LCA in the Semiconductor Industry is Lagging: We Need Better and More-Accessible Data

Modern electronics are present in every walk of life, but we don’t currently have the information needed to understand their environmental footprint

For many years, the semiconductor industry in general has been somewhat insulated from market demands for carbon and water footprints and other environmental impact assessments. This trend has started to shift as investors see global warming and other environmental impacts as potential risks to their investments, putting demands directly on the industry. The supply chain is complex, however, and in many ways, carbon footprints and broader impact assessments are in the early ages of LCA, where data was hard to come by and proxies were sketchy at best. The problem becomes acute for the electronics industry where semiconductors are among the largest contributors to environmental impacts; a situation further complicated by the global nature of the electronics supply chain where a component may be started in China, completed in Malaysia, tested in the U.S., assembled into a product in Vietnam and sold in Europe. Add to that the fact that products have something around an 18 to 36 month month lifetime and by the time any detailed assessment is complete the product is out of date.

The news isn’t awful, though. There are data out there. With the support of Applied Materials, one of the largest suppliers of semiconductor manufacturing equipment, Sarah Boyd performed an amazing, thorough analysis of the semiconductor manufacturing process as her PhD thesis, and while the data are from 2009, they are still useful today. Needless to say, there are a lot of assumptions in those models that need to be refined and updated. Cynthia Murphy, et al., developed a parameterized model of one fundamental semiconductor process, showing that it is possible to make flexible models that can be modified fairly quickly as new devices are developed.

The parameterized model is available in the DataSmart library available in SimaPro and OpenLCA. Much of Dr. Boyd’s work has been incorporated into the electronics data available for GaBi.

Building up from the semiconductors, in addition to the GaBi datasets, several other datasets have been developed. The PAIA consortium, led by Quantis Intl. and based on models developed at MIT, has developed underlying data for product-specific GHG tools, however, these are only available to consortium members and the underlying data are not current and are not available, even to members. iNEMI has developed the Eco-impact estimator for GHGs based on clever algorithms to estimate impacts, but is only available to members. Ecoinvent has some data on commonly used electronics, useful for broad-brush analyses and sensitivity, but less useful for product-specific analysis. Fraunhofer IZM has expanded upon the GaBi data and also developed tools for simplified GHG assessment.

In reality, however, with the exception of Dr. Boyd’s work, none of this is very current or good for anything beyond a rough estimate of GHG impacts. While the extensive use of both water and toxics in the semiconductor industry are well known, it is difficult to perform a water footprint or understand to any extent the effects of the toxics usage. Further, one needs to have access to a specific tool or be involved with a specific group to have access to the data.

What is truly needed is a bottoms-up data development program. Working with companies like Applied Materials, parameterized models can be developed for each process used in a semiconductor fabrication center. These models can then be used by the manufacturers to develop inventory results for ICs and other devices. More data will be needed from the specialty chemical suppliers and others in the supply chain to complete this effort.

More current, accessible data for passive components, connectors and the PC boards is less critical, as these play a minor role in the impacts of most electronics. They do play a role, however, and need to be included in an electronics LCI database. Fortunately, data for other components such as case materials, display materials, metals, and polymers is already up to date.

This type of effort can support all the existing work—providing updates to the GaBi data and supporting the PAIA, ecoinvent, Eco-Impact Estimator and Fraunhofer efforts.

Considering the impact that electronics have on, and their growing ubiquity in, our lives it’s time to consider their impacts using high quality data so that we can truly understand its contribution to the environment.