision. Aquest’s first installation of FabEX technology is in the process of demonstrating this capability in an existing fab, with an already installed vehicle-based AMHS. This kind of interoperability is achieved by leveraging the now-mature industry standards. It is anticipated that compared to existing AMHS, FabEX will provide up to a 400% increase in capacity with significantly shorter delivery times. Stay tuned!
What are the key issues---both technically/technologically and economically---facing the AMHS/automation space?
The key issue facing automation is the inefficiency and reduced productivity due to delays or "waits" throughout the manufacturing process. Examples of "waits" are delays caused by unavailability of vehicles to transport or pick up/drop off FOUPs; delays caused by scheduling complexity in response to hot-lots or equipment down situation; and delays caused at equipment front-ends not optimized for fab automation.
One major factory integration challenge facing the industry is how to deal with high-product-mix/small wafer-lot fab scenarios. What impact does the amount/level of product mix have on FabEx and the No Wait Manufacturing scheme? Also, are you seeing more interest from certain types of manufacturers (foundry, IDM; memory, logic, etc.)?
FabEX is, by architecture, designed for high-volume as well as high-mix manufacturing. Because of its inherent design flexibility (both high-speed and high-capacity), it is adaptable to any form of manufacturing , including the high-mix requirement of a foundry, high-volume requirements of a DRAM manufacturer, or mixed products of a logic or ASIC facility.
You claim FabEx/NWM is faster, has shorter wait times, better cost of ownership, less vibration, etc. than competitive/existing systems. Do you have modeling or early case studies to back this up, and if so, what kinds of specific improvements in cycle times, reliability, particle cleanliness/ESD, tool utilization, capacity degradation, and other tool and factory effectiveness metrics (MTTR, MMBF, MCBF) do you see in your studies?
Aquest has developed a very comprehensive simulation and modeling capability for fab process flow and AMHS. Results for a variety of fab models show that FabEX will shorten delivery times by 40-50%, and that FabEX provides a maximum delivery time that is five times shorter than OHT-based AMHS. FabEX improves delivery time predictability by about 50% and can deliver four times as many FOUPs This increased capacity can support smaller lot sizes and prevent production perturbations from clogging the AMHS. FabEX can reduce overall fab cycle time, by [approximately] 10% and could result in many tens of millions of dollars of increased output or profit for the IC manufacturer.
In the press release info issued last month, it said Aquest had “received significant commitments from some of today’s leading chipmakers†and would begin shipping and installing systems in June. What do you mean by “significant commitmentsâ€â€”are those revenue generating orders or JDPs/betas with little or no initial revenues? Can you say who those “leading chipmakers†are, or at least how many of the top 10 or 20 you have dealings with? And as for those ships/installs, did you carry those out successfully in June?
We announced our first purchase order on July 9 from PowerChip Semiconductor. This is a multimillion-dollar-valued installation that is expected to significantly improve their capacity and shorten delivery times. As you know, many of my key team members and I, through our long and successful experience in building Asyst, are very connected to the top IC manufacturers at all levels. As such, we have targeted specific potential customers. Through comprehensive simulation and modeling, these customers are convinced that FabEX AMHS can provide significantly improved performance over existing vehicle-based AMHS. They are engaged with us and are in line to evaluate our system later in the year. That’s all we can say for now.
Where do you manufacture your systems, and if you work with a contract/outsource manufacturer, which one? What are the manufacturing cycle times for a system, and what kind of capacity/output (systems per month) can you generate?
At present, we are in preproduction manufacturing in Taiwan. We plan to volume manufacture our FabEX subsystems through a "partnered" contract manufacturer in Asia, with proven capability to provide highly responsive, very high-quality, cost-effective products. Because FabEX is modular by design, we expect to be very responsive to customer demands. We expect to be able to ramp our manufacturing by well over 100% per year for the next few years.
What’s the game plan for Aquest moving forward, short and longer term? Do you have any plans for reticle handling systems? What about close(r) cooperation with process OEMs?
FabEX is a very adaptable technology. While we are initially targeting IC manufacturing, our technology is designed for other applications, including reticle transport and FPD manufacturing. In fact, our breadth of patent applications covers such potential opportunities. Leveraging our long association and experience at Asyst, we are very close to many of the major OEMs and are optimizing our FabEX for existing equipment front ends and eventually hope to have their equipment front-ends design optimized for FabEX.
What’s your opinion about an eventual transition to a larger wafer size, i.e., 450 mm? Do you think it’s necessary and if so, what does the industry need to do to make it happen successfully?
300 mm is going to have a long life. The next phase in 300 mm (300 Prime) is going to be driven by significant improvement in fab productivity, not just technology-driven process improvements. The 300 Prime learning will set the stage for 450 mm. The 450-mm transition is going to be much more expensive than the 300-mm transition and may be much further out than is talked about by some. Regardless of when 450-mm occurs, the flexible design of FabEX ensures that the high-speed, high-capacity solution that FabEX offers will be the preferred solution for 450-mm IC manufacturing and beyond.
|
Blogs 
