by Steve Russo, Texas Instruments

Executive overview

Semiconductor fabs face continual pressure to reduce energy consumption and capital costs, resulting in an ongoing effort to recycle and recover resources. Texas Instruments recently identified and capitalized on a significant opportunity to further these key objectives in one of its ion implantation modules.

Ion implant process tools typically are the largest consumers of cleanroom air in a fab. Air is drawn through the implanter to dissipate heat from process tools and to provide a dilution safety factor in the event of an on-board toxic gas leak. Ion implant tools use highly toxic materials which, by design, are integrated, stored and used within the tool itself (instead of by remote supply).

This article describes how Texas Instruments’ 300mm production fab substantially reduced its operating costs and capital expenditures — while maintaining a safe environment — by implementing Type 1 subatmospheric gas sources (SAGs) to transport, store and deliver toxic gases to ion implantation. It walks through the business case behind the change, covers the engineering adaptations made in the module’s exhaust system, and details the positive effect these modifications had on both operating and capital costs. (See the list below for definitions of technical terms used.)

Targeting exhaust-related expenses

Texas Instruments has a long history of innovation within the semiconductor industry, including recognition for pioneering development of the first integrated circuit. More recently, TI designed and built the first fab to attain LEED certification, an independent national benchmark recognizing compliance with sustainable environmental and energy practices. In 1998, recognizing that exhaust-related factors comprised a significant portion of both fixed and variable costs, the company’s staff began exploring possible ways to safely reduce exhaust requirements.

An internal analysis (Figure 1) revealed that ion implant tools required the greatest amount of exhaust — twice the amount of the next-highest consumer of cleanroom air, wet deck processing hoods. This proportion was due in part to the greatly expanding size of ion implant tools in the 1990s. Thus, the staff chose to focus initially on reducing exhaust requirements in the ion implant bay.

Figure 1. Exhaust output per tool, as determined by an internal TI study.

Figure 2 depicts a standard industry configuration for handling implant exhaust. After accounting for process exhaust — which is completely isolated and sent directly to a scrubber — two other major air streams are associated with an implanter: one from the gas box and another from its outer shell. In a high-current implanter, shell exhaust runs as high as 1500 cubic feet of air per minute and consists of primarily of heat load and secondary containment exhaust. This cleanroom-quality air was considered general exhaust and taken outside untreated. Anything from the gas box, however, was deemed acid exhaust and had to be sent to scrubbers before being released. As would be expected, acid exhaust consumed considerably more resources than general exhaust, including energy, and thus contributed more to the cost of ownership in operating an implanter.

Figure 2. Typical implant exhaust configuration.

Texas Instruments’ first attempt to address the cost factors around these air flows was in 1998 in 200mm and 150mm fabs. At the time, only two of three toxic gases used in ion implantation (arsine and phosphine) were being transported, stored and delivered using the SDS gas source, a Type 1 SAG. These cylinders maintain subatmospheric pressure under normal operating conditions and contain gases by locking up their molecules. The underlying physics of this reversible adsorption process renders it essentially impossible for gas to escape — until a vacuum is applied, as is the case when delivering gas to a tool.

Given this, members of the group reasoned, an exhaust system for containing and venting an accidental gas discharge could be adjusted to match the safety margin actually needed. However, a third gas, boron trifluoride (BF₃), was still being stored and delivered to TI’s implanters in standard high-pressure cylinders. Consequently, this initial effort to reduce shell side exhaust had to be suspended for several years.

In late 2000, Texas Instruments opened its first 300mm fab, and the facility’s implanters were installed in the main portion of the cleanroom over the waffle table floor. By 2003, TI required more production capacity from its fab and set in place plans to transform a one-time administrative section into cleanroom space. It was decided that the implanters would occupy this area.

Specifying new exhaust flows

The reconfigured space was designed to house 30 implanters. However, standard technical requirements meant the exhaust load would be extremely dense. Additionally, estimates of the capital costs to retrofit the area were exorbitant, with exhaust and make up (fresh air) elements comprising a large percentage of that expense.

Tasked with reducing up-front capital expenditures, as well as year-over-year energy costs, a team was assembled to address the situation. In addition to meeting stringent safety, energy, exhaust and financial objectives, members of the group faced strict deadline pressure: they would have just four to six weeks to make recommendations that complied with an already-established construction time-line. The team consisted of five mechanical engineers who regularly worked with exhaust systems; two environmental engineers; two safety engineers; and a mechanical engineer who specialized in the design of new cleanroom facilities.

Among the goals TI team members set were to:

• Design a safe BF3 delivery system utilizing a subatmospheric methodology.
• Configure the implanter outer shell exhaust so, without any safety concerns, it could be routed back into the room for energy and cost savings.
• Ensure all new designs comply with SEMI S2 and S8, safety and environmental standards established by semiconductor companies for tool manufacturers.
• Optimize overall exhaust demand to a level as low as possible while maintaining tool and safety requirements.

