VIBFREQ1295: A New Database for Vibrational Frequency Calculations.

Area of Science:

• Computational quantum chemistry
• Astrochemistry
• Spectroscopy

Background:

• High-throughput vibrational spectral data for astrochemistry often uses harmonic frequency calculations.
• Lack of established model chemistry recommendations necessitates benchmarking.
• Accurate prediction of fundamental frequencies requires accounting for anharmonicity.

Purpose of the Study:

• To present VIBFREQ1295, a new database for vibrational frequency calculations.
• To provide a benchmark for computational quantum chemistry methods.
• To establish reliable scaling factors for predicting fundamental frequencies.

Main Methods:

• Compiled 1295 experimental fundamental frequencies and computed 141 molecules' harmonic frequencies using CCSD(T)(F12*)/cc-pVDZ-F12.
• Analyzed chemical space representation for common organic functional groups and vibrational modes.
• Determined optimal scaling factors for different frequency regions.

Main Results:

• A single global scaling factor yielded median differences of 15.9(5) cm⁻¹.
• Region-specific scaling factors (below 1000, 1000-2000, above 2000 cm⁻¹) significantly improved accuracy, with median differences of 7.5(5) cm⁻¹.
• Established a lower bound for the performance of scaled harmonic frequency calculations.

Conclusions:

• VIBFREQ1295 serves as a robust database for benchmarking vibrational frequency calculations.
• The database can train machine-learning models for vibrational spectra prediction.
• Provides a reference for detailed spectroscopic modeling.