Held February 13 and 14, the sixth annual review meeting of the National Science Foundation/Semiconductor Research Corp. Engineering Research Center (ERC) for Environmentally Benign Semiconductor Manufacturing took stock of its accomplishments since its founding in 1996 and provided a vehicle for researchers from a range of universities and companies to discuss their findings as well as their long- and short-term objectives.

ERC's impressive growth is a measure of academia's interest in lessening the harmful and wasteful impact of semiconductor manufacturing processes on the environment, and industry's interest in tapping academia's brains and technical know-how. While the center's founders included the University of Arizona (its flagship institute), the Massachusetts Institute of Technology (MIT), California's Stanford University, and UC Berkeley, today it also encompasses Cornell University, Arizona State University, MIT's Lincoln Laboratory, and the University of Maryland. Six years ago, the center had 20 faculty members working in 7 disciplines and organized around 3 research thrusts. Today, it has 30 faculty members, 11 disciplines, and 4 research thrusts (BEOL and FEOL processes, factory integration, and patterning). The ERC's budget has grown accordingly—from $2,238,000 in 1996 to $3,770,000 today. Not surprisingly, the center's nearly 50 industry affiliates include some of the biggest (and richest) players in North America, Europe, and Asia.

In his keynote address, ERC director Farhang Shadman highlighted the center's expansion as well as its shift in focus over the years: There is now less emphasis on novel environmental, health, and safety (EHS) solutions to existing processes and materials, and more emphasis on alternative processes and materials. At the same time, the center stresses the importance of enabling the fundamentals of ESH science and technology. Shadman, professor of chemical and environmental engineering at the University of Arizona, noted that ERC's format is to be a forum for technological exchange.

The semiconductor industry faces three challenges—the "axis of evils," in Shadman's words. These challenges—performance obstacles (such as defect density and yield loss), cost, and EHS impact—can be visualized as the three corners of a triangle threatening to expand ever outward under the pressures of process needs, business realities, and environmental concerns. The bloated area of the triangle represents the manufacturing burden that must be minimized, no mean task when you consider that the three points of the triangle symbolize seemingly mutually exclusive appetites.

The annual conference offered a rich assortment of papers, presentations, and posters on an array of topics. A presentation on dielectric etch by Ritwik Chatterjee, Ajay Somani, and Rafael Reif from MIT's Microsystems Technology Laboratories pointed out that the semiconductor industry is a relatively small but fast-growing contributor to greenhouse-gas emissions. The use of compounds with atmospheric lifetimes of thousands of years—such as hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and SF₆—has nearly doubled since 1990, posing a potent global-warming potential. After characterizing emissions from conventional C₄F₈-based low-k films on four different tool sets and three different film stacks, the authors concluded that alternative films based on C₄F₆ can reduce global-warming emissions substantially.

For example, the study demonstrated that emissions from organosilicate glass low-k etch on an Applied Materials eMax etch tool decreased 65.1%, from 0.339 kgCE to 0.119 kgCE, when the novel compound was used. Translating their findings into the mundane language of highway travel, the authors showed that a 35% emissions reduction from a 25-wafer dual-damascene run is equivalent to the lowered emissions that would be achieved by driving a four-cylinder Honda Accord 379 miles instead of 1081.

Several papers and posters in the areas of front-end processes and patterning focused on the use of supercritical CO₂ to reduce wet-chemistry consumption. "Supercritical fluids are a very hot topic just now," exclaimed University of Arizona professor Anthony Muscat in a presentation on surface preparation. "What could be more benign than scCO₂?"

While scCO₂ may not be a magic bullet, its advantages caused more than one conference participant to sit up and pay attention. For one thing, its proponents claim it is nonflammable, nontoxic, and nonaqueous. It has the solvating ability of a liquid but the mass transfer properties of gas. Furthermore, it is reusable and inexpensive. Muscat's test results indicated that scCO₂, used in conjunction with very small amounts of chlorine, removed copper from the wafer surface. Although chlorine is on the environmental hit list, Muscat remarked that benign compounds, such as water or O₂ molecules, can take the place of chlorine to oxidize copper.