The team’s first step was to tackle shell-side exhaust requirements. More than 99.9% of the time, this flow consisted entirely of heat and room air; the only way anything potentially toxic could enter the flow was if a system leak had occurred. With arsine and phosphine stored in SDS Type 1 cylinders, multiple compounding failures would need to take place for any accidental discharge of these gases to occur. To elevate on-site BF₃ to a higher safety level, team members recommended eliminating traditional pressurized cylinders. They replaced them with a VAC system, a Type 2 SAG technology that stores gas at high pressure but extends the safety envelope by delivering it at sub-atmospheric pressure.

Once all three gases were being managed in this fashion, the Texas Instruments safety department and one of the company’s implant manufacturers approved the shell exhaust for those tools to be re-circulated back into the room air. In addition to reducing the inherent risk associated with high-pressure gas cylinders, the change had the immediate benefit of avoiding any capital expenditure for installing additional exhaust and make up air systems in the new implanter space. Even after factoring in the cost of new subatmospheric storage and delivery solutions, this approach turned out to be the most economical. Up-front costs entailed a less than one-year simple payback period. Recovering air that had been dumped outside previously and safely reusing it ultimately was calculated to generate $200,000 in annual variable cost savings.

In developing these recommendations, the team carefully examined placement of the life safety sensors (LSS) that monitored whether gas releases occurred within the shell exhaust or outside the tool. Under prior designs, detectors were installed in the exhaust duct as well as under the raised floor at the wafer-loading end of the implanter. An LSS inside the duct, if triggered, would automatically shut down gas flow inside the tool. Similarly, an LSS monitoring the breathing zone would immediately evacuate that space if it detected any gases over acceptable limits and shut down the gas flow inside the tool.

Even with the exhaust system changes being envisioned, no modifications had to be made to LSS placement. However, from an avoided cost standpoint, the potential savings were significant. With the ion implanters situated in a dedicated room, a single breathing-zone LSS detection would evacuate only that one room — not the entire plant. Although this scenario was considered unlikely, it was a key factor in the group’s decision to move in this direction. Figure 3 is a screenshot of the LSS monitor that TI maintains in its 300mm production fab. Software dashboards similar to this one track real time data collected from roughly 300,000 points in the facility every five to 10 seconds. This monitor is specific to safety reading for the ion implant area.

Figure 3. Screen shot of LSS monitor.

Optimizing the overall system

Shell exhaust connections are almost always on top of the tool. Instead of pushing air out the top of the implanter — which might be a source of particles in the cleanroom — the team opted to route this air underneath the raised floor. Many implanters also have a fan on the shell but it may be too small to overcome much of a pressure drop. Team members settled on adding a booster fan on the shell exhaust to overcome the ductwork’s pressure. At the same time, they specified installation of a secondary duct to the general exhaust system. Normally, its damper would be closed. However, if the booster fan underperformed or failed for any reason, the secondary ductwork would provide exhaust for an individual tool during an emergency or maintenance situation. The room’s general exhaust system was sized to allow no more than three of the secondary ducts open at any time.

Work to optimize shell exhaust also led to the development of other cost savings. Implant manufacturers uniformly recommend that air from the gas box be added to the acid exhaust. Disposition of this flow (on the order of 300 cubic feet per minute per implanter) was subject to building regulations and internal Texas Instruments safety codes. Its handling as acid exhaust was not because it innately required treatment, but as a form of insurance against it containing toxic gases that had been accidentally released in the gas box. Indeed, under local environmental codes, such an instance would be treated as an emergency release only: it would not have to be abated, just documented and reported.

Again, the only way an accidental release could occur would be from a failure in one of the subatmospheric delivery systems — an unlikely event given the history of the two technologies, Type 1 and Type 2 SAGs.
Accordingly, the team floated the idea of taking advantage of the superior safety profiles of the SDS and VAC systems and shifted this flow to the general exhaust system (Figure 4). While this air would not be recycled, it would save over $400,000 in capital expenditures for exhaust duct materials and abatement system installation expenses. And by not venting this air to the scrubber, the company would realize more than $50,000 per year in energy savings that would have gone to operating the abatement system.

Figure 4. Final proposed implant exhaust configuration.

Figure 5 contains two images from the renovated ion implanter area at TI’s 300mm production fab. The side view of one implanter shows several of the exhaust and safety features reconfigured in the design team’s capital cost reduction and energy savings project. Looking down the aisle behind several implanters reveals venting leading to ductwork under the floor and the blue booster fans that help vent shell exhaust and recirculate air safely back into the room.

Figure 5. Photographs of final configuration.

