by Matt Wickenheiser, Contributing Editor
Advanced process materials and the continued adoption march of 300mm are driving materials growth in 2007, and fabs seem to be finally learning to make smarter purchasing decisions, according to top analysts tracking the market, interviewed by WaferNEWS. Potentially troubling signs seem to be offset by mitigating factors — yet there is still no short-term answer for stretched-to-the-limit polysilicon demand, particularly from the solar segment.
Semiconductor Equipment and Materials International (SEMI) is forecasting 7.1% growth in the worldwide fab materials market for 2007 (to $22.59 billion), a drop from the 16.9% growth of 2006 that reflects slower growth in chip unit demand, according to Dan Tracy, senior director of industry research and statistics (see table, above).
On the packaging materials side, SEMI said it sees 13% growth in 2007, with the market hitting $16.64 billion, up from $14.73 billion in 2006. Combined, that’s an overall materials market of $39.23 billion forecast for 2007, and a record $35.82 billion in 2008, SEMI noted (see table, below).
Looking ahead, SEMI expects a $25.58 billion fab materials market in 2008 (13.2% growth) and a $26.82 billion market in 2009 (4.9% growth). On the packaging side, SEMI forecasts $18.24 billion in 2008 (9.6% growth) and $20.46 billion in 2009 (12.2% growth).
Key growth drivers in ’07 on the fab materials side include 300mm and SOI, as well as advanced process materials, including 193 photoresists, anti-reflective coatings, CVD precursors, and CMP consumables, said Tracy. “Besides fab materials, many of these new devices are used in portable electronics and have resulted in a tremendous change in the types of materials used in packaging from new substrates, new wire technologies, new die attach materials and more.”
Techcet Group LLC also forecasts 8% growth in total semiconductor process materials in 2007, to $37.1 billion, as the segment shrugs off rumors of a market slowdown. “Word on street on the materials side is the fabs have learned their lessons from the past and have prevented themselves from double-ordering the last six to nine months,” Lita Shon-Roy, senior market analyst, told WaferNEWS. “Now, when wafer production is slower they don’t have to yank out orders from under suppliers’ feet.” Higher-tech materials such as CMP and high-k will continue to growth faster than other materials such as gases, wet chemicals, or quartz, she added.
Polysilicon demand still a short-term risk
Device buildups and a temporary decline in fab utilization rates could potentially cast a shadow on silicon wafer demand in the latter part of ’07, according to Takashi Ogawa, Gartner Dataquest’s semiconductor manufacturing research director. But the semiconductor industry’s emphasis on cautious operation management in recent years means that any market change by device inventory adjustment “will have a short-term, temporary influence over wafer demand,” especially between late 2006 and early 2007, he said.
Wafer demand will still be slow in 1Q07, Ogawa suggested, but under the most likely scenario, Gartner thinks that 300mm wafer demand will remain firm throughout 2008, though temporary shortages may be inevitable in the small-diameter market segments, as well as in some local markets and for some customers. Silicon will recover beginning in the second quarter because of a seasonal increase in production demand, he continued, and annual demand is expected to see 8% growth in 2007.
Red flags are still being waved over polysilicon availability in the near-term, with new silicon wafer fabs in the works but not expected to be online until 2009. “Although the latest forecast indicates that wafer demand will soften in early 2007, it by no means justifies us to ‘call off the alert’ in terms of a polysilicon shortage,” Ogawa said.
Karey Holland, Techcet’s senior managing partner, noted that there’s been high demand for polysilicon for solar energy use — “It’s just been sucking the stuff away.” The firm believes silicon consumption will experience moderate growth in 2007, between 7%-8%. SEMI’s Tracy noted that in addition to capacity expansions to help relieve pressures, solar cell makers are looking to develop alternative sources for polysilicon, such as metallurgical grade polysilicon and new suppliers.
One impact of polysilicon tightness on the semiconductor side that has not been heard about much is the effect on fab process tools, Tracy added. “Some equipment requires consumable components processed from high-purity silicon and the current market conditions have resulted in some supply challenges in this area,” he said.
New developments in CMP, CVD
Elsewhere, the silicon carbide (SiC) wafer market will see 20%-25% growth in 2007 thanks to increased use for LEDs by companies like Cree, as well as for discrete devices such as Shottkey barrier diodes, IGBTs, and power devices for RFID, said Shon-Roy. For high-brightness LEDs, indium gallium nitride (InGaN) is grown on SiC wafers to manufacture blue-white LEDS.
In the area of CMP, Techcet estimates slurry revenues of >$700 million in 2007, and pad revenues of >$500 million, with two main driving factors. As fabs move toward some type of selective polish for shallow trench isolation they are abandoning standard oxide polishes (at $12/gallon) in favor of selective polishes such as cerium oxide-based ones or silica with additives ($30-$50/gallon), and that bumps up the entire slurry market. Selective STI processes will allow fabs to transition from a multistep STI process to direct CMP of the STI oxide, Holland said. Whether eCMP will start taking a significant part of the slurry market as the industry moves down to 32nm processes is still up in the air, she noted.
The other change seen in the CMP market is increased competition, where there are currently 20 companies trying to take market share from big players Rodel and Cabot, said Holland.
In CVD low-k dielectrics, 3MS (trimethylsilane) is estimated to have over 50% market share, much attributed to Applied’s Black Diamond process, and end-user feedback suggests the transition to Black Diamond-2 is going to go well, according to Techcet’s Holland. Most end users are not using 3MS for BD2 and are not saying what the precursor is, she noted, adding that the industry would see about a 10% reduction in cost/gram of other precursors.
In the photoresist segment, there’s a lot of business in 90nm resist with “very, very high-added-value materials,” noted Shon-Roy. Techcet sees the worldwide photoresist market at $1.2 billion in 2007, increasing to about $1.3 billion in ’09. Costs will jump fivefold when fabs transition from KrF for 248nm for lithography to ArF for 193nm, and going to 14 layers of copper means engineers have to do litho on each layer, increasing use of the material, added Holland.
Recouping development costs still No.1 focus
Expanded use of increasingly complex and expensive materials brings up the No.1 concern of material suppliers — the cost-down pressures companies face, and the challenge of realizing a return on their investment for all the new technologies they are being asked develop and bring to the market, noted Tracy. “In this environment there are downward pricing pressures for both older and newer materials, and this affects decisions of how and where suppliers will spend money to innovate new material technologies,” he said.
He suggested that the demands for tailored materials made by newer devices and process technologies is a challenge for both fab and packaging material suppliers. The result is lower volumes of highly engineered materials are required by a given customer, he said — instead of the high volumes that have traditionally generated the payback to materials suppliers. These niche technologies can be less attractive to the larger suppliers who need a return good enough to justify the investment to their stockholders.
“A potential consequence is that more competition could emerge as this creates an opportunity for smaller, quicker players with lower overhead and new ideas to possibly introduce new materials,” said Tracy. M.W.