Common mistakes in molecular modeling involving B3LYP functional

Welcome to atomistica.online

Articles & Tutorials

Common mistakes in molecular modeling involving B3LYP functional

If you find this article useful, please consider supporting the development of atomistica.online by citing:

Stevan Armaković and Sanja J. Armaković
ATOMISTICA.ONLINE – WEB APPLICATION FOR GENERATING INPUT FILES FOR ORCA MOLECULAR MODELING PACKAGE MADE WITH THE ANVIL PLATFORM
Molecular Simulation, 49 (1) (2023), 117 – 123, DOI: 10.1080/08927022.2022.2126865
Download citation as .ris file (right click – save link as)

and

Stevan Armaković and Sanja J. Armaković
ONLINE AND DESKTOP GRAPHICAL USER INTERFACES FOR XTB PROGRAM FROM ATOMISTICA.ONLINE
Molecular Simulation, 50 (2024), 560 – 570, DOI: 10.1080/08927022.2024.2329736
Download citation as .ris file (right click – save link as)

Also, feel free to support this project by becoming a patron at https://patreon.com/atomista

Introduction

As already pointed out on this platform, molecular modeling is a powerful tool in the field of computational materials science, allowing scientists to gain valuable insights into the behavior of molecules and materials at the atomic level. One of the most widely used density functional theory (DFT) methods is the B3LYP functional. However, even experienced researchers can sometimes make common mistakes when applying this functional. In this article, we will explore three common mistakes made by using the B3LYP functional, and provide accurate alternatives.

Mistake #1: Optimization When Non-covalent Interactions Take Place

Non-covalent interactions, including hydrogen bonding, van der Waals forces, and π-π stacking, play a pivotal role in shaping the stability and properties of molecules and materials. Many newcomers to molecular modeling often fall into the trap of utilizing the B3LYP functional for geometry optimization without considering its limitations when dealing with non-covalent interactions. The B3LYP functional has been observed to overestimate the strength of non-covalent interactions, resulting in inaccurate geometries and energetics. To tackle this issue, researchers typically explore one of two solutions:

Empirically derived corrections: One approach to mitigate the limitations of the B3LYP functional when dealing with non-covalent interactions is to apply empirically derived corrections. Empirically corrected density functionals are designed to account for non-covalent interactions by incorporating additional terms into the energy equation. These empirical terms are derived from experimental data or high-level calculations and are added to the standard functional to improve its accuracy in describing non-covalent interactions. The most famous empirically corrected solution is the B3LYP-D3. More about dispersion-corrected functionals will be presented in a separate article.

The use of density functionals specifically designed to reproduce electronic density more accurately: Another approach is to use density functionals that are explicitly designed to reproduce electronic density more accurately, especially for non-covalent interactions. Functionals like M06-2X are prime examples of such specialized functionals.

Empirically corrected density functionals offer a practical solution. They provide geometries very close to those of higher levels of theory, at only a slightly higher computational cost compared to regular density functionals. Conversely, specialized density functionals, such as M06-2X or ωB97X-D, are more accurate in describing electronic density, particularly for non-covalent interactions. However, they come with a significantly higher computational cost. Therefore, if you are dealing with relatively large molecular structures, it is advised to use dispersion-corrected functionals for geometrical optimizations, then you can continue with single-point energy calculations at a higher level of theory.

Mistake #2: Calculations of Excitations and Simulation of UV Spectra

Another common mistake in molecular modeling is the use of the B3LYP functional for electronic excitation calculations and simulating UV spectra. While B3LYP is a reliable choice for certain geometry optimizations and ground-state properties, it fails to accurately predict excited-state properties, such as electronic transitions and UV-visible spectra.

B3LYP‘s shortcomings become evident when it comes to excitation energies, as it tends to either underestimate or overestimate them, leading to significant discrepancies when compared to experimental data. Researchers should explore density functionals explicitly designed for this purpose to achieve precise and reliable results in calculations involving excited-state properties.

Two notable alternatives are the long-range corrected functionals CAM-B3LYP and ωB97X-D. These functionals incorporate specific corrections to address the limitations of standard functionals like B3LYP when dealing with excited states. CAM-B3LYP, for instance, includes a range-separation parameter that improves the description of long-range electron correlation, making it well-suited for predicting excited-state properties accurately. Similarly, ωB97X-D incorporates dispersion corrections, which enhance its ability to handle non-covalent interactions as well.

By choosing the appropriate density functional for electronic excitation calculations, researchers can significantly enhance the agreement between computational predictions and experimental data, ultimately leading to more reliable and insightful results in studies involving UV spectra and electronic transitions.

Mistake #3: Assuming B3LYP Is Always the Best Choice for Energy Calculations

A common misconception among beginners in molecular modeling is the belief that B3LYP is a universally reliable functional suitable for energy calculations. However, alternative functionals may offer better accuracy depending on the research goals and the nature of the systems under investigation. Due to the vast number of various density functionals available, suggesting one or a few for obtaining accurate energies can be challenging and potentially misleading. Nevertheless, it is essential to highlight the Minnesota family of density functionals, with M06-2X being one of the most well-known representatives. M06-2X is frequently recommended for thermochemistry in several research papers.
However, it’s important to note that beyond DFT, highly accurate wavefunction-based methods are available for calculating the energies of molecules. These methods, often referred to as “post-Hartree-Fock” methods, among others include:

Coupled Cluster Theory (CC): CC methods, such as CCSD (Coupled Cluster with Singles and Doubles) and CCSD(T) (Coupled Cluster with Singles, Doubles, and Perturbative Triples), provide exceptionally accurate energy predictions by systematically including electron correlation effects. They are considered the “gold standard” for electronic structure calculations but are computationally demanding and are typically used for smaller systems.
Møller-Plesset Perturbation Theory (MP2): MP2 is a widely used and reasonably accurate post-Hartree-Fock method that includes electron correlation effects through perturbation theory. It is computationally more affordable than CC methods while still providing good accuracy for many systems.