Gas Management Confronts Materials, Cost and Service Issues

Alexander E. Braun, Associate Editor -- Semiconductor International, 12/1/1998

"We're being pressured by customers to bring prices down," said Noel Leeson, vice president, electronic materials at BOC Edwards (Murray Hill, N.J.). "This is complicated by the DRAM market and the acceleration in design rule shrink." He added that customers are speeding down the technology road faster than anticipated.

As design rule shrinks continue, some materials or ways devices are made will not work anymore. This is pushing gas management companies to newer materials sooner than expected. "Everyone knew copper was coming; it was just a question of when," Leeson pointed out.

Gas companies involved in metalization have focused on tungsten, and not particularly on aluminum, PVD rather than CVD. Copper presents both an opportunity and threat depending on which metal layers get replaced. If it is tungsten, that will affect gas companies' business. If it is aluminum, there are opportunities for these suppliers, because the copper material is relatively new and unformed.

Fig. 1. Gas management companies are meeting the challenge of providing higher quality for less by improving means of production and entering into strategic alliances with chemical management companies to provide customers with consolidated, on-site, expertise and service teams. (Source: Matheson)

Presently, it seems plating will be the method of choice for a while. It is certainly the way bulk copper plating depositions will be done, at least in the first generation of copper devices, with CVD coming afterward. With the plating chemicals sector in the hands of a number of small companies, rather than established semiconductor suppliers, another opportunity for materials suppliers is opened.

COO ­ The Magic Word

Customer COO is a vital issue for suppliers. "Typically, when you look at a process that uses gas, the gas' actual cost is a small part of the process cost," Leeson said. "That's why in-situ service is important; it helps to understand the impact of gas cost. Will a higher-purity gas cut down on the use of things like monitor wafers, which are costlier than the gas' simple molecular cost?"

Leeson believes that understanding this at a level where it becomes possible to have a sensible dialogue with the customer is important. "Customers are uncomfortable talking about yield," he said. "It is difficult to have such a dialogue, but the more we understand, the better we can react to things that could cause problems."

Some yield impactors are the effect of particles and moisture levels in gas flow, as device rules shrink. Particles that before could be ignored are having an effect, and reducing moisture levels by an order of magnitude becomes more important, because the detrimental effect to processing is magnified with tighter geometries.

These are areas where BOC Edwards and others have ongoing R&D programs: "How do you reduce by a factor of 10 the moisture in a corrosive gas? How do you go from 0.01-0.001 µm particle detection?" Leeson asked. "Because these factors impact yield, we must be able to deliver gases that meet these demands."

In the particle area, most suppliers recognize a need to change, not fine-tune, current processes. Leeson put it this way, "We need a better understanding of what causes particles to adhere inside cylinders and how to get them out (Fig. 2). We're working on approaches that are different to what is being used now in today's production plants with good results. Soon we should have an improved process that will be well characterized."

Fig. 2. The reduction of moisture in corrosive gases, as well as the detection and elimination of particles as small as 0.001 µm that may adhere inside cylinders are a few of the challenges being faced by the R&D efforts of gas management companies. (Source: BOC Edwards)

A major problem facing suppliers is not technology, but the customer's unwillingness to foot the bill for these new requirements. As Leeson sees it, "Finding a way to pay for this development work and improved production processes while bringing prices down poses an interesting conundrum." One way is building high-efficiency production facilities that reduce production unit costs. Because the initial costs of constructing new plants are nothing short of staggering, only major players seem capable of it.

Michael Pfarr, vice president of marketing for the Matheson Electronic Product Group (San Jose, Calif.) agrees with Leeson's comment: "Cost has always been a consideration, but is higher than ever on the list. Customers demand materials and services that satisfy their requirements with the lowest cost option. It comes at us from two directions. One is the customer's requirements in a very competitive market, in which he's trying to reduce costs as much as possible. The other is that, as fabs close, we as suppliers have fewer opportunities and must be as competitive as possible. Everybody is sharpening their pencil!"

