Improved Infrared Spectra Prediction by DFT from a New Experimental Database.

Area of Science:

• Computational Chemistry
• Spectroscopy
• Quantum Chemistry

Background:

• Accurate computation of infrared (IR) spectra is crucial for characterizing gas-phase cations.
• Experimental IR multiple photon dissociation (IRMPD) spectroscopy provides valuable reference data.

Purpose of the Study:

• To improve the computational accuracy of IR spectra for gas-phase cations using Density Functional Theory (DFT) methods.
• To evaluate the performance of various DFT functionals and basis sets for predicting vibrational frequencies.

Main Methods:

• Assessment of five DFT functionals (B3LYP, M06-2X, CAM-B3LYP, LC-BLYP, ωB97X-D) and three basis sets.
• Comparison of computed IR vibrational frequencies with experimental IRMPD spectra for ten Zn and Ru organometallic complexes.
• Evaluation of prediction accuracy using Mean Absolute Error (MAE) and Root-Mean-Square Error (RMSE).

Main Results:

• B3LYP functional demonstrated the lowest MAE (7.1 cm⁻¹) and RMSE (9.6 cm⁻¹).
• Range-separated hybrid (RSH) functionals, like ωB97X-D, showed higher errors (MAE: 12.8 cm⁻¹, RMSE: 16.6 cm⁻¹).
• Employing linear correlations instead of simple scaling factors significantly improved prediction accuracy across all tested functionals.

Conclusions:

• The B3LYP functional is recommended for accurate IR spectra computation of gas-phase cations within the studied range.
• Linear correlations offer a superior approach to scaling computed frequencies compared to traditional methods.
• Significant discrepancies between experimental and computed spectra can occur even with correct structures, necessitating confidence limit definitions.