Launched in June, the Oasis Clean is a modular dry-in/dry-out system that can clean more than 100 wafers per hour, Applied says. The tool maker asserts a slew of technological and productivity benefits for the tool. The single-wafer system features a megasonic tank that removes particles <0.14 µm without damaging the substrate. Low oxide/silicon loss is an additional technological plus claimed for the wet bench. As for productivity, the four-chamber system reduces cycle time and WIP wafers, Applied says. As befits its name, the Oasis lowers DI water use by a factor of 10, according to the company.

Conversations with customers over a two-year period prompted Applied to begin seriously considering the prospect of getting involved in one of the industry's most venerable equipment segments, says Satheesh Kuppurao, who oversees technical marketing and strategic technology for the company's wet clean division, transistor and capacitor group. The equipment manufacturer began a program at the beginning of 2000 during talks about RTP and LPCVD equipment, Kuppurao says.

"Increasingly, we were getting into gate oxide discussions with customers, all mentioning cleans being a very big part of the front end of the line," he recalls. "That's how the program got started. We didn't want to enter this market, actually. We just wanted to get OEM parts but integrate them with our tools. There simply wasn't any single-wafer clean out there for the front end of the line." Applied has been working with a joint development partner in Europe in developing the Oasis. The company has since begun collaborating with a Taiwanese partner.

Dainippon Screen, the Japanese equipment manufacturer, sells a single-wafer system that it developed with Hitachi, but the Applied marketing executive claims the system is not as "strong" in FEOL processes as it is in BEOL processes. Trecenti Technologies, a wafer foundry, is the first single-wafer 300-mm fab to use the system, notes Kuppurao. "They saw the benefit of cycle-time reduction, being a foundry. They could actually turn around chips in one-third the time."

Dainippon leads the overall wet-bench market segment with a share of approximately 35%, according to Dean Freeman, a principal analyst with Gartner Dataquest. The number of wet-bench players is fairly extensive and includes suppliers such as SEZ, TEL, Semitool, Verteq, Akrion, and Mattson. "Dainippon has a [big] market and didn't want to scavenge its market, with a single-wafer system," Kuppurao asserts.

FRESH WATER: The Oasis is Applied Materials' entry into the FEOL wet-clean market.

Revenues for 300-mm wet cleaning tools are forecast to rise from $671 million in 2003 to approximately $1.4 billion in 2006, Applied says, citing its own internal forecasts as well as market data from Gartner Dataquest and VLSI Research. Those same data show 300-mm single-wafer benches capturing more than 50% of the wet clean market by 2004. In 2001 that figure was approximately 20%. With the migration to 300-mm fabs picking up pace, the executive says Applied expects 2004 to offer "a huge growth opportunity."

One of the primary reasons is that the market for single-wafer cleans "has not taken off" at the front end, Kuppurao insists. "It just hasn't matched batch cleaning in quality and productivity. That gives us a good entry point." He notes that BEOL cleaning "is not a market we want to be in, because the chemical producers make all the money."

The need for FEOL wet cleans is "kind of like a cell phone," the technical marketing head suggests. "When you don't have a cell phone, you don't see the necessity for it. When you have it, you wonder how you ever got by without it."

As chipmakers begin ramping up 130-nm processes, challenges that had remained hidden have begun to surface, Kuppurao says. Problems for FEOL wet cleans at this technology node encompass capacitors, contacts, and gates for DRAMs. In logic processes the issues involve ultrashallow junctions, silicide, and postgate etch.

Of all wet clean applications, Applied says approximately 60% of the market is in FEOL prethermal and poststripping cleans. The current prethermal cleaning sequence involves three steps: oxide removal followed by the traditional two-step RCA cleaning sequence to first remove organics and particles, then metals. The Oasis sequence shortens the cleaning recipe. The oxide-removal step is followed by a single-step clean that removes organics, particles, and metals, Kuppurao says.

After copper/low-k dielectrics, "the next big challenge is transistor scaling," he asserts, because 1% of the thickness poses 60% of the challenges. "We'd been in the gate oxide and front-end market for a long time promoting single-wafer benefits against batch furnaces." Your typical wet bench takes 64 minutes to complete its run. Kuppurao claims the Oasis takes approximately 2 minutes to do the same.

The depth of an ultrashallow junction is 30–35 nm, he says. Typically, a wet bench can remove approximately 3 nm of the junction during each batch clean. "If you take this junction through five cleans you would lose almost half the junction. To compensate for that, the customer would have to do some other work [on the wafer]. After awhile you can't afford to lose those materials, like dopants."

With its claims for the Oasis's high throughput and small footprint, Applied believes the industry will shortly begin clamoring for the system. The bench promises profitability at a time when fabs have been hit with "a double whammy: 300-mm wafers and smaller devices," Kuppurao says. The biggest drawback with single-wafer cleaning has always been matching batch productivity, he points out. "If they can turn chips faster in the market and invest less to begin with, profits go up."

The dry-in/dry-out feature is a major reason for the high throughput of the four-chamber system, he points out. One wafer from the loading-port dry goes into the wet clean chamber and, dried in the same chamber, comes out dry. The tool's robotics always pick up a spin-dried wafer, never a wet one. This capability "basically allows us to get more than 100 wafers per hour" from the system.

Freeman of Gartner Dataquest thinks the throughput numbers can boost Applied's chances with the Oasis. "If [100 wafer-per-hour throughput] is indeed the case, they then start to get cost-competitive with some of the automated wet benches. Then you have the additional factor that your day-to-day operating expenses might be lower than those of a wet bench."

"It appears to be a fairly well-thought-out idea," he continues. "If single-wafer cleaning is going to take off, this tool has the potential to at least get that market segment kick-started in a big way."

Other wet-bench vendors have not been standing still. In June, SEZ purchased L-Tech, a manufacturer of drying equipment for batch wafer processes. L-Tech's systems use IPA for drying wafers. Like Applied, SEZ notes that wet processes are cost-effective when used with 25- and 50-wafer batches or larger. The trend toward single-wafer cleaning prompted SEZ to buy L-Tech, which is also developing several single-wafer cleaning and drying technologies.

The Austria-based vendor had been "playing in the back end of line with spray processes, and they realize there is a huge market in the front end of the line with automated wet benches," Freeman concludes. The acquisition will enable SEZ to "compete across the board with the likes of Dainippon Screen and TEL."

Add Applied to that list. Kuppurao says a popular "misimpression" holds that Applied has no background in wet processes. But, he points out, the equipment manufacturer has been making CMP tools for five years. "A little-known fact is we've had a 'skunk works' program in wet cleaning for a long, long time. Actually, even 10 years back we had people doing wet chemistry."

The division marketing and technology head expects to see Oasis orders coming in over the next six months from logic-device makers and foundries. Indeed, the company already "has systems in the booking process right now."

The rate of market penetration will depend, he says, "on how quickly 300-mm fabs will come onboard. Current plans call for his division to focus only on those customers that can provide the supplier with technical feedback. This list includes manufacturers of ASICs, microprocessors, and DRAMs, as well as foundry companies.