Consolidation and Expertise

Cost reduction goes beyond price reduction. Reducing the number of suppliers also cuts operations expenses. This is a principal reason why chemical and gas companies ­ Air Products and Allied Signal, and Matheson and Olin for example ­ are establishing alliances not only to provide technical and operational advantages with one team handling both gases and chemicals, but also operating cost reductions for the customer with one team doing the work of two.

The marketplace demands that suppliers, rather than the user, provide expertise beyond the fab's core competency. Customers want the resources required to run a fab, whether chemicals and gases or equipment and experts, bundled together. This means suppliers must restructure to optimize operation, improve processes and procedures, as well as concentrate on where costs, overhead and materials are generated. If a distillation column is efficient, for example, it may make sense to go to cheaper, lower-grade raw material and distill it into high grade material.

It certainly requires a global outlook and capability in the search for raw material sources and having processing plants around the world, rather than centralized in one location, to enhance local supplies and services and reduce shipping costs. The gas management industry is meeting what would be an otherwise impossible situation ­ giving more for less ­ by modularizing its operation on a worldwide scale and developing and using better technology.

As Pfarr put it, "As we move to the next level of analytical capability of moisture in ammonia, for example, our R&D people must develop a process capable of analyzing at a level never achieved before, and move it into operations." Companies are not only developing advanced capabilities, but ensuring they are cost-efficient. "We have produced tungsten hexafluoride for years," Pfarr said, "but we're investigating how to improve the production stream's efficiency, to allow us to either produce it at a lower cost, or produce more of it at the same cost." The average gas company spends 3% to 5% in R&D. Demands imposed by copper, dual damascene and CMP will force additional R&D expenditures to meet users' needs.

Some of the new processes require solutions other than conventional high-pressure gas sources. Some CVD precursors are liquids requiring a bubbler instead of a high-pressure cylinder. Pfarr does not see this as insurmountable. "For example, trichlorosilane has always been delivered in a drum, the aluminum CVD precursor has always been a liquid. As we move into exotic chemicals for some of the low- and high-k dielectrics, we're expanding from conventional gas distribution to chemical distribution, which ends up as gas when it gets to the tool but doesn't start out like that."

The marketplace is swinging from customers telling suppliers what they want the design to be, to telling the supplier what they want to do, letting him design and supply the requested capability. "Because we have hundreds of customers," Pfarr said, "and do a lot of work in the field and have a tremendous database to draw from, we learn faster than they can. When we do prototypes, we try all sorts of ideas, and we know what does not work."

EH&S Concerns

The gas industry is doing byproduct abatement by collecting and treating them. Dealing with these by-products is merely an extension of existing technology. Even toxics like ClF₃ and C₅F₈ can be treated easily, as can pyrophorics. The list is just getting longer. As one industry expert put it, "The new materials coming in make no difference. We've been dealing with these categories of chemicals ­ poisons, toxics, hazardous materials ­ for a long time. The new categories make no difference; what was difficult before is just as difficult; what was easier still is."

The main difference is that there will be more on site than before. Certainly with bulk specialty gases gaining favor, having a tube trailer with several hundred times more than a 12,000 gram silane cylinder contains produces unique problems for a catastrophic single release. A tube trailer with 21,000 lbs of hydrochloric acid creates a different scenario if there is a catastrophic release and all seven tubes empty at once, than just one cylinder releasing. From a facilities standpoint, release containment requirements will not be too different, just larger in scale.

Fig. 3. By outsourcing gas and chemical operations, chipmakers can tap into specialists knowledgeable in niche disciplines such as gas handling equipment. (Source: Praxair)

An industry expert who asked not to be identified indicated that it is frustrating for suppliers to convince customers to use a full COO model. "All Purchasing cares about is how big the check is. The purchasing agent doesn't consider the time that process engineers require to do test wafers, or uptime versus how many cylinders must be changed. The industry must agree on what COO models entail. Then it would be easier for us to show how a fab's cost of operations decreases even if the price of services and materials increases."

