In yesterday's opening segment of Chip Shots' story about the official launch of Chinese equipment supplier AMEC, chairman/CEO Gerald Yin shared a bit of the company's background and strategic thinking.
This concluding post focuses on the AMEC's two tool systems, the Primo D-RIE dielectric etcher and HPCVD system.
Since it sees itself as a premium brand, AMEC wanted its tool-line name to reflect top-drawer excellence, so the team chose Primo for its etch and CVD systems. I'm not sure if Gerald and the team realized when they came up with the moniker that they'd stumbled into the surfer-slacker slang domain: "Wow, dude, those waves are really primo!" or "This is some primo stuff, man!" Hey, even dudes know quality when they see it.
So what does AMEC claim makes its Primo tools so primo? It starts with a proprietary, high-productivity, compact minibatch cluster system architecture, which the company claims offers at least a 35% cost of ownership benefit compared to the competition. The tools can be installed quicker, up and running faster---plugged in and playing in about a week compared to the usual month or so. And their modular design makes them easy to maintain, the story goes. Gerald talked about AMEC's analysis of the evolution of equipment technology and how the Primo tools are third-generation systems that have incorporated the best of the previous two generations.
In the case of the architecture, which started back in the day with bell-jar batches of wafers and evolved to single-wafer clusters, "we believe that third generation seems to go back to batch, in a minibatch cluster. With how the cluster is set up, carrying a number of chambers, each chamber produces more than one wafer---two wafers or four wafers. In that case, you have parallel processing to generate much more wafers with a limited capital investment. So you improve productivity and reduce cost. That's the clear trend we believe in."
Gerald delineated some other advantages of the Primo platform. "In the technology development mode, you can throw one wafer on the left (station) or one wafer on the right (station), and just process one wafer. The result is exactly the same as if you threw two wafers in. There are RF systems and endpoint separate controls, sharing a lot of components too. But this is the first time in the industry that you can do that flexible processing. You don't need to requalify the process…. People argue that if you have two stages, how do you make sure those two stages are the same? But my argument is that if you have separate chambers, you still have matching issues between the two chambers, with more variables."
Schematic of AMEC's Primo D-RIE tool, with twin stations.
Gerald ran through reams of data and explanations supporting AMEC's equipment-performance claims. As for the Primo decoupled reactive ion etch system (that's "D-RIE," not to be confused with the "DRIE" tools favored by the MEMS and 3-D crowds), the bells and whistles include a fast-tuning plasma capability and dual-plasma confinement; a very-high-purity, erosion- and contamination-resistant chamber material (developed by AMEC with no less than six key suppliers); critical dimension control that's already hitting sub-3-nm levels; a bottom-powered, high-frequency source; and direct-resistive top electrode heating.
One of the most innovative features is the D-RIE's flexible process regime, where four combinations of density and energy can be applied. Need some very-high-aspect ratios (up to 24:1) or FSG vias? Use the high-density, high-energy setting. For dual-damascene low-k dielectrics, you can twiddle the knobs to a low/low setting. To strip that low-k, dial in high/low.
"Some 70-80 different dielectric etch applications can be covered," Gerald claimed. As an example, he cited a very-low-k trench etch on top-level dual-damascene with FSG: "Without a stopping layer, we can control the corner, with no faceting and no CD infringement. We can tweak the density and energy to get a perfect result." One AMEC data set relates how, after a 3000-wafer marathon, the average etch rate was found to be off by less than a nanometer, just a few angstroms of variability difference.
The other member of the Primo family---the plasma-assisted thermal HPCVD tool---has some special features of its own to address shallow trench isolation (STI) and premetal dielectric (PMD) apps, as Gerald explained. As part of its multichannel distribution technology, a specially designed showerhead and multichamber gas delivery subsystem deliver excellent gap-fill capability (which comes in handy for STI), high deposition rate, and low particulation. A dynamic wafer heater stage offers best-in-class edge-to-edge and center-to-edge wafer uniformity, while the injector reduces pesky particles and facilitates a high liquid flow rate. The data shown for within-wafer uniformity, particle performance, and subsystem reliability are very competitive, with early indications that both systems have good 32-nm extendability potential.
The initial beta customer has made suggestions related to serviceability and software functionality of the AMEC twins. The company has also run hundreds of thousands of wafers, both its own and those from at least 10 different "customers," according to Gerald.
Once the launch takes off, the betas are humming, the first revenue-generating orders come in, and the initial wave of tools are installed, what's next for AMEC? Gerald told me that the next phase will focus on advanced versions of the Primo family, for the 32- and 22-nm process generations. Already not one to stand pat, the young company also has a new chamber concept in the R&D pipeline.
But for now, the focus at AMEC remains the successful launch of its tool line and subsequent production ramp when the first orders come in, all the while knowing that Applied and the other equipment big boys won't stand idly by and play nice as the new upstart competitor tries to move in on their market turf.
|
Blogs 
