Shelved atom experiments appear to show clear evidence for quantum jumps. We use a gedanken interference experiment to demonstrate that a fundamental temporal ambiguity in the photon time-of-emission occurs in such transitions.
The observation of quantum jumps in a Dehmelt electron-shelving scheme was made two years ago.1 Prior to these experiments considerable theoretical discussion was addressed to the question of whether the turn-on and turn-off of fluorescence, as described by Cook and Kimble,2 would actually occur. This discussion focused on photon statistics. It was taken to be self-evident that a prolonged cessation of fluorescence from the strong transition implied that the atom had jumped and now resided in the shelved state. This is surely correct but with some qualification. What observational information leads to the conclusion that the electron is shelved? It is the appearance of an abnormally long time interval without recording a single photoelectron at the detector monitoring the strong fluorescence. But how long is abnormally long? What is the time resolution of our quantum jump? I calculate the reduced atomic state following a photoelectric detection event, conditioned on the requirement that no subsequent photoelectron has been observed. For sufficiently long times without recording a subsequent photoelectron the atomic state collapses to the shelved state with high (but not unit) probability. The waiting time required for a confident judgment that the atom is in the shelved state is calculated as a function of transition rates and detector efficiency.
We consider a three-level atomic system driven strongly on one transition and weakly on the other. The excited state on the weak transition is assumed to be metastable. We give an analysis of the fluorescence from the strong transition in terms of the elementary probability density p[0,t)(t1,t2,..., tn) which gives us the probability density that exactly n photons are emitted at times t1,t2,...,tn by the atom in the time interval [0,t). We show that p[0,t) essentially factorizes into products of conditional densities c\ifmmode \tilde{}\else \~{}\fi{}(\ensuremath{\tau}) that, given a photon is emitted at time zero, the next photon emission occurs at time \ensuremath{\tau}. This enables a simulation of the individual photon emissions to be given which shows directly the existence of prolonged dark windows in the fluorescence corresponding to the shelving of the electron in the metastable state or ``quantum jumps.''