Afterpulse measurement and correction

Abstract

A pulse of electrons in a photomultiplier sometimes generates succeeding spurious pulses, called afterpulses. These are detrimental to statistical analysis of the distribution of photocounts. A method is outlined to determine the probability of occurrence of afterpulses, by analysis of the frequency distribution of output pulses produced by constant light. A related method is outlined to correct for the effect of afterpulsing, so that the original distribution of photocounts can be determined, even if the light is fluctuating. This correction technique is tested by application to measurements of twinkling starlight and shown to work very well. The method to deduce afterpulse probabilities is developed further to produce values in a form which allows comparison with previous measurements. For one tube the results agree with expectations, but not for another. A plausible reason is given for the discrepancy, illustrating the potential of the method to give information about the operation of photomultipliers.