A quantum mechanical study of the reactivity of $(SiO)2$-defective silica surfaces

The reactivity of the strained $(SiO)2$-four atom ring defect at the silica surfaces has been studied in a cluster approach adopting the $ONIOM2[B3LYP∕6-31+G(d,p):MNDO]$ method to compute the ring opening reaction by interaction with $H2O$ and $NH3$⁠. The vibrational “fingerprints” of the isolated defect are computed at 921, 930, and $934cm−1$ in reasonable agreement with experimental evidence on amorphous silica outgassed at $T>900K$⁠. The opening of the $(SiO)2$-four-member ring by the considered molecules is exergonic and the actual value depends on the possible constraints enforced on the reaction products by the silica surrounding. The free kinetic energy barriers result from the interplay between the nucleophilic/electrophilic character of the adsorbed molecule and are 22 and $25kcalmol−1$ for $NH3$ and $H2O$⁠, respectively. All free energy profiles envisage an activated complex in which the nucleophilic part of the molecule interacts on the coordinatively strained silicon atom of the $(SiO)2$ defect followed by the proton transfer from the coordinated molecule towards the oxygen of the defective ring. Calculations show that this step can be speed up by the presence of more than one adsorbed molecule or even more (about seven orders of magnitude), by the copresence of water molecules acting as “proton transfer helpers.” In these cases, the free energy barriers decrease to $≈13$ and $15kcalmol−1$ for $NH3$ and $H2O$⁠, respectively. For the case of $H2O$ adsorption, benchmark test calculations reveal that MP2, BLYP, and B3LYP energy profiles are in very good agreement with each other, whereas for PBE, both the reaction energy and the activation barrier are underestimated. Present data also show that the molecular model mimicking the $(SiO)2$ defect is far less reactive than what appears to occur on the real defect at the surface of amorphous silica. So, only a combination of some further geometrical strains imparted by the solid on the $(SiO)2$ defect, not accounted for by the cluster models, and higher adsorbate loadings are needed to reharmonize experiment and simulation. Notwithstanding, the vibrational features of the reaction products have been characterized and support the available experimental measurements.