Updated Tkatchenko Scheffler method and assessment of dispersion methods for alkali-containing systems

summary

The gist

The present work examines the accuracy of the Tkatchenko-Scheffler method with updated van der Waals radii (abbreviated as TS 2018)…

In short

This study compares the Tkatchenko-Scheffler (TS) method with updated van der Waals radii (TS 2018) against other dispersion methods for predicting structures in semiconductors. TS 2018 corrects errors in alkali atoms, improving binding energy predictions and lattice anisotropy compared to the original TS method. The results show TS 2018 is highly effective for complex systems involving alkali elements.

Key concepts

Tkatchenko-Scheffler (TS) Method
A widely used technique to add missing long-range van der Waals forces into density functional theory calculations. It works by incorporating pairwise interactions between atoms based on their electron density, making it useful for organic and insulating materials.
TS 2018 Update
An improvement over the original TS method that uses updated van der Waals radii derived from atomic dipole polarizabilities. This revision fixes a major flaw in the older method by accurately estimating the size of atoms, especially alkali metals, leading to more accurate structural predictions.
PBE+MBD NL Method
A non-local many-body dispersion method used for calculating structural properties. It is compared against TS methods and often shows superior accuracy in predicting unit cell volumes for materials like perovskites when using experimental reference data.
Alkali Element Overbinding
A problem where the original TS method incorrectly predicts too strong a binding energy for alkali dimers (like Li2, Na2, etc.). The TS 2018 correction successfully addresses this by using better atomic radii, bringing the predicted binding energies closer to experimental values.

Terminology used across episodes

This episode discusses

The paper

Updated Tkatchenko Scheffler method and assessment of dispersion methods for alkali-containing systems · Read on arXiv

Department of Chemistry, University of North Carolina, Chapel Hill Department of Chemistry, Duke University School of Engineering Brown University Department of Physics and Materials Science at the University of Luxembourg Department of Physics and Astronomy at the University of North Carolina

Transcript

Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.

Kai: We're looking at the paper: "Updated Tkatchenko Scheffler method and assessment of dispersion methods for alkali-containing systems".

Mira: Here's the gist of the paper. The present work examines the accuracy of the Tkatchenko-Scheffler method with updated van der Waals radii (abbreviated as TS 2018), coupled with the semilocal Perdew-Burke-Ernzerhof density functional, for structural predictions of semiconducting and insulating materials in comparison to the non-local many-body dispersion method and the original TS method (TS 2009) The Tkatchenko-Scheffler (TS) pairwise method is a widely used approach to incorporate missing long-range van der Waals contributions in semilocal and hybrid density functional calculations Despite numerous refinements of the approach to include many-body terms, the original formulation still remains highly relevant as an efficient and robust method, especially for organic and/or insulating materials The present work examines the accuracy of the TS method with updated van der Waals radii (abbreviated as TS 2018), coupled with the semilocal Perdew-Burke-Ernzerhof density functional, for structural predictions of semiconducting and insulating materials in comparison to the non-local many-body dispersion method and the original TS method (TS 2009) Special attention is paid to materials containing alkali elements, for which the TS 2009 method exhibits a large overbinding, associated with potentially large errors in predicted atomic structures TS Method Refinements The TS 2018 method improves upon the original TS 2009 method by deriving the free-atom vdW radii from atomic dipole polarizabilities and by revising the approach to determine the interatomic vdW distances This revision successfully addresses key limitations of the TS 2009 method that arise from an underestimation of the atomic radii for alkali elements The TS 2018 scheme improves upon the original TS 2009 method by deriving the free-atom vdW radii from atomic dipole polarizabilities and by revising the approach to determine the interatomic vdW distances Performance on Alkali Dimers The TS 2018 and TS alkali methods correct an unphysically large binding energy in five alkali dimer test cases (Li2, Na2, K2, Rb2 and Cs2) The overbinding becomes more pronounced from Li2 to Cs2, and can be attributed to the underestimation of the free-atom radii for alkali metals in the TS 2009 method The PBE+TS 2018 and PBE+TS alkali methods can reach a level of accuracy comparable to the mathematically more complex PBE+MBD NL method for both alkali halides and a variety of other binary semiconductor systems Structural Predictions in Solids In the non-alkali containing materials of Figure 4, all vdW methods probed here yield improved results over the plain PBE functional For twenty-five binary semiconductors that do not contain alkali elements, the unit cell volumes calculated using PBE+MBD NL, PBE+TS 2009 (which is equivalent to TS alkali when alkali atoms are not present), and PBE+TS 2018 methods are within 7 % of the reference values The uncorrected PBE functional shows a consistently larger overestimation of the lattice volume Perovskite System Analysis For three selected Cs-based three-dimensional lead-halide perovskite test cases (CsPbCl3, CsPbBr3 and CsPbI3), the PBE+TS 2018 and PBE+TS alkali methods give better lattice anisotropies than the original PBE+TS 2009 approach However, the PBE+MBD NL method outperforms both on the unit cell volume error, when considering low-temperature experimental reference data These results establish the PBE+TS 2018 and PBE+TS alkali, as well as PBE+MBD NL, as effective methods for modeling complex organic-inorganic hybrid systems where vdW interactions play a key role, but alkali elements are also present The TS 2018 and TS alkali methods can reach a level of accuracy comparable to the mathematically more complex PBE+MBD NL method for both alkali halides and a variety of other binary semiconductor systems Finally, for three selected Cs-based three-dimensional lead-halide perovskite test cases (CsPbCl3, CsPbBr3 and CsPbI3), the PBE+TS 2018 and PBE+TS alkali methods give better lattice anisotropies than the original PBE+TS 2009 approach The results establish the PBE+TS 2018 and TS alkali, as well as PBE+MBD NL, as effective methods for modeling complex organic-inorganic hybrid systems where vdW interactions play a key role, but alkali elements are also present Acknowledgement This work was funded by the U.S.

