A semiconductor industry exposure reconstruction and risk assessment

A high quality study was conducted to determine exposure profiles for three historical semiconductor manufacturing lines.  Exposures to non-carcinogens as well as carcinogens were characterized.  Exposure was assumed for any process area in which a Group 1 human carcinogen recognized to cause leukemia was used in the  process or was potentially present as a by-product. Exposure was considered likely in a process area in which a Group 2 (probable or  possible human) carcinogen was measured at or above 50% of the OEL.  We characterized exposures in 35 SEGs, reconstructed exposures for the three lines using QLRAs, Bayesian statistics and quantitative  data. With regard to a historical and now obsolete process, we also modeled exposures using 2-zone, near field/far field techniques.  We combined the exposure profiles with time, frequency, proximity, ventilation, and employment duration to reconstruct doses. We  compared the reconstructed average and cumulative exposures with exposures reported in epidemiological studies for the purpose  of estimating risk.  We formed a number of opinions and drew a number of conclusions:  • Reconstructed exposures yield dose calculations which were generally low or trivial.  • Excess lifetime risks were within the range of acceptable excess lifetime risks (a range between 1x10-6 and 1x10-4) for the general  population.  High quality exposure assessments are required for health studies where exposure data are largely censored.