Abstract
Dilute solutions of the alkali metals in liquid ammonia are assumed to consist of the following constituents: M+, the solvated metal cation; S, the solvent; M−, the solvated metal anion; S−, the solvent anion (solvated electron); the ion pairs M+·M− and M+·S−. These species are assumed to be in equilibrium via
M++2S−⇌M−+2S
M
+
+2
S
−
⇌
M
−
+2
S
and
M++A−⇌M+·A−,
M
+
+
A
−
⇌
M
+
·
A
−
,
where A− is either M− or S−. For simplicity, the activity coefficient of each species is assumed to be given by the extended theory of Debye and Hückel, with a fixed value of 5.5 Å for the interionic size parameter; the ion‐pairing theory of Fuoss, with the same value for the interionic size parameter, is also assumed. The only adjustable parameter then remaining for a complete characterization of the composition of dilute alkali‐metal—ammonia solutions is the redox equilibrium constant.
With generally good results being obtained, the foregoing model has been applied to account for the following properties of alkali‐metal—ammonia solutions which are more dilute than about 0.10M: (1) the spin paramagnetic susceptibilities reported by Hutchison and Pastor; (2) the vapor‐pressure‐depression results of Kraus; (3) the electrical conductances measured by Kraus and the temperature coefficients of conductance, reported by Gibson and Phipps; (4) the Knight shifts for 14N and 23Na, reported by McConnell and Holm, Acrivos and Pitzer, and O'Reilly; (5) the optical‐absorption results of Gold and Jolly, and those of Dye and Douthit. An a posteriori argument supports the neglect of ion triples for the solutions which have been considered, and reinforces a distinction to be made bewteen the metal anion Na− and the analogous ion triple S−·Na+·S−.