1、290 2009 ASHRAEABSTRACTCurrently, the electronics industry such as integrated circuit (IC) industry is primarily composed of high-tech Orig-inal Equipment Manufacturers (OEM) and Original Design Manufacturers (ODM) in Taiwan. The electronics IC industry is not only capital- and technology-intensive,
2、 but also energy-intensive. Since the Kyoto Global Warming Conference in December 1997, understanding and mitigating the emission of carbon dioxide (CO2) from industrial processes has become an important issue. The goals of this study are to develop refer-encing information by obtaining specific ene
3、rgy consumption data from the IC industry, which can be used by the industries, plant builders, energy companies, and government agencies to identify opportunities and enhance energy efficiency in the IC industry. Specifically, this study focuses on SEC for two sectors in the IC industry - the IC as
4、sembly and testing that are the back-end stage processes in IC manufacturing. We quantified levels of SEC through benchmarking the energy consumption data for a selected group of fabrication plants (hereafter referred to as “fabs”) at their maximal capacity, which was defined as the greatest number
5、of output that the fab can produce under normal conditions with existing resources. The “capacity utilization” of a fab refers to the ratio of actual total output to the maximal capacity. In this study, the total 11 fabs capacity accounts for more than 60% of the total national capacity for IC indus
6、try in Taiwan. We have found that facility systems and process tools shared the major portion of the power consumption by 53% and 47% respectively in all partic-ipated IC A/T fabs. The average cooling load of the IC A/T fabs ranged from 0.09 to 0.31 USRT/m2. For IC assembly fabs stud-ied, when their
7、 capacity utilization were 100%, the SEC ranged from 21.2 kWh/k pieces to 119.4 kWh/k pieces, with an average of 48.1 kWh/k pieces. For IC testing fabs studied, when their capacity utilization were 100%, the SEC ranged from 76.3 kWh/k pieces to 201.1 kWh/k pieces, with an average of 122.2 kWh/k piec
8、es. We estimated that if all the IC A/T fabs in Taiwan achieved the level of specific energy consumption of the best performer (i.e., lowest SEC level) among the 11 surveyed fabs studied in this paper, the annual energy savings and of CO2-emission reduction from the IC A/T industry Taiwan would be e
9、xpected to reach 258,847,000 kWh/year and 164,886 tons, respectively. In addition, an innovative energy efficient design of the cleanroom ventilation system is proposed to save energy use in the IC A/T fabs. A better design and control can save fan energy by more than 20%. INTRODUCTIONWorld producti
10、on of semiconductors was valued at US$256 billion in 2007, with average annual growth 12-13% per year over the last two decades (Semiconductor Industry Association 2008). Previous studies have shown that the energy and materials intensity of semiconductor manufactur-ing is surprisingly large given t
11、he tiny physical scale of prod-ucts (Williams et al. 2002, Murphy et al. 2003). The high energy intensity required is due to needs for pure inputs and tightly controlled environments for semiconductor process-ing. In addition, purity standards for input materials are elevated with the decrease in fe
12、ature size for microelectronic circuits (Krishnan et al. 2008). Attaining and maintaining the purity standards in controlled environments induces substan-tial energy and materials uses. Since the Kyoto Global Warm-ing Conference in December 1997, understanding and mitigating the emission of carbon d
13、ioxide (CO2) from indus-trial processes has become an important issue.Specific Energy Consumption (SEC) for the Integrated Circuit Assembly and Testing (IC A/T) Industry in TaiwanA. Chang S.-C. Hu, PhD T. Xu, PhD Member ASHRAE D. Y.-L. Chan, PhD R. T.-C. HsuA. Chang is a PhD student and S.-C. Hu is
14、a professor in the Department of Energy and Refrigerating Air-Conditioning Engineering, National Taipei University of Technology, Taipei, Taiwan. T. Xu is staff scientist at the International Energy Studies Group, Environmental Energy Tech-nologies Division, Lawrence Berkeley National Laboratory, Be
15、rkeley, CA. D. Y.-L. Chan and R. T.-C. Hsu are staff researchers at the Energy and Environmental Research Labs, Industrial Technology Research Institute, Hinchu, Taiwan.LO-09-025 2009, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHR
16、AE Transactions 2009, vol. 115, part 2. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.ASHRAE Transactions 291In Taiwan, up to 98% of energy supply is from imports. In 2006, the i
17、ndustrial sector accounted for 58% of the total national energy consumption annually, with the integrated circuit (IC) industry used 12% of the total annual energy. Since the Kyoto Global Warming Conference in December 1997, understanding and mitigating the emission of carbon dioxide (CO2) from indu
