Abstract
In mid-1997, semiconductor suppliers and the companies that produce
chip-manufacturing equipment were set to implement building ICs on 300mm
(12-in.) wafers. A massive R&D effort, estimated at slightly more than $4
billion expended so far, has been under way for three or more years to develop
the technologies and equipment for the big wafers.
Copper and low-K helped push back the 300mm front. The new interconnect
technologies don' t just increase chip performance; they also help make higher
chip density possible. The more you improve density, the more ICs you make per
wafer, and the less you need a bigger wafer.
The move to copper is going to take longer because the industry expended so
much of its reserves on 300mm. Badly burned industry players are going to be
twice as cautious moving forward.
Overcapacity in the semiconductor industry moved 300mm processing back to the
forefront in 2000.
In a shift away from the recent trend toward outsourced semiconductor
production, Integrated Device Manufacturers (IDMs) appear to be returning to
their initial approach to chip making: using internal manufacturing to achieve
economies of scale.
This trend is manifesting itself in the number of 300mm fabs that will be
operated by IDMs vs. those owned by foundries. In 2002, 14 300mm facilities
were in production. An estimated 21 percent of those fabs were owned by
pure-play foundries. In 2004, the number of 300mm fabs reached 24, with 29
percent of them being owned by foundries.
However, by 2006 there were 47 300mm fabs in production, but only 19 percent
will be owned by pure-play foundries. Over the next two years, only one new
300mm fab owned by a pure-play foundry will be built. With their 300mm
capacity additions, IDMs will transition production from external foundries to
internal fabs.
The pure-play foundries will continue to serve the expanding fabless
semiconductor companies and serve as capacity buffers for the major IDMs.
However, the key question facing the pure-play foundry service providers is
how much demand will shift out of the foundries once internal 300mm capacity
becomes available at the IDMs.
New construction will be 300mm related, as DRAM fabs and foundries continue to
move to the larger wafers. But future planned 300mm fabs also show a
significant increase in the number of fabs devoted to logic and flash memory
products, with a corresponding reduction in the dominance of DRAM and foundry
applications.
All of this expansion most likely will leave the semiconductor industry
vulnerable to a new round of overcapacity and severe price declines.
The benefits of copper are well known, but the problems encountered by chip
manufacturers were underestimated when then began evaluating the technology.
In facts, initial enthusiasm turned to euphoria for the copper tool
manufacturers.
The most significant problems encountered with the copper dual damascene
process center around copper deposition and copper CMP, and equipment
suppliers are working to enhance the quality of their equipment, processes,
and materials.
CMP of copper is more complex than other metals because of the need to remove
the tantalum or tantalum nitride barrier layers and copper uniformly without
overpolishing any features. Copper' s physical properties add to the polish
difficulties. Unlike tungsten, it is a soft metal and subject to scratching
and embedded particles during polishing. If one uses a traditional one-step
process to planarize copper, dishing (overpolishing) of the copper results as
the pad reaches the much harder Ta or TaN barrier.
This report discusses the technological and economic impact on the
300mm/Cu/Low-K convergence. This report discusses the timing, trends, issues,
and market analysis of 300mm wafers/tools, copper deposition/etch/CMP, and
low-k materials. Markets are analyzed and projected.
View Cart
