Precision Prediction of NMR Shielding Constants and Chemical Shifts

A benchmark study from 2026 evaluates XYG3-type doubly hybrid density functionals for predicting nuclear magnetic resonance parameters. The work by Xinming Lai compares eight xDH@B3LYP variants and xDH-PBE0 against the NS372 dataset of main-group species, using CCSD(T)/pcSseg-3 as a reference. Results show the XYGJ-OS variant provides the best overall balance with mean absolute deviations of 2.40 and 1.77 ppm for shielding constants and chemical shifts, respectively.

Use Cases

• Benchmarking density functional approximations for NMR property prediction based on the NS372 dataset.
• Developing new density functionals targeting magnetic-response properties based on performance analysis of xDH variants.
• Rationalizing shielding behavior based on the balance between opposite-spin and same-spin PT2 contributions described in the study.

Strengths

• Benchmarks against the curated NS372 dataset of main-group species.
• Reports specific performance metrics, including a mean absolute deviation of 1.77 ppm for chemical shifts with the XYGJ-OS variant.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Author: Xinming Lai. Platform: figshare.

Collection Method

Results from computational benchmarking using analytic gauge-including atomic orbital (GIAO) implementation for XYG3-type doubly hybrid density functionals.

Freshness

Last updated 2026-05-16 14:16:56; freshness should be verified.