Using optical microscopy, we investigated the efflorescence of ammonium sulfate (AS) in aqueous AS and in aqueous 1:1 and 8:1 (by dry weight) poly(ethylene glycol)-400 (PEG-400)/AS particles deposited on a hydrophobically coated slide. Aqueous PEG-400/AS particles exposed to decreasing relative humidity (RH) exhibit a liquid liquid phase separation below similar to 90% RH with the PEG-400-rich phase surrounding the aqueous AS inner phase. Pure aqueous AS particles effloresced in the RH range from 36.3% to 43.7%, in agreement with literature data (31-48% RH). In contrast, aqueous 1:1 (by dry weight) PEG-400/AS particles with diameters of the AS phase from 7.2 to 19.2 mu m effloresced between 26.8% and 33.9% RH and aqueous 8:1 (by dry weight) PEG-400/AS particles with diameters of the AS phase from 1.8 to 7.3 mu m between 24.3% and 29.3% RH. Such low efflorescence relative humidity (ERH) values have never been reached before for AS particles of this size range. We show that these unprecedented low ERHs of AS in PEG-400/AS particles could not possibly be explained by the presence of low amounts of PEG-400 in the aqueous AS phase, by a potential inhibition of water evaporation via anomalously slow diffusion through the PEG coating, or by different time scales between various experimental techniques. High-speed photography of the efflorescence process allowed the development of the AS crystallization fronts within the particles to be monitored with millisecond time resolution. The nucleation sites were inferred from the initial crystal growth sites. Analysis of the probability distribution of initial sites of 31 and 19 efflorescence events for pure AS and 1:1 (by dry weight) PEG-400/AS particles, respectively, showed that the particle volume can be excluded as the preferred nucleation site in the case of pure AS particles. For aqueous 1:1 (by dry weight) PEG-400/AS particles preferential AS nucleation in the PEG phase and at the PEG/AS/substrate contact line can be excluded. On the basis of this probability analysis of efflorescence events together with the AS ERH values of pure aqueous AS and aqueous PEG-400/AS particles aforementioned, we suggest that in pure aqueous AS particles nucleation starts at the surface of the particles and attribute the lower ERH values observed for aqueous PEG-400/AS particles to the suppression of the surface-induced nucleation process. Our results suggest that surface-induced nucleation is likely to also occur during the efflorescence of atmospheric AS aerosol particles, possibly constituting the dominating nucleation pathway.