Pure polysilicon furnaceware could pave way for Integrated Materials success
Furnaces may not be the most glamorous tools in the fab, but they are certainly among the workhorses. “More than 50% of the time under the tool for making a device is spent in a vertical furnace these days,” says Tom Cadwell, president and CEO of Sunnyvale, CA–based Integrated Materials (IMI). But as feature sizes dip below 100 nm, the challenge to meet cleanliness, uptime, and other process and operational specs in front-end thermal equipment has become more difficult. More-mature fabs also need to tighten their belts and squeeze any bit of additional yield they can out of their processes.
Cadwell and his IMI colleagues believe their SiFusion line of towers, boats, pedestals, and other fixtures—the first high-purity all-silicon furnaceware—offers an alternative to existing quartz and silicon carbide consumables that can help chipmakers and OEMs achieve significantly improved performance.
Before Cadwell joined IMI in December 2002, he did his homework. He saw a company working with polysilicon, a material he had a great deal of experience with from his days at MEMC and Strasbaugh. He discovered that the chipmakers were definitely interested in the potential benefits of what IMI had to offer. In addition, he found some venture capitalists that might be convinced to invest in the company.
Once he took over, he saw three immediate challenges. “First, the company needed to be funded.” Then, although the basic technology was sound, “a cleaner fusing material” had to be developed. Finally, “the company needed to improve its operations to be more reliable and cleaner.”
On the operational front, the company did not have any cleanroom space and “all fusing had been done in brick kilns until then. Within three to four weeks, we stopped all production of towers for high-temperature applications, pending development of cleaner fusing material and having a tube furnace to fuse in.”
To generate a little cash flow and survive, Cadwell remembers how “we were able to drag in some receivables, collect insurance on a few towers that had been broken during shipment, and raise a little angel money.” With that bit of revenue, they managed to complete work “up to beta level” on a cleaner fusing material and bought a used furnace element and quartz tube on the cheap by August 2003.
A few months later, the first units shipped of what Cadwell calls the company’s current version of SiFusion. “The initial response from four customers was that the product was ionically clean,” he recalls. “Then in early 2004, we started to get feedback from customers about particle performance and having no need to clean [the parts] for the silicon nitride application. Most of what was going on then became qualifications that extended for a good period of time, because people wanted to put on a fairly thick coating [of silicion nitride] to make sure about the long-term proposition of not needing to clean unless there was some major upset.”
“The first commercial order began in early third-quarter of 2004 with a Korean customer” for nitride applications. “Then in fall 2004, we got some very good feedback from a large American chipmaker, and by spring 2005, they’d validated a significant improvement in high temperature and placed a big volume order in the second quarter.”
Cleanroom facilities helped IMI improve its operational capabilities.
Things started to turn around on the investor front as well, with an $8.5 million Series B funding in spring 2004, and then a Series C funding of $8.8 million in fall 2005 added needed capital. “During the time frame from [the first] funding in April 2004 to spring 2005, the acceptance of our product in terms of believing that it might be something that will help improve performance in furnaces” advanced considerably. “It went from, in the beginning, we would actually get laughter when they were told they wouldn’t have to clean the towers, to a year later, when that laughter had gone away because we had so much data from a number of customers showing that, indeed, you did not have to clean it.”
Although confident in their products, Cadwell and his cadre were pleasantly surprised by the performance data. “With LPCVD processes, customers will see an overall particle reduction in the 30% range for small particles and 50% range for large particles,” he notes. “But in a well-controlled environment, the reduction occurs in most cases regardless of the absolute level of existing performance.” In other words, “whether a guy is running at 100 particles or 40 particles, you see the same size reductions.
“The second thing that is surprising is how wide a variation in particle performance there is between customers. With our performance improvement, it really doesn’t seem to matter much whether we are comparing to quartz or silicon carbide, with maybe a little bit bigger improvement against quartz.”
IMI touts several other advantages for its pure polysilicon towers, boats, and other furnaceware. With LPCVD, in addition to the reductions in particle-generated defects, the components do not need routine cleaning, resulting in greatly reduced tool downtime. When it comes to diffusion tools, the company says that SiFusion components eliminate trace metal contamination and wafer damage, offer a better functional life span, and have more-precise manufacturing tolerances. Slippage is another area where IMI claims an edge over the competition.
The company recently commissioned David Jimenez of Wright Williams & Kelly to perform a cost of ownership study on LPCVD nitride deposition. The report states that for both 200- and 300-mm applications, the “SiFusion boats and liners provided the lowest cost of ownership by a substantial margin,” compared with quartz and silicon carbide components. “The higher purchase price of the SiFusion material is more than offset by its long life and drastically reduced risk of breakage,” the report continues. It also revealed wide differences among customers in the amount of tower or boat breakage and also big disparities in the time between cleanings, especially in 300-mm fabs.
One customer from a leading semiconductor manufacturer (who asked to remain anonymous) told MICRO via e-mail that IMI faces some key challenges. “For us, it was that the first, second, and third trials did not work. If they want to be accepted, the product has to work right the first time.” Although the materials engineer believes that the components have cleanliness and defect density reduction advantages, the furnaceware company needs to focus on long-term reliability and product recycling issues as well as further cost reductions to fit in with the chipmaker’s push to improve standard materials. But the customer sees a definite upside. “As devices get smaller and defect density specs get tighter, I believe that is where they can make an impact.”
The game plan going forward, according to Cadwell, is “to try to develop a commercial business with the numerous customers where we have ongoing qualifications and to try to cooperate with the furnace OEMs for our mutual benefit.” As part of their expansion plans, he hopes to have an Asian manufacturing site selected in the third quarter of this year.
The industry veteran is bullish on IMI’s future. “The biggest development in furnaces was somewhere back in the mid-’80s when someone decided to stand the furnace tube up because they couldn’t get the deposition uniformity with the wafers being vertical [in horizontal tubes]. Other than that, what we’re doing with silicon internals is the first major development in 20 years in furnace development.”—TC
“The BIG Idea” is a new feature that will report on interesting start-up and early-stage companies and technologies in the semiconductor and related micro/nanoelectronics industries.