An example of this is safe delivery source (SDS) cylinders that go into an ion implanter. "On a per-gram basis they're more expensive than a high-pressure cylinder," stated the same expert. "But you don't change them so often, and they are at sub-atmospheric pressure, so they're safer. This minimizes downtime, false alarms and plant evacuations. What cost savings do you assign to improved safety; how do you factor the price of one life saved?"  

One-Stop Shopping

Steve Sidorchuk, product manager for Praxair (Danbury, Conn.), is not unnerved by the increase in materials. "Since 1995 the number of specialty gases we offer has grown fourfold and will continue to increase. We're ready."

Sidorchuk also views with equanimity the trend toward having the gas supplier own the whole system (Fig. 3). "We have a turnkey offering, whereby we take complete accountability for building the gas, chemical, toxic monitoring and abatement systems," he said.

Like some of its competitors, Praxair not only builds the systems and gets them up and running and certified but also provides services where its own people run the operation onsite. Sidorchuk predicts the trend will escalate. "In the future the customer will say, 'A tool supplier sent us the tool; you take care of installing and certifying it.'"

Fig. 4. Technician completing a container change at a customer's fab. Total gas and chemical management, provided at the fab by the supplier, is becoming the industry standard, rather than the exception. (Source: Air products)

Gerald Ermentrout, vice president and general manager of the Electronics Division of Air Products and Chemicals Inc. (Allentown, Pa.), sees the concept of total gas and chemical management as the standard model for the typical state-of-the-art fab (Fig. 4). "Virtually every fab built today in the U.S. and the majority of those in Europe, and increasingly those in Asia, use this concept. Today, if someone built a new fab and didn't look to an outside supplier to provide gas and chemical management services, it would be unusual."

Within a decade, Ermentrout believes this will extend beyond gases and chemicals. "Increasingly, these kinds of services will be outsourced, with the manufacturer just focusing on making circuits," he predicted. "All support activities: gas and chemical supply, things like the DI water supply and the running of other support systems in the fab will be outsourced."

Like its peers, Air Products is developing new materials for the copper arena and for low-k dielectrics, but Ermentrout believes the old standbys will remain. "Existing materials have significant lifetimes left," he said. "Many of the new materials seemed tied to 300 mm, and the shrinks are postponing it. Existing equipment and processes are being pushed to tighter and tighter technology, and some of the expected changeovers have not occurred."

Particle contamination is not a problem for bulk gases ­ particularly when generated onsite ­ with most facilities capable of consistently delivering them with parts-per-trillion impurities. This is more of an issue with specialty gases. "With spec gases, particularly as you look to 0.18, 0.15, 0.13 µm technology there's room for improvement," Ermentrout conceded. "We're all trying to improve purity levels. For some of those spec gases, such as HCl and the corrosive gases, the analytical equipment set is not where it needs to be to measure those low levels of impurity."

Ermentrout believes customers will stay with the one-stop shopping concept and buy from companies that meet their needs in the broadest number of areas for materials. "On the equipment side they are now looking for turnkey packages. Three years ago, most of our bids would be for a gas cabinet sent to the site where the customer or a contractor installed it," Ermentrout recalled. "Now, half of our requests for quotes are for turnkey options."

Gas delivery system component manufacturers, such as Unit Instruments (Yorba Linda, Calif.), are also gearing up for the new materials and COO challenges. According to David Sheriff, vice president and general manager of Unit's Z-Bloc '99 Division, "The biggest challenge we're facing is reducing cost of the entire foodchain that supplies gas systems. Right now you have a component manufacturers, a distributor, an integrator, system designers, weld shops, integrators and so forth. There are too many players and not enough value added at each level," Sheriff said. "It's a sign of the times that distributors are starting to integrate systems, and system integrators are dealing directly with component manufacturers. In our case," Sheriff pointed out, "we have an agreement to merge with USF Kinetics. They are creating a larger, more vertically integrated organization to improve efficiency."