Effect of a changing energy picture

In 2008, spiking energy prices triggered a new effort to capture additional exhaust and energy reduction savings. At the same time energy prices were reaching record highs, overall economic performance was slipping dramatically, so an effort to cut utility costs in the semiconductor business became even more important.

The team started by reexamining implanters that had not been selected for exhaust optimization in 2003 — exploring options around routing the shell exhaust from these tools back to the room. However, doing so would have incurred expenses around reconfiguring duct work and outfitting booster fans, as had been done earlier.

Instead, they considered an alternative that involved no additional cost: reducing the shell exhaust requirement directly. Following plans developed collaboratively by the team’s facilities and equipment engineers, they installed a temporary data logger in the shell exhaust and beamline area of a high current machine to monitor its temperature. Figure 6 graphs the temperature recorded at the source pump and the beamline while the exhaust flow was reduced. Lowering it gradually over an eight-week period generated tool data from all different modes of operation. The exercise clearly confirmed that flow could be scaled back by more than 50 percent with little to no change in temperature.

Figure 6. Implanter exhaust reduction to 250 from 530CFM.

The team decided against dropping the flow any further to ensure the implanter shell remained at a negative pressure. Nonetheless, their evaluation cleared the way to reduce the flow on a total of eight implanters. Owing to less power needed to run exhaust fans and a diminished make up air requirement, the move resulted in a reduction of roughly 2,000 cfm, or about $10,000 in annual energy savings.

Designing in savings

The significant capital, energy and cost savings TI achieved by incorporating SAGS technologies in a limited expansion project allude to larger and sustained economic opportunities. Even greater returns may be generated when similar approaches are applied at the time fabs are being designed and built. By engineering air handling systems, abatement units, life safety monitor systems, etc., to correspond with the safety margins that SAGS solutions provide, these measures contribute to reducing up-front capital costs while lowering both the cost of ownership of tools and a facility’s overall operating costs.

Conclusion

Texas Instruments’ entire process to incorporate SAG technologies took place in phases, spanning several years, and required a group of individuals to implement the changes. The payoff for the team’s perseverance was an effort that delivered both safety and economic benefits. At the time of this article’s publication, there have been no gas releases at Texas Instruments with either an SDS or VAC system. Overall, the effort resulted in close to $1 million in capital savings and approximately $150,000 in recurring annual energy savings. Anticipating sustained upward pressure on energy prices over the long term, the company’s ROI for this project will most assuredly continue to increase.

Acknowledgments

The author acknowledges the contributions of the following TI engineers and technicians who participated in the company’s energy and capital cost savings project: Kevin Ditzler, Tina Gilliland, Rene Graves, Russell Hill, Bob Johnston, Kathy Meissner, Jeff Miller, John Miller, Mike Mitchell and Jay West. He also acknowledges Jim Mayer of ATMI for his contributions to this article.

Steve Russo is a graduate of Old Dominion U. in mechanical engineering. He is a senior member of the technical staff at Texas Instruments, PO Box 655012, Mail Station 361, TX 75265 USA; ph.: 972-927-7983; email [email protected].

Definitions and Typicals

• Shell exhaust: the two lines in blue off the top of the implanter in the diagrams. Typically each connection is 8″ to 12″ and the quantity of exhaust for the combined connections is 500-1400 cubic feet per minute (cfm).
• Gas box exhaust: The red line going to acid exhaust off of the top of the gas box. Typically one connection that is 8″ to 10″ that uses about 500 cfm.
• Pump exhaust: Usually a 4″ to 6″ connection with very little flow (<100 cfm). There is usually a shell around the vacuum pump that will also be sent to acid exhaust that is included in this number. This is not depicted in the drawing below but should be.
• Waffle table: A concrete structure that has preformed holes in it that supports the raised floor of the clean room as well as the tools. The purpose of the holes are to let air return to the second level and to allow the utility piping from the second level easy access to the manufacturing tools.
• General exhaust: Ventilation that is taken outside with no means of removing any gasses entrained. Typically heat and inert gasses go in to this stream.
• Acid exhaust: Ventilation that is taken outside that is treated to remove acidic fumes from the air. Typically a wet scrubber is used.
• Make up or “fresh air”: Air brought in to the building to replace the air that is exhausted out of the building. A clean room has positive pressure to reduce particles from being brought in to the building. Therefore, more make up air is brought in to the building than is ventilated out the exhaust systems.
• SDS: A Type 1 SAG that stores and delivers its contents at a pressure of less than 14.7 psia at NTP.
• VAC: Vacuum actuated cylinder, a Type 2 SAG — a pressure-controlled for delivering ion implant dopant gases at sub-atmospheric pressures.
• Life safety sensors (LSS): Sensors to detect gasses identified that are hazardous to health. These are dictated by local codes and TI’s safety department.