18、strial processes has become an important issue. The electronics industry such as integrated circuit industry is primarily composed of high-tech Original Equip-ment Manufacturers (OEMs) and Original Design Manufac-turers (ODMs) in Taiwan. The electronics IC industry is not only capital- and technolog
19、y-intensive, but also energy-inten-sive. The characteristics of energy use in a wafer- or IC-fab include: 1. A high recirculation rate of filtered airflow in order to maintain a high air cleanliness level, resulting in high fan power demand.2. Precise and tight controls of cleanroom air temperature
20、and humidity (e.g., 240.5oC and 405% RH), resulting in high energy demand and use for the simultaneous cool-ing and heating of air.3. The manufacturing tools and process equipment consume significant energy, which in turn impose high heat load that need to be removed from the controlled environments
21、.4. A large amount of ultra-pure water and pure gases are used, which must be supplied by processes with high-energy consumption.5. High flow rates of exhaust air and outside air require significant transport and treatment energy. Limited open literatures on investigating energy uses in wafer or IC
22、fabs exist. Yao et al. (2003) compared energy use to produce two generations of computer chip products (i.e., Pentium Pro and a Pentium IV), which showed that producing a latter generation chip required 29% less energy than its processor. Xu (2003, 2004, 2008) reported energy consump-tion and partic
23、le control of facility systems of various micro-electronics fabs including minienvironments in the USA and characterized fab energy use in terms of energy use or power demand normalized by cleanroom floor area. Hu and Chuah (2002) studied the energy consumption of wafer fabrication plants in Taiwan.
24、 Naughton (2007) proposed SEC data in meeting the requirements or goals set in the International Roadmap for Semiconductors. Prior to this investigation, however, published energy demand data for IC A/T fabs in Taiwan are essentially non-existent. This is perhaps due to a combination of the followin
25、g reasons: energy consumption data of the mission facilities in IC A/T fabs are often not read-ily available or sharable. In addition, a certain level of sophis-tication and expertise is required for obtaining the data through measuring and monitoring power demand or consumption of the complex facil
26、ity and process systems. Added to the complexity, energy conservation has not yet been on the highest priority list in the semiconductor industry in Taiwan, and very few survey studies have been conducted. Given that Taiwan is now a major manufacturing base for semiconductors in the world, with IC A
27、/T industry most nota-ble in that Taiwan has a worldwide market share of more than 45% of IC assembly and 60% of IC testing (Industry Technol-ogy Research Institute 2006), it is important to develop infor-mation on energy use and assess energy saving potentials in the industry.The overall goals of t
28、his study are to develop referencing information by obtaining specific energy consumption (SEC) data from the IC industry, which can be used by the industries, plant builders, energy companies, and government agencies to identify opportunities and enhance energy efficiency in the IC industry. Specif
29、ically, this study focuses on SEC for two sectors in the IC industry - the IC assembly and testing that are the back-end stage processes in IC manufacturing.The technical objectives of this study include (1) to quan-tify and compare the SEC levels for IC A/T fabs, and (2) to identify the opportuniti
30、es for energy-savings in the IC fabs in Taiwan. This study covers all the power uses in the fabs, including the facility systems, process tools, offices and garages, etc. The facility systems include chiller plants, make-up air, recirculation air, exhaust air, nitrogen, compressed dry air, process c
31、ooling water, vacuum and ultra-pure water systems. METHODOLOGYCharacteristics of Microelectronics FabsThe whole supply chain for the semiconductor manufac-turing process is shown in Figure 1. Wafer fabrication and IC A/T are the two major energy-intensive processes, which are carried out in well-con
32、trolled environment, i.e., cleanrooms. Normally, the two processes are accomplished in two distinct fabs, a wafer fabrication fab and an IC A/T fab. Some IC A/T fabs are in the same plant and some are separate. Even in the same plant, the IC A/T fabs are somewhat different in type of process tools,
33、utility system, facility system etc. The energy inflow/outflow and support facility system serving typical microelectronics fabs (wafer fabrication and IC A/T) is shown in Figure 2. Facility systems include the chiller plant, makeup air, recirculation air, exhaust air, gases, compressed dry air, pro
