Supporting Information for “Development of new photon-counting detectors for single-molecule fluorescence microscopy” Supporting Information for “Development of new photon-counting detectors for single-molecule fluorescence microscopy” by X. Michalet

SBR does not depend on the measurement duration, whereas SNR increases with the measurement duration and SBR. These are well known results and do not need further discussion (1-2). However, these simple expressions also show an interesting connection between signal count rate (s) and background count rate (b) illustrated in Fig. S6. Fig. S6 shows curves of constant SNR at a given SBR, as a function of relative variation in background and signal count rates. Starting from signal count rate s0 and background count rate b0, these curves indicate what is the relative signal count rate increase s/s0 needed to cancel out a relative increase in background count rate b/b0. For instance, a twofold increase in background count rate can be compensated by a 30% signal count rate increase if the original SBR is equal to 1, but by only a 10% signal rate increase (respectively 3%) if the starting SBR is equal to 10 (respectively 30). In other words, in an experimental situation where the typical SBR is ~30, if the detector quantum efficiency can be increased by ~3% with a less than 2-fold corresponding increase in background count rate, an increased SNR will result. Note that in doing so, the final SBR will be reduced (by a factor Q/Q0×b0/b ~ 2). Obviously, the usefulness of a minor increase in SNR paid for by a large decrease in SBR needs to be evaluated on a case-by-case basis. The cases of single-molecule burst detection and single-molecule imaging are discussed in separate sections below.