# Gold Nanocluster–Amino Acid Interactions: Assessment of DFTB with Dispersion Corrections

## TL;DR

This study evaluates how well a computational method called DFTB with dispersion corrections models interactions between gold nanoclusters and amino acids, showing it's generally reliable but with some limitations.

## Contribution

The paper introduces a systematic assessment of DFTB + D3(BJ) for modeling gold nanocluster–amino acid interactions, highlighting its accuracy and limitations.

## Key findings

- DFTB + D3(BJ) reproduces qualitative binding preferences for amine over carboxyl adsorption with energies close to DFT results.
- Au–X bond lengths are systematically longer in DFTB calculations, with larger deviations for Au13 and nitrogen-containing cyclic amino acids.
- DFTB remains reliable for larger clusters like Au20 but shows growing quantitative deviations as cluster size increases.

## Abstract

Understanding the interaction between gold nanostructures
and biomolecules
is critical for advancing applications in nanomedicine, biosensing,
and bioelectronics. Here, we assess the performance of density functional
tight binding (DFTB) with Grimme’s D3­(BJ) dispersion correction
by comparing it to available density functional theory (DFT) results
in the literature for gold nanocluster–amino acid complexes.
Five clusters (Au3, Au8, Au13, Au20, Au32) interacting with ten amino acids were
investigated at both amine and carboxyl binding sites. System selection
was guided by the availability of the corresponding DFT results in
the literature. DFTB reproduces the qualitative binding preference
for amine over carboxyl adsorption, with interaction energies typically
within 2–3 kcal/mol of DFT for Au3 and Au8. Au–X bond lengths are systematically longer by ∼0.4–0.7
Å, and larger deviations appear for Au13 and nitrogen-containing
cyclic amino acids. For Au20, good agreement with DFT is
obtained for alanine at the amine binding site and tryptophan at both
the amine and carboxyl sites, suggesting that DFTB remains reliable
beyond the smallest clusters. For Au32, although a direct
comparison is not possible, the data indicate similar binding preferences
to those found for smaller clusters. In addition, nitrogen-containing
cyclic amino acids (such as histidine, proline, and tryptophan) tend
to show larger discrepancies, reflecting the increased importance
of polarization, charge redistribution, and multiple competing binding
motifs in these systems. As the cluster size increases from Au3 to Au32, DFTB generally preserves qualitative
trends but exhibits growing quantitative deviations, indicating that
transferability toward more metallic, nanoparticle-like regimes must
be assessed with caution. Our results demonstrate that DFTB + D3­(BJ)
provides an efficient and sufficiently accurate framework for screening
Au–biomolecule interactions.