Life-Cycle Assessment of Semiconductors

Abstract

This thesis is the first complete and transparent study of the life-cycle environmental impacts of semiconductor chips using process-level data, as well as the first analysis of the changes in these impacts over time. LCA of complementary metal oxide semiconductor (CMOS) logic, flash memory and dynamic random access memory (DRAM) are presented. CMOS logic is the most common form of digital logic used today. This thesis provides a life cycle assessment for CMOS chips over 7 technology generations with the purpose of comparing impacts by life cycle stage, examining their trends over time and evaluating their sensitivity to data uncertainty and changes in production metrics such as yield. A hybrid life cycle assessment (LCA) model is used; Wafer production, electricity generation, water supply and certain materials are represented by process LCA data, while the remaining materials are described using economic input-output (EIO) LCA methods. It is determined that, in the case of CMOS logic, life-cycle impacts in all but one category are dominated by the use phase, and impacts are most sensitive to those variables which define use phase energy demand (chip power demand, usage patterns, power supply efficiency and chip lifetime). Using the same methodology, LCA of flash memory over 5 technology generations is presented. The most recent generation of flash memory is compared with magnetic storage, using a laptop hard-drive as a functional unit, and it is determined that in most impact categories, a flash-based drive will result in fewer impacts. Life-cycle impacts of DRAM, over 6 technology generations, are presented using as the functional unit, the memory requirements a popular operating system in each year of production. The influence of the choice of functional unit on results in semiconductor LCA is examined, and it is argued that functional unit is best defined as a computational power or memory capacity required for a given function, within a set time period. The LCA impact and inventory results for these three types of semiconductor products allow more accurate assessment of the environmental ben-1 efit or costs of information technology relative to traditional products and services.

Developing a parametric carbon footprinting tool for the semiconductor industry

“…Moreover, the calibration factors (i.e., carbon emission factor of electricity and OWE) can be adjusted in the proposed parametric carbon footprinting tool to strengthen the application. Although several fast carbon footprinting approaches for the semiconductor industry have been proposed in previous studies (Boyd et al 2006;Dessouky et al 2011;Villard et al 2012), which integrated material and energy consumption with carbon emission at each equipment unit of the manufacturing process, these approaches are suitable for use by a semiconductor company that is capable of completely in-house production, from design to the manufacturing stage. However, this approach may be inapplicable to some semiconductor companies that have separate design and manufacturing stages due to the confidentiality of this internal information of other enterprises.…”

“…The quantity of mask layers represents the level of complexity for wafer (Taiariol et al 2001;Yao et al 2004). Boyd et al (2006) implemented life cycle inventory for a comparison between a six-layer wafer and an eight-layer wafer with the same specifications. The results indicated that the increase in the quantity of layers and the consumption of energy and material will affect the wafer's CFP.…”

“…However, these companies should finalize CFP calculation before developing their own parametric tools. Calculating CFP may not be a considerable problem because several life cycle inventory approaches for semiconductor manufacturing have been proposed in previous studies (Murphy et al 2003;Boyd et al 2006Boyd et al , 2009. Second, the parametric tool proposed in this study can be applied to six different wafer function types.…”

“…Although PAS 2050(BSI 2011 and the latest ISO/TS 14067 (ISO 2013) have helped develop carbon footprint of product (CFP) as a sub-discipline of LCA, performing CFP of IC is often unattainable because of the relatively short development cycle and rapid change in design and manufacturing technologies. Thus, the assessment of existing chips has become obsolete in a few years (Harland et al 2008;Boyd 2012). The manufacturing process of semiconductors from upstream to downstream of the value chain is composed of four phases, they are as follows: IC design, wafer fabrication, IC test, and IC packaging (Wang and Chiu 2014).…”

“…Some researchers have proposed methodologies to achieve fast carbon footprinting in the semiconductor industry. Boyd et al (2006) designed a tractable, parametric, and secure structure to evaluate energy use, material input, and emissions data, such as carbon footprint, to comparatively assess the environmental implications of semiconductor fabrication. Dessouky et al (2011) proposed a user-friendly carbon footprint model to promote student awareness of the effect that changes in manufacturing technologies and transportation modes could have on the carbon footprint of the semiconductor industry's supply chain.…”