A hypothesis concerning FAB mechanisms, referred to as a 'bubble chamber FAB model', is proposed. This model can provide an answer to the long-standing question as to how fragile biomolecules and weakly bound clusters can survive under high-energy particle impact on liquids. The basis of this model is a simple estimation of saturated vapour pressure over the surface of liquids, which shows that all liquids ever tested by fast atom bombardment (FAB) and liquid secondary ion mass spectrometry (SIMS) were in the superheated state under the experimental conditions applied. The result of the interaction of the energetic particles with superheated liquids is known to be qualitatively different from that with equilibrium liquids. It consists of initiation of local boiling, i.e., in formation of vapour bubbles along the track of the energetic particle. This phenomenon has been extensively studied in the framework of nuclear physics and provides the basis for construction of the well-known bubble chamber detectors. The possibility of occurrence of similar processes under FAB of superheated liquids substantiates a conceptual model of emission of secondary ions suggested by Vestal in 1983, which assumes formation of bubbles beneath the liquid surface, followed by their bursting accompanied by release of microdroplets and clusters as a necessary intermediate step for the creation of molecular ions. The main distinctive feature of the bubble chamber FAB model, proposed here, is that the bubbles are formed not in the space and time-restricted impact-excited zone, but in the nearby liquid as a 'normal' boiling event, which implies that the temperature both within the bubble and in the droplets emerging on its burst is practically the same as that of the bulk liquid sample. This concept can resolve the paradox of survival of intact biomolecules under FAB, since the part of the sample participating in the liquid-gas transition via the bubble mechanism has an ambient temperature which is not destructive for biomolecules. Another important feature of the model is that the timescale of bubble growth is no longer limited by the relaxation time of the excited zone ( approximately 10(-12) s), but rather resembles the timescale characteristic of common boiling, sufficient for multiple interactions of gas molecules and formation of clusters. Further, when the bubbles burst, microdroplets are released, which implies that FAB processes are similar to those in spraying techniques. Thus, two processes contribute to the ion production, namely, release of volatile solvent clusters from bubbles and of non-volatile solute from sputtered droplets. This view reconciles contradictory views on the dominance of either gas-phase or liquid-phase effects in FAB. Some other effects, such as suppression of all other ions by surface-active compounds, are consistent with the suggested model.