The phenomenon of ion evaporation from charged liquid surfaces is at the basis of electrospray ionization; a source of a stunning variety of gas phase ions. It is studied here by producing a monodisperse cloud of charged droplets and measuring the charge q and diameter d(r) of the residue particles left after complete evaporation of the solvent. When the droplets contain small monovalent dissolved ions, the electric field E on the surface of their solid residues is found to be independent of d(r). One can thus argue that the source of small ions in electrospray ionization is field-emission, and not other proposed mechanisms such as Dole's charged residue model. A consequence of the observed independence of E on d(r) is that the rate of ion ejection is simply related to the rate of solvent; evaporation, estimated here as that fora clean surface of pure solvent. The reduction G(E) brought about by the electric field E in the activation energy for ion evaporation has thus been inferred asa function of the measured field E in the range 1.5<E(V/nm)<3.25. It agrees surprisingly well with the so-called Schottky hump from the image potential model (IPM), G(IPM)=(e(3)E/4 pi epsilon(0))(1/2)). This remarkably simple result is paradoxical in view of two major objections raised earlier against the use of-the IPM for ion evaporation from liquids. However, the correct mechanism (first introduced by Iribarne and Thomson) leading to an attractive interaction between the liquid surface and the escaping ion is not the creation of an image charge, but the polarization of the dielectric liquid by the ion. In the limit of a large dielectric constant epsilon much greater than 1, the image force and the polarization force coincide numerically, though the later sets in much faster and is apparently free from the paradox raised by Rollgen. Also, the dielectric nature of the liquid and its strong screening of the net charges near its surface resolves another paradox raised by Fenn regarding the discrete distribution of charges. This screening also introduces a correction in the model proposed by Iribarne and Thomson for G(E), making its predictions virtually indistinguishable from those of G(IPM)(E) In conclusion, small ions observed in electrospray ionization are produced by field-emission. Measured ionization rates are well represented by results from a ''polarization potential model'' which appears to he physically sound. These predictions coincide with those from the IPM in the limit epsilon much greater than 1, the only case studied so far. (C) 1995 American Institute of Physics.
