Ion-Induced Nucleation: The Importance of Chemistry Shawn M. Kathmann, Gregory K. Schenter, and Bruce C. Garrett Chemical Sciences Division, Pacific Northwest National Laboratory, Richland, Washington 99352, USA (Received 5 October 2004; published 24 March 2005) Experiments have shown that ions can substantially increase vapor-to-liquid nucleation rates. However, interpretation of these experiments is complicated by ambiguities arising from the manner in which the ions are produced. Several studies have concluded that water has a general preference for anions over cations. We show that specification of the ion’s sign alone is insufficient to provide an understanding of the aqueous ionic cluster thermodynamics and that classical ion-induced nucleation theory does not treat the cluster physics properly to describe ion-induced nucleation accurately. DOI: 10.1103/PhysRevLett.94.116104 PACS numbers: 82.60.Nh, 64.60.Qb, 68.03.-g, 82.20.-w Nucleation requires surmounting an activation barrier via rare event processes. The presence of trace species— especially ions—can reduce this barrier and increase the rate of formation of ionic embryos [1–23]. Considerable attention has been paid to the sign effect, the relative en- hancement of nucleation rates due to an ion’s sign, without reference to chemical identity [1–6,8,11–15]. Wilson dis- covered that x rays induced droplet formation in super- saturated water vapor. Employing suitable electric fields, he found that water prefers ‘‘anions’’ to ‘‘cations’’—but the chemical identity of these ions was never determined. Subsequent ion-induced nucleation experiments have pro- duced a variety of interesting results and interpretations, sometimes contradictory, depending on physical condi- tions, ion chemical identity, and electric field strength. The majority of these studies found that ion-induced nu- cleation is more favorable than homogeneous nucleation of the host vapor and confirmed a ‘‘negative’’ sign preference for water. These studies [1–6,8–15], however, produced ions via  rays, x rays,  particles, and nonresonant UV photoionization, resulting in a cascade of ions whose chemical identities and lifetimes were unknown [7]. Definitive experimental resolution of whether anions or cations enhance nucleation rates is difficult because it is not possible to simply change an ion’s charge without altering its electronic structure and thus its chemical prop- erties. Castleman and Tang [7] stated over 30 years ago ‘‘the sign of the charge is not the only prerequisite and even ions of like sign may nucleate at different supersaturation ratios. The interpretation of the data obtained in most ion- nucleation experiments is therefore very uncertain and appropriate account of the actual species is necessary in understanding the molecular nature of the nucleation phe- nomena.’’ This statement is in stark contrast to the con- clusion [15] that ‘‘understanding the effect of ions would be very difficult, or even impossible, if the ion’s specific chemical characteristics had a significant effect on their nucleating efficiency.’’ Further underscoring the impor- tance of chemistry, Castleman suggested, based on electron affinities and ionization potentials, that halides and metals would be appropriate primary negative and positive ions, respectively. Indeed, experimental data [24,25] on ionic hydrates explicitly demonstrate the sensitivity of monohy- dration enthalpies on an ion’s chemical identity (for ex- ample, Li  34:0, Na  24:0, K  17:9, Rb   15:9, Cs  13:7, F  23:3, Cl  13:1, Br   12:6, I  10:2; all energies in kcal=mol). Theoretical approaches offer a means of isolating differ- ent effects, such as the sign of the ionic charge, while keeping all other factors constant. The first theory of ion- induced nucleation modified classical nucleation theory (CNT) [26] using the electrostatic model of Born [27] and Thomson [28] to yield classical ion-induced nucleation theory (CIINT) [29,30]. CIINT models cluster thermody- namics using the bulk liquid surface tension, density, and dielectric constant. Although CIINT does predict that ions have large effects on nucleation rates, it does not explicitly treat cluster chemical physics. Nadykto [23] addressed discrepancies between ion-induced nucleation experiments and CIINT by including a mean field charge-dipole free energy to account for the host vapor dipole interaction with the electric field of the seed ion. Although this effect may be important, it does not address the inherent limitations of CIINT. Moreover, it is unclear how these modifications could be validated against experimental results using chemically unidentified ions. Molecular theory and simulation provide a means to examine the chemical nature of ions, not just the sign effect, on the nucleation process. Kusaka [17,18] predicted a sign effect for a simple attractive hard sphere molecule with multipole moments using a mean field density func- tional theory to compute the critical cluster reversible work. They found a sign preference resulting from an asymmetry in the electrostatic interaction. However, their approach is semiquantitative in model representation and theoretical treatment. Oh [20] performed Monte Carlo simulations of metastable water vapor to obtain the work of formation for aqueous ionic clusters in which only the sign of the ion was changed. Their studies concluded ‘‘generally’’ that water prefers anions—consistent with Wilson’s early observations on unidentified ‘‘anions.’’ Furthermore, it is difficult to compare and contrast their PRL 94, 116104 (2005) PHYSICAL REVIEW LETTERS week ending 25 MARCH 2005 0031-9007= 05=94(11)=116104(4)$23.00 116104-1  2005 The American Physical Society