Selection of integration time intervals for quartz OSL decay curves

In quartz optically stimulated luminescence (OSL) dating protocols, an initial integral of the OSL decay curve is used in the calculation of equivalent dose, once a background integral has been subtracted. Because the OSL signal commonly contains a number of exponentially decaying components, the exact choice of time intervals used for the initial-signal and background integrals determines the composition of the net signal. Here we investigate which combination of time intervals will produce the net signal most dominated by the fast OSL component, while keeping an acceptable level of precision. Using a three-component model of OSL decay, we show that for a specified level of precision, the net signal most dominated by the fast component can be obtained when the background integral immediately follows the initial signal and is approximately 2.5 times its length. With this 'early-background' approach, the contribution of slow components to the net signal is virtually zero. We apply our methods to four samples from relatively young deposits. Compared to the widely used 'late-background' approach, in which the background integral is taken from the last few seconds of OSL, we find less thermal transfer, less recuperation and a higher proportion of aliquots yielding an equivalent dose in agreement with expectations. We find the use of an early background to be a simple and effective way of improving the accuracy of OSL dating, and suggest is should be used in standard protocols