The Fluoride Ion Affinity Revisited: Do We Need the Anchor-Point Approach?

A large and diverse high-level benchmark data set of computed gas-phase fluoride ion affinities (FIAs) for 71 small main-group Lewis acids is presented. It has been used to evaluate quantitatively DFT approaches with 52 functionals and 4 composite methods. Two widely used indirect anchor-point methods based on isodesmic reactions with fluorophosgene or the trimethyl silyl cation are compared to the direct computation of the FIA. It has been frequently stated that anchor-point methods are to be strongly preferred over direct FIA computations at DFT levels, as they avoid treatment of the naked fluoride ion. Here it is shown that this widespread assumption does not hold when modern functionals with low self-interaction errors and suitable basis sets with diffuse functions are used. In these cases, an anchor-point approach based on ${\mathrm{OCF}}_2$ has little or no advantage, and the widely used anchor-point calculations based on ${\mathrm{Me}}_3{\mathrm{Si}}^{+}$ even deteriorate results in most cases. It is shown that this is due to a break-down of often prevailing error cancellations in the anchor-point approach that help to improve results when using less suitable functionals or basis sets. Overall, the direct computation of FIAs at appropriate DFT levels including diffuse basis functions is the clearly preferable route.