'The IC industry faces three challenges: performance obstacles, cost, and EHS impact.' — Farhang Shadman, University of Arizona

In "Emerging scCO₂ Applications," Kevin Albaugh from Praxair Semiconductor Materials joined the academicians in praising scCO₂. The supercritical fluid can minimize or even replace the use of hazardous, regulated solvents while reducing water consumption and noxious emissions, he claimed. Industrial CO₂ has a minimal impact on atmospheric CO₂ and is largely recyclable. In its supercritical form, it can be used for wafer cleaning, photoresist residue removal, chemical fluid deposition of thin films, and as developer in 157-nm lithography processes.

Rounding out the program were several papers containing hair-raising statistics on materials waste in the semiconductor industry. B. (Pierre) T. Khuri-Yakub from Stanford University pointed out that a gallon of DUV photoresist can cost as much as $5000 and that waste from a photoresist spin-on process can exceed 98%. He received oohs and ahs from the audience with his flex-tensional ejector model, a transducer built into the silicon that can deposit precise amounts of liquids onto the wafer surface. Having demonstrated the workability of the concept with a large-scale prototype, Khuri-Yakub is fabricating a micromachined ejector containing multiple nozzles with a single driver and is designing one containing multiple nozzles with multiple drivers.

CMP, the dirtiest of all semiconductor processes, is also one of the most wasteful. In "Waste Minimization through Tribological and Fluid Dynamics Characterization," Arizona's Ari Philipossian showed that 99% of the slurry used in CMP goes down the drain because of the physics of fluid dynamics. Under the direction of Philipossian and MIT's Rafael Reif, the ERC is exploring several novel technologies to lessen CMP waste.

One approach includes the development of sensors for CMP modeling and control, thermal modeling, and the modeling of abrasive-free polishing. Other approaches strive to reduce slurry waste by improving CMP pad technology. For example, Philipossian and his team are attempting to eliminate the need for particle-containing slurry by developing fixed abrasive pads that will release particles during planarization.

"Getting particles out of the slurry and putting them into the pad," commented Philipossian, will make it possible to generate "slurry on demand," drastically cutting the wasteful squandering of unused slurry typical of current CMP processes. Another potential breakthrough is the so-called self-refreshing pad, which contains solid particles that dissolve in contact with water so that a fresh pore surface can be generated constantly during polishing. "You can actually hear the pad pop when the top layer of particles pops off," Philipossian said. "Thus, the pad constantly replenishes itself."

The ERC's stated mission is to "create and develop the science, technology, and educational methods that will lead to future semiconductor manufacturing facilities with minimal consumables and minimal emission of environmentally harmful, unsafe, and unhealthy waste materials." A noble endeavor, but how does it square with the needs of the manufacturing community, whose prime objective is safeguarding the bottom line? Shadman explains that the center does not seek to address all aspects of environmental protection, but concentrates on those that involve cost savings. "We have to compromise between cost and performance."

The manufacturing sector backs the center's work in myriad ways and profits from its research. Together with Texas Instruments, Motorola, Litmas, and International Sematech, ERC developed a commercial point-of-use plasma abatement scheme to reduce emissions of PFCs and HFCs. In the critical area of water use, a joint project between ERC and International Sematech developed an award-winning recycling program that has led to overall reductions in water consumption of between 25 and 70% at 10 fabs in the United States, Europe, and Asia. Pall's Barry Gotlinsky reported at the meeting that his company commercialized gas-purification technology that was developed jointly with Shadman's group.

Summarizing the center's strategy, Shadman stresses that the ERC's preferred modus operandi is to initiate a seed project and then whet companies' appetites to pursue further work. "It's a very subtle form of technology transfer," Shadman remarks. "We give them this fundamental information and let them take it from there."

—Bob Michaels

For more information on the ERC, visit www.erc.arizona.edu.