Kai: National Science Foundation under awards No.

Mira: DMR2323803 and DMR-2323804 Research Computing at the University of North Carolina at Chapel Hill is acknowledged for providing computational resources Data Availability All data calculation including input and output files, processed data, and benchmark data gathered from references is publicly available on Github, Ref.

Kai: 94 Implementations of the TS 2018 and TS alkali methods in the open-source libMBD package are forthcoming Appendix A Details The revised formula of Tkatchenko-Scheffler Dispersion Method uses an effective atomic volume calculated from the electron density to scale free-atom polarizabilities and C6 coefficients to account for the environmental dependence The new TS 2018 formula outperforms the TS 2009 formula with an average relative error, ⟨R.E.⟩, of 0.2% and an average absolute relative error, ⟨R.E.⟩, of 1% compared to experimental values of the noble gas dimers The new TS 2018 formula outperforms the TS 2009 formula with an average relative error, ⟨R.E.⟩, of 0.2% and an average absolute relative error, ⟨R.E.⟩, of 1% compared to experimental values of the noble gas dimers The new TS 2018 formula outperforms the TS 2009 formula with an average relative error, ⟨R.E.⟩, of 0.2% and an average absolute relative error, ⟨R.E.⟩, of 1% compared to experimental values of the noble gas dimers Appendix B Reference Values The experimental reference data for alkali halides are listed in Table 2 The experimental values used for the alkali halides are listed in Table 3 The zero-point corrections are calculated and listed in Table 4 Table 6 Performance Summary The performance of PBE, PBE+MBD NL, PBE+TS 2009, and PBE+TS 2018 methods in predicting the unit cell volume of alkali halides and semiconductor systems is summarized in Table 1 The average relative error (⟨R.E.⟩) and absolute relative error (⟨R.E.⟩) values show the overall deviation magnitude between predicted unit cell volumes from the experimental references The updated PBE+TS 2018 method corrects the atomic vdW radii and thus largely eliminates errors stemming from the alkali elements, allowing the method to reach a similar accuracy as the PBE+MBD NL method The updated PBE+TS 2018 method corrects the atomic vdW radii and thus largely eliminates errors stemming from the alkali elements, allowing the method to reach a similar accuracy as the PBE+MBD NL method The updated PBE+TS 2018 method corrects the atomic vdW radii and thus largely eliminates errors stemming from the alkali elements, allowing the method to reach a similar accuracy as the PBE+MBD NL method The updated PBE+TS 2018 method corrects the atomic vdW radii and thus largely eliminates errors stemming from the alkali elements, allowing the method to reach a similar accuracy as the PBE+MBD NL method Table 5 Perovskite Structure Data The experimental and predicted lattice parameters (lattice vectors a,b, and c; lattice angles α, β, and γ) of CsPbCl3, CsPbBr3, and CsPbI3 are presented in Table 5 The PBE+TS 2018 method gives better lattice anisotropies than the original PBE+TS 2009 approach The PBE+MBD NL method outperforms both on the unit cell volume error, when considering low-temperature experimental reference data The PBE+TS 2018 and TS alkali methods give better lattice anisotropies than the original PBE+TS 2009 approach The PBE+MBD NL method outperforms both on the unit cell volume error, when considering low-temperature experimental reference data The PBE+TS 2018 and TS alkali methods give better lattice anisotropies than the original PBE+TS 2009 approach References (3) Jarvis, S.

Title and authors: Kai: Moving on, the title of this paper is "Updated Tkatchenko Scheffler method and assessment of dispersion methods for alkali-containing systems". It clearly frames the scope of the work.

