The CMP slurry market will triple in size by 2005, even as engineers struggle with today's gritty demands: new formulations for lowering defectivity, dual damascene structures, low-k insulators, and next-generation problems, such as premetal dielectric polish for inlaid gates.

When all is said, the market for slurries will see 29% compound annual growth between 2000 and 2005, moving from $175 million this year to $520 million in 2005 (Fig. 1), according to market researcher Kline & Co., Little Falls, NJ. This echoes increases seen in the CMP tool market, which has averaged a 36% annual growth rate (see "CMP: Market trends and technology," Solid State Technology, June 2000, p. 67). At the same time, there are some alternatives to conventional CMP mounting challenges to the slurry market.

No doubt, there's more room for growth. Only 15-20% of all semiconductors today receive CMP treatment, said analysts from Kline, and the need for CMP rises as design rules shrink and the number of interconnect levels increase. "The slurry market has all the earmarks of an industry that is young and growing and just as confused as it should be with the onslaught of new technologies," said John Davis, manager of client services at Kline & Co.

Rick Foster, director of marketing for slurry supplier Cabot Microelectronics Corp., says his firm has its eye on 60 new fab projects around the globe, and notes, "all 60 of these fabs will be CMP fabs."

Kline & Co. estimates that the number of CMP steps will increase from roughly 88 million CMP operations/year in 2000 to 315 million in just five years (Fig. 2). The number and variety of applications will multiply, pressing the market for new slurry formulations. By 2005, dual damascene processes will drive metal planarizations to rival oxide CMP steps (Fig. 3). CMP of copper is a key driver, though tungsten and aluminum CMP will also see strength (Fig. 4). "Very few people are running copper; it's not a high volume manufacturing process," notes Kevin Witt, advanced products marketing director for Rodel. "I think by the year 2005 only 25% of the world's 8-in. equivalent starts will be copper." Conventional fumed silica dispersions, used primarily for oxide and tungsten CMP applications, will continue to grow and hold a majority share of the slurry market, accounting for 58% of the market this year, and 40% in 2005 (Fig. 1). Alternative colloidal formulations, however, will gain popularity, taking 34% of the market share by 2005, up from 18% this year. Likewise, the alumina-based slurry sector for metal planarizations will grow, but its overall share of the expanding market will dip a slight 5%. A band of new slurries, including ceria, manganese dioxide, and zirconia, will make up the 8% "other" grouping in 2005.

Cabot leads the pack
For all its commotion, the slurry market appears to be attractive to investors. Cabot Microelectronics, Aurora, IL, went public with an $80 million offering this year. While specific slurry market share numbers are debated, most agree Cabot holds the leading market share, followed by Rodel. The remaining market share is divided among EKC Technology, Fujimi, Bayer, Hitachi Chemical, and a growing list of newcomers, such as the new Planar Solutions slurry joint venture between Arch Chemicals and Wacker Silicones. In addition, there are some slurry developers that have not yet surfaced in the commercial market.

Kline's Davis expects the industry will see consolidation in the future, with some of the more established firms collaborating with these seedling companies. "Cabot and Rodel will continue to have strong, dominant positions," said Michael Corbett, Kline & Co. director. Going forward, Corbett and Davis expect to see more companies adopt an integrated portfolio in which slurry is but one component of an overall strategy.

Polished cheese?
As with any technology still in its youth, CMP— and by extension, slurry— has its share of challenges ahead. In its monthly Electronic Materials Report, materials research firm Rose Associates, Los Altos, CA, says, "the challenges, though, are seemingly never-ending. . . Each new application seems to require a new slurry formulation and different process conditions to achieve the desired selectivity and optimal removal rates."

Planarization of low-k materials poses one of the greatest challenges to the CMP community. Rose Associates likens low-k CMP to polishing a block of cheese with sandpaper. "Low-k insulators are much softer than the silicon dioxide films chipmakers are accustomed to processing. They encompass a whole new set of problems with respect to slurry formulation and process development," noted the Rose report. "The likely porous nature of very low-k films will cause entrapment of residual slurry particles, creating a cleaning nightmare."

In addition, 300mm wafer processing could lead to a reduction in the volume of consumables, say the Kline analysts, who point out that average slurry usage rises from 560 ml/wafer to 650 ml/wafer, which only marks a 20% increase for more than a 200% increase in polishing area. However, Cabot's Foster says a broader perspective is needed here. Despite the increased area/die, the transition from 200mm to 300mm wafers will not decrease the total number of wafers processed annually. Overall growth in the market will drive demand. "The number of wafers processed will be on the increase, and that's what drives consumables," Foster said.

