Designing a Specialty Gas System

Specialty gas system engineering involves the distribution of cylinder gases from the source to the manufacturing tool. The configuration of specialty gas systems varies significantly across fabs in the United States, and even more so with fabs overseas. These differences are due in part to the requirements of applicable national and local codes. They are also due in part to the perceptions and preferences of various stakeholders (operators, customers, suppliers, etc.).

Changing existing distribution concepts can be daunting. It involves first a clear understanding and interpretation of the codes, a significant challenge in itself. Further, it requires the integration of issues from numerous internal groups, including environmental health and safety, operations, process engineering, equipment engineering, and gas equipment and material suppliers.

Our experience with cost reductions in our existing Fab1 saved >$750,000 in capital and operating costs over several years. It also enabled us to achieve a 20% cost reduction in the specialty gas systems for our new Fab2. Changes from the original Fab1 design were implemented during the build out of Fab1, then incorporated into the Fab2 design.

The use of simplified inert gas valve manifolds (inert gas trees, IGTs) was implemented for a quarter of the total cost of the original valve manifold panels (VMPs). The original design used fully automated eight-stick VMPs with remote monitoring. The installed cost of a VMP was >$65,000. The revised design uses an eight-stick manifold of manual valves with manual purge capability (Fig. 1 ). The installed cost of these units is $15,000.

1. A simplified inert gas tree, as shown here with an eight-stick manifold of manual valves with manual purge capability, was implemented for a quarter of the total cost of the original valve manifold panels.

Successful implementation of this change involved working closely with manufacturing, operations and other personnel to ensure that necessary capabilities were not compromised, and to provide proper procedures and training.

The use of purge carts in place of fixed purge equipment saved >$250,000 in gas equipment for Fab2. Transitioning to purge carts involved the design and fabrication of a mobile, single-cylinder purge cart rather than fixed, dual-cylinder, automatic crossover racks (Fig. 2 ). Purge gas is used during commissioning and decommissioning of gas sticks, tools and/or gas equipment (i.e., valve manifold box, VMP and IGT). The cost of the cart is similar to a cabinet or rack, but significantly fewer units are required (e.g., four to six carts vs. 20+ fixed units).

2. The use of a single-cylinder purge cart (right) in place of fixed, dual-cylinder purge racks (left) saved >$250,000 in gas equipment for Fab2.

The choice of using individual purge connections at the gas equipment (VMB, VMP or IGT) or creating a purge header to manifold the purge to compatible gases is likely to be based on specific site conditions (e.g., layout, clustering approach, equipment design, site preferences, etc.). We have used both configurations in different situations. Fixed purge systems are still required for gas cabinets, which are purged much more frequently for cylinder changes.

Although the purge cart is less convenient than using a fixed purge system, its use is relatively infrequent (with the possible exception of an initial tool ramp). In addition to significantly lower initial costs, the purge cart approach has also increased the available sub-fab space and reduced the inventory cost of purge cylinders.

Bulk specialty gas supply (BSGS) units are standard vendor-supplied packaged systems to supply inert or hazardous specialty gases from bulk containers to distribution panels (VMBs, IGTs, etc.). Bulk containers can be Y cylinders (24 in. diameter horizontal cylinders), isotainers, hydril tube trailers or bulk tanks. The systems may be located in the fab building, remote from the fab at a back pad, or both. BSGS systems are typically easier to apply to new factories, unless provisions were made for future bulk systems.

These systems can significantly reduce capital and operating costs. An assessment of costs involves such factors as comparing the BSGS costs, savings from purchasing bulk materials, and the cost of gas cabinets. BSGS can be justified based on high-volume usage (material savings) and/or a large number of points of use (driven by installed cabinet costs). The Table is a general guide for evaluating the cost of BSGS for a new factory.

Prior to performing this cost comparison, it may be useful to develop a short list of likely gases by determining which have:

• The highest usage per month.
• The highest cost per month.
• The highest number of points of use.
• The largest product cost delta (cylinders vs. bulk containers).

Typically, several gases will appear in more than one of these categories. These will have the highest potential for a favorable payback and should be evaluated first. The values for items A, B and C should be readily available. The costs for BSGS systems (items D, E and F) may require some additional investigation. The installation costs for a BSGS system can be especially difficult to define, particularly for an existing facility. In the case of an existing facility, it is also unlikely that the material savings will offset the total BSGS costs (except perhaps inerts).

Once a solid design concept is developed, the cost assessment can be straightforward. Particularly for hazardous gases, though, the design must be thought out carefully and thoroughly. In addition to the usual considerations of specialty gas system design, bulk systems have additional unique issues. For instance, most standard BSGS systems are designed strictly to replace the gas cabinet as the supply source and do not always address the distribution issues inherent to bulk supply. Usually, a gas cabinet supplies a single VMB that is installed with the gas cabinet. With a bulk supply system, one or several VMBs may be installed with the BSGS, and several more may be added in the future. Some consideration must be given to future expansion and modifications without service interruptions. For example, standard VMBs, or custom engineered valve boxes that can be used as "primary VMBs" (that is, VMBs that supply other VMBs) or branch VMBs. The number, location(s), and control functionality of these units requires careful consideration, as they can be crucial to the performance of the system.

