Strategies for energy reduction in semiconductor manufacturing
10/01/2008
EXECUTIVE OVERVIEW
ISMI’s Environment Safety Health (ESH) Program has undertaken specific projects to demonstrate efficiency and conserve energy in semiconductor manufacturing operations. The strategies articulated in this article could potentially reduce the industry’s contribution to climate change and reduce overall production costs. This paper highlights the ISMI ESH Program’s energy reduction strategies, addressing both fab facilities and processing equipment.
Facilities work conducted by ISMI’s ESH Program includes studies of optimal cleanroom air management and humidification methods, fab energy and heat recovery opportunities, and development of semiconductor fab-specific Leadership in Environmental and Engineering Design (LEED) criteria. Process equipment energy reduction efforts have also focused on characterizing process equipment energy use during idle and processing modes, and demonstrating a low utility consumption idle mode for vacuum pumps and point-of-use abatement systems. The process equipment energy studies include the application of the SEMI S23 standard - A Guide for Conservation of Energy, Utilities and Materials Used by Semiconductor Manufacturing Equipment.
 |
null
 |
null
Facilities energy
For over a decade, the ESH Energy program at SEMATECH/ISMI has characterized energy consumption in semiconductor facilities at both the fab and process level. Fab-wide power consumption studies show that process equipment uses 40%-50% of the total power used by a device manufacturing facility. Data collection periods are 1997 (150 and 200mm fabs) and 2007 (200 and 300mm fabs) (Fig. 1).
 Figure 1. Energy use by system based on fab wide characterization studies in 1997 and 2007. |
null
The allocation charts in Fig. 1 have inspired further studies focusing on specific central utilities. ISMI’s most recent project compared the best available technology for recirculating air in a cleanroom. Figure 1 shows that the average recirculation air energy requirement dropped from 11% of the fab total energy use in 1997 to 5.3% (200mm) and 3.9% (300mm) in 2007. The data correlates well with two significant changes that occurred over the ten year period: 1) fab designs have progressed from traditional recirculation air handlers to fan filter units (FFUs) and 2) the transition to mini-environments on process equipment has relaxed the flow-rate requirement for HEPA-filtered recirculation air. Additionally, ceiling coverage for filtration has reduced from 100% to 25-40% in process areas with wafer traffic. The ISMI study compared four recirculation air configurations:
- Fan filter units -- HEPA filters with dedicated fan motors mounted above the filter.
- Recirculation air handlers -- Traditional air handlers with ducted supply and returns to individual HEPA filters.
- Fan tower -- Large axial fans or fan walls that draw air from the subfab and pressurize a plenum above the fab.
- Plenum module -- Similar to recirculation air handler but without ductwork since the unit incorporates the HEPA filters and supplies a pressurized plenum directly above the filters. The comparison study showed that on a power per unit of flow (kWh/cfm) basis, FFU cleanrooms have the lowest energy consumption (Fig. 2).
 Figure 2. Comparison of energy consumption for various recirulation air handling techniques. |
The FFU configuration minimizes pressure drop, and therefore energy consumption. Another key finding was that despite the 0.25hp (0.19kW) FFU fan motors having the poorest efficiency rating of 78%, those cleanrooms operated at lower specific energy use than all other systems with larger motors, at +6hp (+4.5kW) and 86% efficiency. The study also identified that existing fan tower and recirculation air handler configurations could be improved by reducing losses caused by prefilters, flow control, and noise attenuation devices.
Energy reduction at the facility level is more mature with cost and production impacts well understood. However, ISMI’s member companies still see opportunities for energy reduction through a comparison of system design and performance. ISMI’s most recent study is a cost analysis comparing four different makeup air humidification techniques:
- Electric resistance heat-generated steam
- Wetted media
- Compressed-air atomizing spray nozzles
- High-pressure water atomizing spray nozzles
 |
null
Table 1 shows that steam is the most energy efficient, followed closely by high-pressure water atomization. However, this comparison does not comprehend differences in operating cost. For instance, steam generation cost can vary significantly depending on the type of heating (i.e., electricity, natural gas, or a recovered heat source). Table 2 provides a comparison of capital, maintenance, and annual energy cost.
Steam looks favorable according to data in Table 2, but the high-pressure water atomization and wetted media options become much more attractive when using a recovered heat source to preheat the makeup air. For example, heat recovery chillers have been used successfully to provide most of the preheat energy necessary for the humidification process.
