Issue Date: August 3, 2009
In 1999, researchers at one of Air Liquide’s customer companies approached the French industrial gases giant for help with a problem: They needed a nonchlorinated alternative to dichlorosilane, a precursor used to deposit silicon nitride films during semiconductor manufacturing.
What started as an effort to answer that simple request is now a major business for Air Liquide. The company’s 10-year-old advanced precursor unit has millions of dollars in annual sales and, until this year’s downturn, was doubling in size every year.
Jean-Marc Girard and Olivier Letessier were two of the Air Liquide chemists who worked with the customer, Toshiba, to solve the problem. Along with Ashutosh Misra, another former research colleague, Girard and Letessier are now executives running the business. Their charge is to continue its rapid growth in the face of stiff competition and customer demands that make that first request seem easy.
Throughout the world, semiconductor fabricators are seeking ultrathin films for performing critical tasks in their computer chips. Low-dielectric-constant films insulate circuit wires from each other, and high-dielectric-constant films help form transistors in advanced logic chips and capacitors in memory chips. Other films create barriers between chip layers.
The films must stay thin and uniform over the hills and valleys of a silicon wafer’s surface. Increasingly, the best way to apply them is with techniques known as chemical vapor deposition and atomic layer deposition. In both cases, precursor molecules are added to a deposition chamber, where they are heated and vaporized. When the molecules land on the wafer’s surface, ligands fall off, leaving the desired film.
Toshiba was after a precursor that it could react with ammonia to create a silicon nitride barrier. Air Liquide’s solution was trisilylamine. At first, it supplied the compound on an R&D basis, but interest grew. “Month after month we got requests for other applications,” recalls Girard, who is now vice president for technology at Air Liquide Electronics.
Other opportunities followed, yet for several years the business was an “internal start-up” that Air Liquide mostly left alone. Girard and his colleagues named it Aloha, figuring they would eventually think of clever words to form an acronym.
The words didn’t come and, at first, neither did success. Air Liquide is expert in handling bulk specialty gases such as ammonia, nitrogen trifluoride, and silane that are used by the electronics industry, but trisilylamine is a liquid, and many other advanced precursors are solids. Most require multistep syntheses. “The business wasn’t born in a company with a strong chemical culture,” Girard acknowledges.
Still, by 2005, the business had progressed enough on its own that top Air Liquide executives took notice. Liking what they saw, they stepped up the allocation of resources to create a global infrastructure that today includes labs and plants in the U.S., France, and Japan.
Mike Corbett, managing director of Linx Consulting, a market research firm that serves the electronic materials sector, confirms that Air Liquide has grown to become a leader in the precursor market. In advanced precursors that are turned into films via atomic layer deposition, Air Liquide has a 21% market share, Corbett estimates, second only to South Korea’s UP Chemical. And in precursors for low-dielectric-constant insulating films, Air Liquide’s 32% share puts it in the number two spot, just below Air Products & Chemicals, according to Linx figures.
Misra, who is director of the Aloha business, attributes the firm’s success to heavy spending on R&D from 2004 to 2006. “It was an extremely critical time in terms of customers deciding on the materials they would use,” he says. “Companies that weren’t dedicating the resources at that time have missed the boat.”
The Aloha business employs 40 chemists and materials scientists across its three laboratories, Misra says. Unlike researchers at chemical companies who may have multiple duties, “these are true Aloha people,” he boasts. “They are doing nothing but precursor development.”
That R&D has paid off in a number of product hits. One is hexachlorodisilane (HCDS), which, like trisilylamine, is used to deposit silicon nitride. Whereas others synthesize HCDS by reacting chlorine with silicon alloys, Air Liquide uses the disproportionation of silane to yield a high-purity product. The company also has worked with Selete, a Japanese semiconductor research consortium, on special additives that improve the performance of HCDS at low temperatures.
“We supply the world,” is how Girard describes Air Liquide’s market share for this molecule.
Corbett says Air Liquide also has a winner with ZyALD, a zirconium-containing precursor used to deposit high-dielectric-constant zirconium oxide films for memory chip capacitors. Memory chips with 65-nm circuit lines generally rely on a precursor known as tetrakis(ethylmethylamino)zirconium. But for the upcoming 50-nm generation of chips, Corbett says, companies want a precursor with improved thermal stability that can better stand high processing temperatures.
Indeed, Girard says Air Liquide scientists came up with the proprietary ZyALD formula after a customer described its need for heat stability. “The opportunity to develop a molecule only comes if the customer tells you what the problem is,” he observes.
Air Liquide launched its latest product in January. Called ToRuS and developed in cooperation with South Korea’s Seoul National University, it’s a ruthenium tetroxide-based precursor used to deposit ruthenium films. The company is expanding its plant in Fremont, Calif., to meet demand for ToRuS.
Corbett has his doubts about the widespread adoption of ruthenium in the next generation of logic chip, given that precursors based on the metal can cost upward of $10,000 per kg. Misra counters that ToRuS is already being used in hard drives and that its high surface reactivity translates into a reasonable cost per silicon wafer.
Keeping costs down is a growing challenge for precursor suppliers, notes Geoff Irvine, vice president of business development at SAFC Hitech, a unit of Sigma-Aldrich that competes with Air Liquide. While the number of materials being integrated into electronic devices is going up, Irvine points out, the amount of each material being consumed is shrinking.
The result is that process chemistry and chemical manufacturing know-how are increasingly critical to making money in the business. “We actually have manufacturing assets, and a number of our competitors in these materials do not,” Irvine says.
Economics aside, the Air Liquide executives see many more opportunities for advanced precursors as electronics firms shrink their devices further. After zirconium, memory chip designers will be looking for strontium- and barium-based compounds that offer even higher dielectric constants, Misra says. And companies want to extend photolithography with double-patterning techniques that require thin sacrificial films based on specialized silicon-containing precursors.
Such new applications have helped the Aloha business double in size every year since 2005, according to Misra. This year, growth has been flat compared with 2008—not bad during a recession—and he predicts that sales will increase by 30% or more annually in subsequent years. By 2015, advanced precursors will be a $1 billion global market, Misra predicts.
Likewise, Corbett foresees high growth, although it might be less robust than Air Liquide’s projections. Moreover, given the inevitable changes in chemistry as chips evolve, he cautions that it’s harder to predict which companies will capture that growth. “If they win at 50 nm, that’s great,” he says of Air Liquide, “but they must go through the same battle again at 40 nm and 30 nm.”
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society