Molecular dynamics simulation of silicon nanoparticle crystallization during laser-induced forward transfer printing

Laser-induced forward transfer (LIFT) printing is a versatile technique to realize micro/nano-scale additive manufacturing of functional materials, including metals and semiconductors. However, the crystallization phenomena during LIFT printing have not been well understood, which is critical to determine the resulting microstructure and properties. In this work, we systematically investigate silicon crystallization during LIFT printing using molecular dynamics (MD) simulations. Specifically, MD simulation with Stillinger-Weber (SW) potential is used to investigate the size effect and surface influence on the crystallization of Si nanoparticles during transportation in air. We found that with a decrease in nanoparticle size, crystallization becomes increasingly rare, even at low cooling rates. The nucleation location of different particles is also analyzed and almost always starts at a sub-surface location (below 5 Å). The evolution of the atomic structure during solidification is also monitored to guide LIFT printing of Si. Our simulation results indicate that nano-confinement induced by the surface layer can lead to single-crystal structure formation, which may shed light on additive manufacturing of single-crystal structures and devices.

💡 Research Summary

This paper investigates the atomistic mechanisms of silicon (Si) nanoparticle crystallization during laser‑induced forward transfer (LIFT) printing using molecular dynamics (MD) simulations. The authors employ the Stillinger‑Weber (SW) potential within LAMMPS to model spherical Si particles ranging from 2 nm to 20 nm in diameter, all placed in a vacuum environment to isolate the effects of size and cooling. Each particle is first heated from 300 K to 2000 K over 10 ns, held at 2000 K for 5 ns to ensure a fully amorphous melt, and then cooled from 1700 K under a micro‑canonical (NVE) ensemble. Cooling rates are controlled indirectly by imposing an interfacial thermal conductance (e.g., 6.38 × 10⁵ W m⁻² K⁻¹) that mimics heat transfer to the surrounding gas.

Key findings are threefold. First, both melting points and crystallization temperatures increase with particle size, yet remain below bulk values. The smallest particle (2 nm) shows no discernible latent‑heat release or crystallization, indicating that extreme curvature and associated Laplace pressure suppress nucleation. Second, cooling rate critically determines the solidification pathway. Fast cooling (≈10⁸ K ns⁻¹) prevents latent‑heat spikes for 2 nm and 4 nm particles, yielding amorphous structures, while 8 nm and 12 nm particles exhibit clear temperature jumps corresponding to latent‑heat release and crystallization. When the thermal conductance is reduced to 1.99 × 10⁵ W m⁻² K⁻¹ (average cooling ≈13 K ns⁻¹), all ten simulations of an 8 nm particle crystallize, demonstrating a 100 % success rate at sufficiently slow cooling. Further reduction to 7.81 × 10⁴ W m⁻² K⁻¹ yields similar results, confirming that a cooling rate below ~15 K ns⁻¹ is sufficient for reliable crystallization of particles ≥8 nm.

Third, the spatial origin of nucleation is consistently sub‑surface, within ~5 Å beneath the particle surface. The authors use bond‑order parameters (BOP, specifically the local q₃ invariant) to distinguish crystalline from amorphous atoms. In cases where crystallization occurs, the largest crystalline cluster initially forms a thin spherical shell just below the surface, then migrates inward as the temperature jump commences. Growth rates differ dramatically among outcomes: a single‑crystal‑like particle (case 10) grows at ~1575 atoms ns⁻¹, a polycrystalline particle (case 3) at ~142 atoms ns⁻¹, and an amorphous particle (case 1) at only ~15 atoms ns⁻¹. This demonstrates that sub‑surface nucleation followed by rapid inward propagation yields high‑quality single crystals, whereas slower or multiple nucleation events lead to polycrystalline or amorphous structures.

The study provides practical design guidelines for LIFT‑based additive manufacturing of Si. To obtain single‑crystal nanoparticles, one should aim for diameters ≥8 nm and ensure a low cooling rate, achievable by reducing the surrounding gas thermal conductance or by engineering the donor‑substrate gap to limit heat extraction. Conversely, very small particles (≤4 nm) are unlikely to crystallize regardless of cooling conditions due to curvature‑induced thermodynamic barriers.

Limitations include the use of a vacuum environment (ignoring gas‑particle heat transfer complexities), a fixed SW potential (which may not capture all high‑temperature behaviors), and the omission of particle‑substrate impact dynamics. Future work should incorporate multi‑particle interactions, realistic gas atmospheres, and explicit laser energy deposition to bridge the gap between simulation and experimental LIFT processes.

In summary, the paper delivers a comprehensive atomistic picture of size‑dependent Si nanoparticle crystallization during LIFT, elucidates the pivotal role of cooling rate and sub‑surface nucleation, and outlines how controlled thermal environments can enable direct printing of single‑crystal Si nanostructures for advanced electronic and photonic applications.