Kinetically Trapped Nanocrystals with Symmetry-Preserving Shapes

The shape of nanocrystals is crucial in determining their surface area, reactivity, optical properties, mechanical strength, and self-assembly behavior. By modulating kinetics at terraces, ledges, and kinks using kinetic Monte Carlo simulations informed by energy models, this study reveals that the primary factors controlling nanocrystal morphology are the adatom nucleation energies and the geometry of growth islands.

The gist

Transient sites dominate the growth process, leading to kinetically trapped, metastable shapes.

Nanocrystal Growth Simulations

The researchers simulate NC shape formation using a rejection-free kinetic Monte Carlo (rfKMC) method, which bridges the gap between MD simulations and geometric construction models. This approach dynamically positions growth sites around the NC surface and continuously updates them as the NC evolves. Two Monte Carlo moves are considered: Etching moves remove a surface atom at rate re,i, while Growth moves add an atom at a growth site i with rate rg,i. These rates are given by equations dependent on the energy Ei of adding an atom at site i:

Growth and etching rates are given by:

(1) ri ∝ exp ∓ Ei / kT,

where kT is the Boltzmann constant multiplied by the temperature, and Ei represents the (free) energy change of adding an atom at the growth site i. The energy Ei depends linearly on the coordination number zi as depicted in Fig. 1b, with discrete levels given by Ei = −ziϵ, where 1 ≤ zi ≤ 12 and ϵ is the bond energy, with dimensionless counterpart ϵ∗ = εkT.

Shape Diagrams and Kinetic Trapping

The study aims to identify the minimum requirements to form diverse polyhedral NC shapes by linking TLK crystallization theory with kinetic Monte Carlo simulations. The primary facets have coordination numbers zad: zad(111) = 3, zad(100) = 4, and zad(110) = 5, which predicts growth velocities v(111) < v(100) < v(110). The key finding is that the NC shape is not solely determined by facet surface energies but by the growth pathway that kinetically traps NC shapes into metastable equilibrium.

Facet selection:

(a) primary facets have coordination numbers zad: zad(111) = 3, zad(100) = 4, and zad(110) = 5 for the primary facets,

(b) surface energies of these facets are E(111) < E(100) < E(110), resulting in the equilibrium Wulff shape [35] of fcc being an octahedron.

The researchers hypothesize that a small set of key determinants derived from the energy model, representing the combined effects of precursors, ions, ligands, and solvents, guides the growth.

Influence of Kinetic Parameters and Energy Ratios

To bias the formation of other shapes beyond the equilibrium shape (octahedron), one must adjust growth rates associated with coordination numbers 3 ≤ z ≤ 5 that represent adatom nucleation. The study summarizes the effect of the ratios E(z=3)/E(z=5) and E(z=4)/E(z=5) on NC shape formation in two shape diagrams:

For irreversible growth (p = 1, Fig. 2a):

(c) The coexistence of kinetic and equilibration effects when growth is reversible enables the formation of the facets for the first time.

The analytic model derived from Eqs. (2) and (3), which relate growth velocity ratios to energy differences, reveals that:

(13) v(110)/v(100) = a exp [E(z=4) − E(z=5)] / kT,

(14) v(111)/v(100) = b exp [E(z=4) − E(z=3)] / kT,

where b ∈ 【bI, bII】 indicates limit I and II.

Facet Roughening and Isotropic Growth

The impact of varying bond energy ϵ∗ on NC morphology is explored. At high values of ϵ∗, NCs exhibit faceted shapes because atoms are preferentially added to growth sites with higher coordination numbers (lower energies), following TLK crystallization theory, resulting in a layer-by-layer growth mode. As ϵ∗ decreases, the growth rates of sites with different coordination numbers become more similar. This leads to the formation of multiple growth islands on the NC surface, causing surface roughening. When diffusion length is such that atoms cannot hop and instead attach to the first site they encounter, the NC grows isotropically, forming a spherical shape.

As a fastidious researcher, I have analyzed this paper on kinetically trapped nanocrystal (NC) shapes using kinetic Monte Carlo (KMC) simulations and analytic modeling based on terrace-ledge-kink (TLK) theory. The core contribution is bridging atomistic energy models with macroscopic shape prediction by focusing on the kinetics of adatom nucleation at low coordination number sites.

Here are specific, actionable improvements for AI systems based on the insights derived from this paper:

Improvements for AI Systems

The primary improvement lies in developing a multi-scale predictive framework that incorporates kinetic barriers and surface topology directly into generative design algorithms for nanomaterials. This moves beyond static energy minimization (equilibrium Wulff shapes) to dynamic, process-aware shape prediction.

1. Development of Kinetic Shape Predictors (KSP)

AI systems should be trained not just on the final geometric outcome but on the kinetic pathways that lead to it.

Improvement: Implement a Kinetic Energy Landscape Mapping module within generative AI models (e.g., Graph Neural Networks or specialized Reinforcement Learning agents). This module must ingest local atomic environments (coordination numbers, bond energies) and learn the probability distribution of growth/etching moves described by the rfKMC algorithm (Equation 1).

Specific Capability: The system can predict not just what shape is stable, but what is the most likely shape formed under specific synthesis conditions (defined by temperature, ligand concentration, precursor type), directly addressing the kinetic trapping phenomenon.

2. Integration of TLK Theory into Generative Design

The analytic models derived in the paper (Equations 13 and 14) provide explicit functional relationships between energy ratios and growth velocity ratios across different crystal facets.

3. Enhanced Feature Recognition for Surface Roughness Control

The paper shows how decreasing bond energy (increasing surface roughness) leads to isotropic growth and spherical shapes, whereas high bond energy favors faceted shapes.

4. Multi-Scale Data Fusion for Robust Modeling

The paper highlights the complexity arising from incorporating ligand adsorption and solvent effects (Equations 31, 25).

Summary of Improved AI System Capabilities

The improved system moves from being a simple pattern recognizer or thermodynamic optimizer to a sophisticated, process-aware materials designer capable of:

1. Predictive Shape Synthesis: Accurately predicting the final morphology of nanocrystals based on synthesis parameters (temperature, ligands) by modeling the kinetic trapping mechanisms (TLK theory).

2. Kinetic Control Design: Designing precursor chemistries and solution environments specifically to bias growth rates toward a desired non-equilibrium shape, moving beyond equilibrium constraints.

3. Inverse Shape Engineering: Taking a target morphology and calculating the precise energy/growth rate ratios required to kinetically achieve that structure in the simulated environment.

4. Multiscale Parameter Optimization: Optimizing synthesis parameters across multiple scales (electronic structure, surface adsorption, lattice growth) simultaneously to ensure robust and predictable material outcomes.