Another previous study by ISMI explored energy and heat recovery inside semiconductor facilities. With many continuous heat transfer processes, there are additional opportunities for heat recovery. The objective of this study was to identify opportunities, document energy recovery concepts, evaluate feasibility, and highlight the most cost-effective options. The study identified the process cooling water (PCW), ultra-pure water (UPW), and clean dry air (CDA) process streams as having the greatest potential. It identified a total of 19 heat recovery ideas; the bulk of them including UPW and CDA processes. An example of an identified concept is transferring waste heat from process tools and air conditioning through a heat recovery chiller to incoming city water for more efficient water deionization. Heat rejected by plant air compressors can also be recovered within an integrated heat recovery scheme.
Implementation of these types of energy reduction strategies are usually driven by cost savings. However, there are additional benefits to implementation such as potential credit towards LEED certification. LEED is a rating system developed by the United States Green Building Council (USGBC) that has emerged as the de facto approach to green building certification. Recognizing the critical importance of proactive environmental design for its members’ fabs, ISMI has initiated the development of fab-specific environmental performance criteria, including green building guidance for designers and operators of fabs. The current credit structure for LEED certification does not acknowledge semiconductor-specific processes and energy reduction strategies. Key elements of ISMI’s work in this area include an initiative to establish fab-specific LEED credits, and development of a LEED Application Guide for High Tech Facilities that will describe approaches and best practices for achieving LEED certification. ISMI also provides LEED training and workshops for its members, which highlights proven approaches for achieving LEED certification for semiconductor manufacturing facilities.
Process equipment energy
Figure 1 also shows that over a ten-year span, process equipment energy consumption continues to be a significant portion of the total fab energy usage. The data led to process equipment energy studies to determine power allocations on a process tool and process tool component level (Fig. 3).
 Figure 3. Process equipment energy use at the component level. |
Studies showed that 70-90% of equipment power is used for heating and vacuum pumps. In some cases, equipment suppliers have responded by making improvements in energy consumption at the component level. As an example, power consumption was reduced in vacuum pumps by transitioning to high-efficiency motors and reducing cooling water flow requirements. A key finding in the process equipment studies was that for some processes, the energy consumption levels did not change significantly during idle and processing operating modes (Fig. 4).
 Figure 4. Year 2000 comparison of tool energy use in idle and processing mode. |
It is important to note that this data was collected in the year 2000, before reduced energy consumption idle-mode features were available from support equipment suppliers. Since 2000, equipment suppliers have measured utility consumption and progressed towards lower energy idle-mode capabilities. However, idle-mode implementation, with associated energy reductions in process equipment, is limited by concerns with process and production impacts. Consequently, ISMI studies have focused on high energy components having a low probability of process and production impacts.
ISMI recently completed two studies examining a low-power mode for vacuum pumps and point-of-use abatement devices. A vacuum pump study was completed in 2006 and showed that significant savings is achieved by upgrading to newer pumps with further savings by utilizing pump speed control to run in reduced idle mode (Fig. 5).
 Figure 5. Comparison of utility consumption on an existing, new, and speed-control-capable vacuum pumps. |
Figure 5 also shows reductions for all utilities: power, cooling water, and nitrogen. As cooling water flow decreases, the temperature delta is increasing, therefore, maximizing temperature delta will ultimately reduce the pumping energy requirement for the facility. The potential energy savings become significant if we assume 600 pumps for a typical production facility. The study estimated that annual savings in utility reductions would be roughly $600,000 and peak infrastructure requirements would be reduced by 0.5MW.
 Figure 6. Utility consumption comparison for process and reduced utility consumption idle modes on a burn/wet abatement device. |
The next project was focused on point-of-use (POU) abatement devices, another support item used throughout a typical sub-fab for etch and CVD processes. Figure 6 shows that the abatement study yielded similar results. The data illustrates that cooling water flow is held constant, but the temperature delta decreases as the unit goes into an unloaded idle state. Circulating more cooling water than required results in poor temperature deltas, which requires the facility to use more pumping energy than necessary. Typical facilities have 4-6°F temperature deltas, while they are designed for 10°F. Lower consumption in idle mode is possible by reducing utilities for incineration and scrubbing processes while the tool is not processing wafers. Again, this capability becomes significant considering that a typical fab will have 80 units with a potential of 4 hours of idle time/day (i.e., tool is waiting for next wafer cassette). Savings are estimated at $1,400 to $2,100/year per POU device. An important criterion for abatement is destruction rate efficiency (DRE), which was also measured during the testing. The data showed that DRE performance impact was not significant.