3. With inert gases, it is relatively easy to plan future build-out valves for backup sources and future tools.

Consider the inert BSGS system depicted in Figure 3. With inert gases, it is relatively easy to plan future build-out valves for backup sources and future tools. With hazardous gases, it is much more difficult to achieve these capabilities because of the need for exhausted enclosures, toxic gas monitoring, and automatic isolation of the supply. A bulk silane system is depicted in Figure 4 . This is a fully redundant system with future expansion capabilities from spare gas sticks in the primary VMBs. The operation of this system was revised from its original design to prevent interruption of service caused by high process pressure, or leak detection in a single primary VMB.

4. This bulk silane schematic shows a fully redundant system with future expansion capabilities from spare gas sticks in the primary valve manifold boxes (VMBs).

Although BSGS systems can have significant cost advantages, they also inherently involve a higher level of risk because a larger number of tools are dependent on a single system. The cost of an unplanned interruption could exceed several years of BSGS cost savings. This is especially important when considering the use of BSGS for multiple wafer processing tools. In the case of hazardous BSGS systems, a formal analysis (e.g., failure mode and effect analysis, fault tree analysis, etc.) of failure modes is warranted.

Although not strictly a BSGS system, the use of bulk helium has become more common in recent years. The material cost of bulk helium is orders of magnitude less than cylinders. Helium can be supplied from a hydril tube trailer, Y cylinders, or a manifolded rack of cylinders, and distributed similarly to traditional process bulk gases (e.g., argon, oxygen, etc.). Purification of the helium may be required.

These examples highlight just a few of the many issues and opportunities that must be carefully considered with bulk specialty gas systems.

5. This high-flow inert gas rack includes bulk Y cylinders, which contain the equivalent of about six standard cylinders.

The high-flow inert gas rack (HFIR) was developed as an alternative to inert BSGS. It was installed in the existing Fab1 with a simple payback well under one year. The HFIR approach involves the use of a standard dual-cylinder, automatic crossover inert rack with a controller and remote monitoring. The gas panel has been upgraded with high-flow components, and is connected to bulk Y cylinders rather than standard cylinders to supply multiple VMPs and IGTs (Fig. 5 ). Y cylinders contain the equivalent of about six standard cylinders.

In this case, additional efforts were required to enable safe Y cylinder handling (cart and guardrails) and connection of Y cylinders rather than standard cylinders to the HFIR (additional regulation, flex tubing, etc.). This system was also designed to supply product from the Y cylinders to several existing VMPs (as well as the new IGT) by connecting from the HFIR to the existing inert racks on the standby side of the gas panel. This enabled the supply of all users from Y cylinders without interruption of service. The payback did not include labor savings due to decreased cylinder changes.

Application of this concept involves careful consideration of operational issues (cylinder handling), usages and Joule-Thompson cooling to ensure safe, reliable system performance.

Blending of hydrogen (H₂) mixtures can be very economical compared with cylinders. The equipment costs for blending are roughly equal to that for a single-cylinder supply. The material cost of using bulk hydrogen and nitrogen is one to two orders of magnitude less than that of cylinders. We have found the annual cost savings of a blending unit is >$100,000.

Blenders are not without their challenges, however. For instance:

• It can be difficult to find high-purity blend system manufacturers (a custom engineered and manufactured unit can be expensive and time-consuming to develop and implement).
• Maintaining the desired H₂ concentration during significant fluctuations in demand can be challenging.
• Blenders may be difficult to integrate into existing gas vendor monitoring systems.
• Blenders do not typically have the redundancy of cylinder systems.

Using a pressure-based cylinder backup source can offset some of the risks of blenders. This provides redundancy with significantly lower operating costs. In one case, a 10% H₂/N₂ blend system was installed with a payback of less than one year. The blender was integrated with an existing cylinder system. Based on the relatively high cylinder usage, additional tools, and a ramp in wafer starts, the material savings easily offset the capital costs. The challenges of blenders can be significant, but the potential savings are worth pursuing.

While the following represent relatively smaller opportunities, they are generally performed more frequently. For example, bending of electropolished tubing (up to 0.5 in. standard and coaxial) to a radius of 10× the tubing diameter is faster and less expensive than using manufactured fittings.

Initially, our approach to gas stick change outs was very conservative. When a gas line was removed from a tool, the tubing and the gas stick in the VMB (or VMP) were demolished. A new tool connected to that position of the gas panel would require a new gas stick. This practice was reviewed and guidelines were developed to allow for the inspection and reuse of most gas sticks. Due in part to the dynamic nature of DRAM manufacturing, this change saved $50,000 per year in parts alone.

The preceding approaches are not revolutionary or uncommon, but they are not universal either. Implementing changes to existing specialty gas systems can be difficult, but the potential savings are significant. The approaches described have saved several hundreds of thousands of dollars in capital and operating costs. In addition, these approaches have not adversely impacted system uptime, reliability or operational manpower.

Naturally, none of these concepts was, or should be, applied without careful coordination; consideration of specific site conditions, customer and operational needs, and specific gas properties; and documented revisions to procedures and training.