Predicting Frequency from the External Chemical Environment: OH Vibrations on Hydrated and Hydroxylated Surfaces

Abstract

Robust correlation curves are essential to decipher structural information from IR-vibrational spectra. However, for surface-adsorbed water and hydroxides, few such correlations have been presented in the literature. In this paper, OH vibrational frequencies are correlated against 12 structural descriptors representing the quantum mechanical or geometrical environment, focusing on those external to the vibrating molecule. A nonbiased fitting procedure based on Gaussian process regression (GPR) was used alongside simple analytical functional forms. The training data consist of 217 structurally unique OH groups from 38 water/metal oxide interface systems for MgO, CaO and CeO 2 , all optimized at the DFT level, and the fully anharmonic and uncoupled OH vibrational signatures were calculated. Among our results, we find the following: (i) The intermolecular R(H•••O) hydrogen bond distance is particularly strong, indicating the primary cause of the frequency shift. (ii) Similarly, the electric field along the H-bond vector is also a good descriptor. (iii) Highly detailed machine learning descriptors (ACSF, SOAP) are less intuitive but were found to be more capable descriptors. (iv) Combinations of geometric and QM descriptors give the best predictions, supplying complementary information.

Water in Crystals: A Database for ML and a Knowledge Base for Vibrational Prediction

“…It is also interesting to note that the difference between the resulting RMSE values for the R­(O···O) and R­(H···O) descriptors in Figure a is smaller than for the same descriptors in the often quite distorted water H-bonds present on metal oxide surfaces (cf. ref ). This can be understood from the angle distribution being more concentrated toward straighter hydrogen bonds in the crystalline bulk data (for completely straight bonds R­(O···O) = R­(H···O) + r­(OH)).…”

“…The situation is quite different for the electric field descriptors where water and hydroxides in our data set follow two disjunct distributions (Figure b). This finding is consistent with water and hydroxides on metal oxide surfaces as presented in ref . To achieve a good prediction using the electric field descriptors for hydroxide species, a separate regression must be performed.…”

“…4a is smaller than for the same descriptors in the often quite distorted water H-bonds present on metal oxide surfaces (cf. ref 35). This can be understood from the angle distribution being more concentrated toward straighter hydrogen bonds in the crystalline bulk data (for completely…”

“…Here, we will make use of the Behler−Parrinello atom-centered symmetry function (ACSF) descriptor 36 and label it a machine-learning descriptor because it is much used in machine-learning applications to generate interatomic potentials. The ACSF descriptor and another machine-learning descriptor, smooth overlap of atomic positions (SOAP), 37 both came out favorably in our recent study of surface OH oscillators on hydrated surfaces in ref 35.…”

Composition-dependent morphologies of CeO2 nanoparticles in the presence of Co-adsorbed H2O and CO2: a density functional theory study

“…34 It has been shown that small-molecule adsorbates such as CO 2 and H 2 O can stabilise the {100} facet over the {111} facet and therefore inhibit the formation of more stable octahedral nanoparticles. 25,26,35 The interaction of individual molecules such as H 2 O 26,[36][37][38][39][40] and CO 2 25,[41][42][43] with CeO 2 surfaces has thus drawn considerable interest. The behaviour in the presence of CO 2 is critical in automotive three-way catalytic converters, as the reactions progress through surface-bound CO 2 .…”