As one of the strongest growth areas, the copper market is stirring activity among slurry suppliers (both existing and new); however, some chipmakers "continue to rely on internally formulated slurries (known affectionately as 'home brews'), for copper CMP, as they wait for acceptable commercial slurries. This captive versus commercial situation is a repeat of the scenario that occurred for tungsten CMP," said the researchers from Rose, adding that "chipmakers do not relish formulating their own slurries, but commercial copper slurries have either not delivered the specific performance need or are perceived to be too costly."

From lab to fab
Suppliers, not surprisingly, are moving quickly to provide solutions. Rose points out that at least 40 CMP-related patents have been issued in the US since the beginning of 1999, with Cabot and Rodel each receiving patents for advances in this area. In addition, a number of leading chipmakers have been issued CMP-related patents.

Recently, Cabot introduced a new dielectric slurry — dubbed Semi-Sperse D7300 — for advanced interlevel dielectric (ILD) applications, and says it offers a tenfold reduction in post-CMP microscratching over available slurries. The company also is addressing the hard disk market with the second generation Lustra 3010 ultra-super polishing slurry for rigid disk substrates — a key growth area for slurry suppliers. "A lot of change is taking place in that market," said Kline's Davis. "It's not as mature as people think."

The new Planar Solutions LLC joint venture in Adrian, MI, has developed Cu10K-2, a single component, second step copper CMP slurry that has been qualified on several polishing platforms. Cu10K-2 removes Cu, Ta and SiO₂ at approximately the same rate creating a uniform surface with minimal dishing and erosion.

Move over slurry . . .
As CMP progressively becomes less and less of a perceived black art, other exotic but slurriless solutions are arriving, threatening to cut into the conventional slurry market's growth.

While it's too early to know the full impact of alternative CMP technologies, the industry is taking all options into consideration. "It's an indicator that the industry is looking for some kind of alternative, that they're not completely happy with CMP," said Davis. "I think [slurry suppliers] would be foolish not to recognize it as a real threat to their business."

Perhaps most advanced of these alternative CMP approaches is the slurry-free approach championed by 3M and Rodel. The technology has been in development for at least six years; 3M worked closely with CMP startup Obsidian (now part of Applied Materials) in developing a fixed abrasive pad for Obsidian's web-based polisher, though the technology is now used more with conventional pads. The process, targeted at copper and shallow trench isolation applications, doesn't require slurry, but uses a polishing liquid (such as ammonia or potassium hydroxide), or a reactive liquid for copper CMP, says Rodel's Witt.

Each fixed-abrasive pad displaces about 14 gallons of slurry from the market, according to Kline estimates. While Witt believes that figure may be even higher, he is quick to point out that the process does require the companion polishing liquid in lieu of the slurry.

Hitachi Chemical is working on a nonabrasive wet chemical solution for copper CMP. Rose reports that the solution contains no grit, combines an oxidation agent, an etching agent, and a corrosion inhibitor, and can be disposed of with the same neutralization process as rinsing solutions. The formulation is said to reduce scratches and dishing to less than 20% of the usual.

Austria-based spin-etch firm SEZ and Honeywell (formerly AlliedSignal) are making progress with a spin-etch planarization method for copper dual-damascene interconnects. The method could potentially sidestep the challenges of using CMP in devices with advanced low-k films, which can be damaged by the mechanical downforce in CMP.

Also targeting this problem, ACM Research, Fremont, CA, introduced a stress-free polishing technology that doesn't require slurry or pads. ACM's system — the Ultra SFP — uses an electropolishing method with layer-by-layer atomic level control. The company is seeking strategic partners to commercialize the tool for the 0.1µm node.

Intel has received a patent for CMP of low-k materials, except this technology uses a magnetic slurry process and a magnetic coil to replace the mechanical downforce with a magnetically controlled force, according to Rose. The process is said to not use a polishing pad, and is defined as either a wet or dry solution.

In spite of these forward-looking technologies, slurry suppliers today face more mundane challenges — meeting the high growth demands of the "tried and true" oxide, tungsten, STI, and polysilicon CMP segments, notes Rodel's Witt. "One of the big opportunities right now is just keeping up."

The analysts at Kline & Co. are undertaking a new study of the CMP and materials market that will include a five-year forecast, profiles of the materials and tool suppliers, and cover emerging suppliers and alternative technologies. This study, entitled The Global Outlook for CMP Technology and Materials, 2000--2005, follows the group's 1998 analysis of the market. For more information, contact Michael Corbett at Kline & Co., ph 973-435-3457, e-mail: Mike_Corbett@klinegroup.com.

This article was based on data provided by Kline & Co., individual companies, and Rose Associates' August, 2000 Electronic Materials Report. The monthly report is available from Rose Associates, 4 Main Street, Los Altos, CA 94022-9990; ph 510-792-5566.