Figures 5 and 6 illustrate a clear progression from the Fig. 4’s year 2000 data, where little to no change in energy consumption was observed between idle and processing states. While idle-mode demonstrations for vacuum pumps and POU abatement had promising results, the capability is far from actual implementation. Several issues that need to be resolved are communicating with the process tool, establishing reliable vacuum levels during idle periods, and evaluating response time for vacuum pumps and POU abatement devices to recover from idle mode. Initial studies showed that the response times should not be a factor for pumps and POU abatement devices, but further demonstrations are needed to establish credibility. This year, ISMI is conducting a feasibility study for operating process equipment in an idle mode. The project will examine existing SEMI standards and how they address idle mode, as well as examining idle mode on a tool component level (Fig. 7).
 Figure 7. Electrical energy reduction for components on a typical etch or CVD tool. |
One of the steps needed to advance energy reduction efforts was standardizing how equipment suppliers characterize energy consumption on process equipment. The existing SEMI guideline, SEMI S23-0705, “Guide for Conservation of Energy, Utilities, and Materials Used by Semiconductor Manufacturing Equipment,”prescribes a protocol/method for collecting, analyzing, and reporting energy/utility data, as well as a method to convert utility consumption data into kWh power consumption using standard utility-to-power conversion factors.
In 2007, ISMI developed the SEMI S23, “Application Guide and Total Equivalent Energy (TEE) Conversion Tool: Selecting and Using Measurement Instruments to Conserve Resources,”which provides assistance in the application of SEMI S23-0705. The objective was to provide guidance in the selection and use of utility measurement instruments, make recommendations for energy-use reduction, and provide a software tool with user instructions to simplify the analysis and presentation of measured utility data. The software tool converts utility data to equivalent kilowatt-hour per year values based on S23-defined utility energy conversion factors. The Application Guide summarizes measurement methods, including recommended practices, for suppliers fulfilling the S23 reporting requirements. It describes measurement methods for temperature, pressure, liquid/gas flow, and power, and discusses the relative ease of application, method accuracy, and relative costs of instruments. It also recommends practices for reducing the energy consumption of semiconductor manufacturing equipment and facility systems directly impacted by the tools.
Conclusion
A decade of energy reduction research at ISMI has yielded results at both the facility and process equipment level. Examples of facility efforts are reduced energy to recirculate cleanroom air and humidify makeup air, as well as integrated heat recovery. A program is under way to establish criteria for certification of semiconductor manufacturing facilities under LEED. Likewise, energy studies have highlighted the need for process equipment energy reduction, and achievable results have already been demonstrated. However, to successfully implement these techniques, end users need to communicate energy reduction priorities to equipment suppliers/fab designers and be willing to demonstrate these techniques in a high-volume production environment.
It will be critical to start with equipment furthest away from the process, such as pumps and POU devices, to overcome concerns of process performance and production impacts. Existing resources (S23, Application Guide and TEE Tool) are available to guide equipment suppliers towards establishing a tool-specific baseline for process energy usage and a roadmap for improvement.
Acknowledgments
The author would like to thank the following individuals for their contributions to this report: Ralph Cohen, Ralph M. Cohen Consultancy; and James Beasley and Walter Worth of ISMI.
LEED is a registered trademark of the United States Green Building Council.
References
- J. Quisenberry, K. Fenstermaker, “Summary Facilities Energy Consumption in 200 and 300mm Fabs,”ISMI Technical Transfer Publication #08024920A-TR, Mar. 2008.
- R. Cohen, “Field Survey of Wafer Fab Cleanroom Air Management,”ISMI Technical Transfer Publication #07104875A-ENG, Oct. 2007.
- R. Cohen, “Methods and Efficiency of Adiabatic Humidification for Semiconductor Cleanrooms,”ISMI Technical Transfer Publication #08034927A-ENG, May 2008.
- A. Giles, P. Breder, “Analysis and Energy Usage for Three 300mm Process Tools: RTP, Cu CMP and PECVD,”ISMI Technical Transfer Publication #08024923A-ENG, Mar. 2008.
- S. Hinson, D. Ranum, T. Huang, “Analysis and Energy Usage in Three Process Tools,”SEMATECH Technical Transfer Publication #99113843B-ENG, Jan.2000
- P. Naughton, J. Weale; “ISMI Vacuum Pump Idle Mode Energy and Characterization Demonstration Projects,”ISMI Technical Transfer Publication #06094784A-ENG, Sept. 2006.
- J. Holsbrink, J. Gompel, S. Pampel; “Idle Mode Energy Savings for Point-of-Use (POU) Abatement Equipment,”ISMI Technical Transfer Publication #07124899A-ENG, Dec. 2007.
Thomas Huang, received his BSME from the U. of Texas at Austin, is a registered professional engineer in the state of Texas, and is a project manager at International SEMATECH Manufacturing Initiative, 2706 Montopolis Drive, Austin, Texas 78741 USA; ph.: 512-356-3628, email: [email protected].
 |
null