34、cess cooling water, vacuum and ultra-pure water systems.Table 1 shows the characteristics of the eleven IC A/T fabs (six for IC assembly and five for IC testing) in this study. The profiles include production volume, cleanroom cleanliness level, recirculation system, floor area and cooling load. The
35、se IC A/T fabs were selected as they accounted for more than 60 percent of annual production volume in Taiwan. Given the variety and the scales of the fabs in this study, energy and power consumption of these eleven fabs may well repre-sent typical IC A/T industry in Taiwan. Data were collected over
36、 from April of 2007 to September of 2008 using central monitoring systems. Additional data collection was from 292 ASHRAE Transactionsaudited reports and on-site observations. Calibrations and verifications of data were performed to ensure data accuracies.Performance Comparison Using SEC IndexThe fa
37、bs studied varied in size, cleanroom cleanliness class, product type, and loads. For uniform comparison of energy performance of the IC A/T fabs, SEC was calculated based on total energy consumption (including electricity, oil, and gas) in the process tool and facility systems per number of IC proce
38、ssed (see Figure 3).We quantified levels of SEC through benchmarking the energy consumption data for a selected group of fabs at their maximal capacity, which was defined as the greatest number of output that the fab can produce under normal conditions with existing resources. The “capacity utilizat
39、ion” of a fab refers to the ratio of actual total output to the maximal capac-ity. In this study, the selected 11 fabs total capacity accounts Figure 1 Supply chain of the semiconductor industry.Figure 2 The energy inflow/outflow and support facility system of an IC A/T fab. ASHRAE Transactions 293f
40、or more than 60% of the total national capacity for IC A/T industry in Taiwan.The heat content of various energy commodities in Taiwan is shown in Table 2. The CO2emission mass per unit of electricity is 637 g/kWh in 2007 (ITRI, 2007). The relationship between the SEC and the capacity utili-zation o
41、f a fab was described as the power law equation:SEC = A (Capacity Utilization)BThe coefficients A and B were determined by the least square regression method, which can be applied using Excel software. The baseline SEC was determined based on the specific energy consumption data of the best performe
42、r (i.e., lowest Table 1. Characteristics of the Participated IC A/T FabsUnits of Production(in Thousand Pieces), kCleanroom Cleanliness Level/ Recirculation SystemAir-Conditioned Area,m2Cooling Load,USRT/m2A-1 2,123,253 ISO Class 57; AHU system 70,000 0.1A-2 483,370 ISO Class 67; FFU system 7900 0.0
43、9A-3 108,838 ISO Class 57; AHU + FFU systems 7000 0.1A-4 1,629,346 ISO Class 67; AHU system 50,000 0.1A-5 172,666 ISO Class 67; AHU system 8870 0.1A-6 728,386 ISO Class 67; AHU system 45,000 0.1T-7 1,737,056 ISO Class 57; AHU system 19,735 0.2T-8 483,370 ISO Class 57; AHU system 6492 0.28T-9 820,716
44、 ISO Class 78; AHU system 9825 0.31T-10 2,459,096 ISO Class 78; FFU system 58,330 0.26T-11 623,050 ISO Class 57; FFU system 14,000 0.28Note: The AHU equals the ducted air-handling unit system, and the FFU system equals the fan-filter unit system.Figure 3 The calculation process of SEC.294 ASHRAE Tra
45、nsactionsSEC among all the fabs) when its capacity utilization was 100%. For the long term, the capacity utilization of a fab depends on the increase in capital investment, but for the short term, the capacity utilization of a fab can be increased by giving work-overtime to the employees or by incre
46、asing the number of temporary workers, in order to increase output. In such a case, capacity utilization may exceed 100%. The energy saving potential and the associated reduction of CO2emission was estimated by: Energy Saving (kWh/y) = (averaged SEC of all the fabsstudiedminimal SEC among all the fa
47、bs studied) (averaged product amount of all the fabs studied)The potential reduction of CO2emission (kg/yr) was esti-mated by: Energy Saving (kWh/y) 0.637 kg/kWhRESULTS Power consumption allocationFigure 4 shows the average allocation of energy consump-tion by different categories of the IC A/T fabs
48、 studied. The facility systems account for 53% of all power consumed in the fabs, and the process tools account for 47% of the total power consumption. For the facility systems, chiller plants are the largest energy users, accounting for 31%of the total power consumption. We also found that the port
49、ion used by air-conditioning systems in an average IC testing fab was larger than that of an average IC assembly fab. The major reason is due to the air-cooled testing tools (i.e. testers), that result high cooling load in cleanroom.SEC AllocationFigures 5 and 6 show the relationship between SEC and the capacity utilization of the six IC assembly fabs (named IC A1-A6) and five IC test fabs (named IC T7T11). For each of lated SEC as a function of the changing capacity utilization. The regression equations were formulated by the least square regression method. Most of t