Like his counterparts in the gas industry, Sheriff believes gas system design ought to be left to experts. "It's a sign that something's wrong when everyone wants to tweak the design of their system and specify their favorite brand of components." Sheriff pointed out that the difference in performance is undetectable and designing one's own system variation skyrockets costs. "Many gas systems can be radically simplified if you don't think of them as a collection of individual components. Most costs are designed in from the start. It's time to treat the tool gas system and its electronics controls as a 'black box,'" he said, adding that outsourcing the basic design to experts who fully understand how the elements function and interact is the industry's biggest opportunity to affect positively gas system cost and performance.

Looking Forward

It seems certain that for the next three years there will be a continuing evolution of materials; high- and low-k dielectrics, etc. and copper will grow in importance. Gas, chemicals, and materials development will continue responding to the needs of increasingly smaller architectures. Strategic alliances, teaming and consolidation between companies (and end users) will probably not only continue, but expand. Even when we are finally out of the current slump, cost pressures are unlikely to let up.

There is much to be done in the ES&H area. While most gas and chemical suppliers continue working on how to best deal with used supplies and expand recovery options, few of their customers are comfortable with the thought of taking a gas, treating it and putting it back into their process. The gas (and chemical) management industry will have to deal successfully with this prejudice before recovery on a truly major scale becomes mainstream.

But the opportunities are endless.

Cost of Ownership of Gas Distribution Systems Helen Armer Aeroquip Corp., Ann Arbor, Mich. The high cost of gas distribution equipment is being examined. Gas distribution equipment in our industry costs several times its industrial counterparts. Cost is partly driven by contamination control requirements that mandate using special designs (e.g., minimum dead volume, corrosion-resistant materials) and clean manufacturing processes (e.g., electropolishing, precision cleaning, cleanroom assembly). We applied cost of ownership modeling to gas distribution equipment and improved our understanding of how costs can be lowered. Our model calculated COO for the building block of a gas distribution system, a gas stick. The gas stick modeled consisted of three valves, a pressure regulator and transducer, a particle filter and a mass flow controller. COO was calculated using the metrics defined in SEMI E35-95A. Input data came from the I300I Equipment Performance Metrics and a gas box maintenance procedure from a leading gas box assembler. The I300I metrics represent a 300 mm fab line running 0.25 µm (250 nm) design rule. When maintenance has to be performed on a gas box, the process tool must be shut down. Hence, our model included the opportunity cost of lost production associated with maintenance. We computed COO for various tools, including poly etch, oxide etch, tungsten CVD, thermal diffusion and MeV ion implant. COO for a five-year stick life ranged from 3.6 to 9.6X the cost of the components. Stick purchase price accounts for 10% to 20% of COO. Installation and prove-in costs account for another 10%. Fig. 1. Gas stick COO is largely determined by the production cost due to system failure and MFC life. Replacement of consumables, such as filters, accounts for about 10%. Cost to replace failed components can contribute greatly to COO, depending on the failure frequency. Hence, mean time between failure was varied in the model. Most of the remaining COO is opportunity cost of lost production. Fab production personnel who know the opportunity cost of downtime for a given process tool can calculate gas stick COO (Figure). See http://www.Semiconductor.Net for the complete article, "Applying Cost of Ownership Modeling to Gas Distribution Equipment." The analysis gave several insights into reducing COO. Reducing component price has a relatively small impact on COO (10% to 30%), although this impact increases as the component life increases. Improving component reliability has the greatest impact on COO. Increasing the mean time between failure of components in the stick from two to five years has a 200% impact on COO. There are several other things that can dramatically reduce COO. These include: reducing the time needed to purge after component replacement, providing easier access to the gas box and the components in it to reduce component replacement time, standardizing component dimensions to reduce installation time and developing new coupling designs that eliminate line twisting problems with existing face seal fittings.