Mira: The authors are Schankler, Song, Bhattacharya, Lucas, Hampton III, Tkatchenko, Kanai and Blum. They're a good mix of condensed matter theory and experimentalists involved in computational chemistry.

Lev: I see a lot of heavy lifting here from the team that is building error correction for real hardware on the quantum side.

Kai: They’re using methods to calculate dispersion forces, which are these long-range attractions between atoms, especially important when we talk about molecular crystals or materials with flexible structures.

Mira: And this paper focuses specifically on how these forces change when alkali elements are introduced into the system because that's where the old methods struggle.

Lev: So they’re taking a standard pairwise method and updating its parameters to account for those specific atomic properties, moving from two thousand nine to two thousand eighteen <ref:2608.01501#pg1>.

Kai: The paper is essentially saying that if you use TS two thousand eighteen instead of the old TS two thousand nine you get better predictions for materials containing alkali ions <ref:2608.01501#pg1>.

Mira: And it’s not just about accuracy; it’s about making sure the underlying physics assumptions—like how we derive those atomic radii—are more sound.

Lev: The complexity is that they are comparing this to other approaches, like the non-local many-body dispersion method, which is mathematically much heavier.

Kai: So they're showing that even with a simpler method like TS two thousand eighteen you can get close to the performance of a more complicated one for these specific problems <ref:2608.01501#pg1>.

The paper's summary: Kai: To summarize what the authors are doing, they are evaluating the accuracy of using the Tkatchenko Scheffler method with updated van der Waals radii, which they call TS two thousand eighteen when paired with the semilocal Perdew Burke Ernzerhof density functional <ref:2608.01501#pg1,method with updated van der Waals radii>.

Mira: The paper says this combination is used to make structural predictions for semiconducting and insulating materials.

Lev: They are comparing it against two other benchmarks: the non-local many-body dispersion method and the original TS two thousand nine method <ref:2608.01501#pg1>.

Kai: And they highlight that the main focus is on materials containing alkali elements because those have a known problem with overbinding in the old TS two thousand nine version <ref:2608.01501#pg1>.

Mira: They are demonstrating that by using TS two thousand eighteen which uses updated radii, they significantly reduce those errors in predicted atomic structures <ref:2608.01501#pg1>.

Lev: In essence, they are showing that this refined pairwise approach is more robust for these materials than the original formulation when alkali ions are present.

Kai: It’s about taking a method that was efficient and reliable for organic materials and making it much better suited for inorganic systems with these challenging ions.

The paper's improvements: Mira: The paper points out the specific changes they made to the TS method itself, which is that they derived the free-atom van der Waals radii from atomic dipole polarizabilities and revised how interatomic van der Waals distances are determined.

Lev: That means they are using a different way to calculate those fundamental parameters instead of just plugging in some fixed numbers.

Kai: So by deriving them from the polarizabilities, they are accounting for how the environment around an atom affects its size.

Mira: Exactly, and that's what allows them to address the key limitations of TS two thousand nine which came from underestimating those atomic radii for alkali elements <ref:2608.01501#pg1>.

Lev: And this revision is what actually makes TS two thousand eighteen better than TS two thousand nine in terms of handling the input data <ref:2608.01501#pg1>.

Kai: So, the improvement isn't just a patch; it’s a fundamental rethink of how you calculate those necessary inputs for the method.

Mira: It’s about making sure that what you use to start your calculation is physically more accurate for these systems than what was available before.

Conclusion: Kai: So, wrapping up, the main point here is that the TS two thousand eighteen and TS alkali methods can reach an accuracy level comparable to the mathematically more complex PBE plus many body non local method for alkali halides and other binary semiconductor systems <ref:2608.01501#pg1>.

Mira: And they also show better lattice anisotropies in perovskite systems when comparing them to the original PBE plus TS two thousand nine approach <ref:2608.01501#pg1>.

Lev: From my perspective, this means that for these specific complex materials, PBE plus TS two thousand eighteen and TS alkali methods are effective tools <ref:2608.01501#pg1>.

Kai: They seem like a solid way forward for modeling systems where vdW interactions are important but you're dealing with alkali ions.

Mira: Ultimately, this work establishes these methods as effective ways to model those complex organic inorganic hybrid systems where van der Waals interactions play a key role, but alkali elements are also present.

Lev: So it’s a validation that we have something that works for these materials, even if the computational cost is still a bit high for real-world experimental design.

Kai: We're excited to see how this theoretical accuracy translates into what we can actually measure with our hardware in the near future.

Mira: Looking forward to seeing where this line of research goes next and what other materials they tackle with these refined methods.

More episodes

← Home