Minimization of composition bias in beta transmission thickness gauges

The beta-particle intensity transmitted through sheet material exhibits a small but significant variation that depends on composition of the material. This causes errors in on-line thickness gauges when a gauge is used for measurement of a substance having a different composition from that used for calibration. The problem is particularly troublesome in aluminum alloy sheet, where thickness accuracies of better than 0.5% are desired, and many different alloys are measured on a single instrument. First, accurate calibration samples are expensive to produce, validate, and maintain, and only a fraction of the existing aluminum alloys possess adequate suites of well-characterized thickness standards. Second, even within-alloy composition variations (i.e., variations allowed by the composition specification) can result in significant errors, e.g., at least 50 microinches in 11-mil can stock (1 mil = 10-3 in. = 0.00254 cm). If a functional dependence between composition change and thickness error were found, then composition-induced bias could be corrected. Gauge calibration could be updated according to the known composition of alloys (or even individual ingots) for which calibration samples are unavailable