How Irregularities, Vacancies and Defects Tip the Crossing Sizes

September 2, 2026

The silver nanoparticle morphologies are identified thanks to ab initio models up to 4 nm, including original defects and complex motifs, and compared to measurements at targeted sizes. The crossings between structures are predicted at very accurate computational conditions, which are compulsory to solve the fierce morphological battle between key symmetries. The provided database will allow the training of machine-learning approaches to examine larger sizes.

At the nanoscale, each atom counts. Indeed, reaching a high degree of control of nanoparticle (NP) morphology is mandatory to design well-defined structures and to engineer their properties. Over the last decade, the synthesis protocols of metallic NPs and advanced characterization techniques have reached a good level of maturity. However, the determination of the competitive morphologies at the one-atom scale remains a timely challenge. To date, the scientific community is constrained to work around a specific size, on a too large range of number of atoms (typically ±10-20 atoms), in addition to other difficulties such as kinetic trapping. This clearly prevents to solve the question of morphological competition, even more when it is close.

A very revealing example is that of silver NPs, for which many efforts are made to identify the competitive structures in controlled conditions, without being able to determine the NP geometries at each size. In this context, Palmer’s group evidenced a very narrow competition of key symmetries at various targeted sizes below 3 nm [1]. This impressive work remained uninterpreted due to several lacks in the literature. To solve this question, we have conducted an ab initio study of an exhaustive morphological search, including defects and complex motifs. An interpretation of the experimental measurements is proposed for the first time.

We show that the defective families have to be considered to propose an interpretation of morphological statistics. In the range 2-4 nm, a close competition is observed between (irregular and defective) face-centered cubic and (regular and defective) Marks-decahedral NPs, which dominate icosahedra in number, not in stability (below 4.5 nm). The crossing diagrams are highly complex, with successive stability inversions occurring at sizes that may differ by as little as a single atom. As a consequence, to imagine predicting the crossing sizes between morphologies, without ensuring a high accuracy in energetics, is hopeless.

This work warns the scientific community about the necessity of accuracy on such systems, for which each atom matters and defects are undeniably competitive. It directly impacts the development of machine learning potentials. In addition, our database will be useful for generating other ones for metals.

[1] D. M. Foster, Production and characterisation by scanning transmission electron microscopy of size-selected noble metal nanoclusters, University of Birmingham. Ph.D., 2017.

Contact:
Nathalie Tarrat | nathalie.tarrat[at]cemes.fr

Publication:
The Fierce Morphological Battle of Silver Nanoparticles: How Irregularities, Vacancies and Defects Tip the Crossing Sizes
N. Tarrat and D. Loffreda
Small (2026): e75018
DOI: https://doi.org/10.1002/smll.75018