Material sputtering at low ion energies: From atomistics to engineering models with molecular dynamics, Monte Carlo, and data-driven approaches

Tran, Huy Dang

Permalink

https://hdl.handle.net/2142/127186 Copy

Description

Title

Material sputtering at low ion energies: From atomistics to engineering models with molecular dynamics, Monte Carlo, and data-driven approaches

Author(s)

Tran, Huy Dang

Issue Date

2024-11-13

Director of Research (if dissertation) or Advisor (if thesis)

Chew, Huck Beng

Doctoral Committee Chair(s)

Chew, Huck Beng

Committee Member(s)

• Rovey, Joshua L
• Levin, Deborah
• Yalin, Azer P
• Williams, John D

Department of Study

Aerospace Engineering

Discipline

Aerospace Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

space electric propulsion, sputtering by design, noble gas ion bombardment, graphite sputtering, general sputtering model, sputtering threshold energy, erosion morphology evolution

Abstract

Sputtering is a fundamental phenomenon describing the ejection of particles from a material surface under the bombardment of energetic ions. The removal of these atoms, under a controlled process, has revolutionized nano- and micro-scale manufacturing through surface patterning, and fabrication of thin film structures with enhanced physical, mechanical, and optical properties. However, sputtering is generally undesirable for plasma-facing materials, such as those in fusion and space electric propulsion devices, since it erodes the surfaces of critical material components and alters the plasma operational conditions. Regardless, sputtering predictions are highly unreliable at low ion incidence energies, typically less than 2 keV, due to the severely sparse and conflicting experimental data available, as well as the lack of understanding of the fundamental sputtering mechanisms at these ion energies. For instance, measurements conducted in different laboratories show that the sputter yield (number of ejected surface atoms per impact ion) of xenon (Xe) ions on carbon surfaces can vary by nearly an order of magnitude at low ion energies. Additionally, the sputter yield, differential yield (ejection angle), and energy distribution of the sputtered carbon atoms are very sparse for the noble gas bombardment of carbon at off-normal ion incidence angles. This dissertation seeks to harness the power of highly-parallelized supercomputing facilities, along with recent advancements in data science, to provide fundamental insights into the sputtering mechanisms of carbon under the lower-energy bombardment of noble gas ions, with the long-term goal of developing physics-based predictive models, which are applicable to general ion/target combinations across scales. My thesis research is divided into three parts. The first part is motivated by the reported large scatter in the sputtering data of carbon materials under the bombardment of low-energy noble gas ions. Here, I conducted scale-bridging molecular dynamics (MD) simulations on the xenon bombardment of carbon substrates across ion energies of 75 eV to 2 keV, and at ion incidence angles of 0° to 75°, to resolve uncertainties in the sputtering data. My results showed rapid amorphization of the carbon subsurface with ion bombardment, but the structural characteristics (sp/sp2/sp3 bond proportion, atomic density) eventually plateau once steady-state sputtering is achieved. In addition, my MD simulations showed that virtually indistinguishable steady-state amorphous sub-surface carbon structures are obtained across the range of ion energies and ion incidence angles, as well as for several different initial carbon structures (graphite, diamond), which suggests that the steady-state sputtering yield data obtained from MD is independent of the initial carbon structure and prior sputtering history. In the second part of my thesis research, I accounted for the evolving surface morphology and sputtering yield with ion fluence by upscaling my MD simulation results to a Monte Carlo (MC) model, which considers the shadowing of the incident ion flux, redeposition of sputtered carbon material, and secondary sputtering induced by the surface impact of the carbon sputterants. Results show that initially rough surface morphologies are consistently smoothened and flattened by a normal ion flux, resulting in sputtering yields that approach MD predictions. Under a highly oblique ion flux, however, the activation of multiple cooperative roughening and smoothening mechanisms at different scales lead to the formation of characteristic surface steps at the microscale, with steady-state sputtering yields that are up to an order-of-magnitude lower than MD predictions. While the observed surface features and the ensuing sputtering yield at steady-state are generally not sensitive to the initial surface morphology, the initial morphology controls the critical ion fluence required to attain steady-state sputter yield. Based on these observations, different initial surface topologies are proposed to delay and abate sputtering, which is a key step towards achieving sputtering-by-design. The parameters of a semi-empirical sputtering model are calibrated based on the MD-MC data at steady-state sputtering yield using a Bayesian approach. While this reduced-order semi-empirical model captures the sputtering predictions for the xenon ion bombardment of carbon substrates, extending this model to other ion/target combinations requires recalibration of the model fitting parameters. The key parameter that has the largest uncertainty is the sputtering threshold energy, which is the minimum ion incidence energy to initiate sputtering. In the third part of my thesis research, I focus on obtaining a data-driven expression for the threshold energy for sputtering applicable to general noble gas ion/target combinations. Using MD simulations, I quantify the sputtering threshold energies of monoatomic surfaces under the bombardment of noble gas ions across a broad range of ion-target combinations. The resulting threshold energies are related to the ion-target properties through an evolutionary algorithm for symbolic regression, and show a strong functional dependence on the nucleus charge governing ion-target repulsion and the target density, in addition to the heat of sublimation and ion-target mass ratio in prior semi-empirical models. This new data-driven formulation has order-of-magnitude improved threshold energy predictions and is applicable to crystalline and amorphous targets. I conclude my thesis with some proposed strategies to arrive at a new general closed-form expression for the sputtering yield applicable to a wide range of monatomic solids under noble gas ion bombardments. A road map for quantifying the sticking coefficients of carbon sputterants on the surfaces of different monoatomic substrates is also discussed.

Graduation Semester

2024-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/127186

Copyright and License Information

Copyright 2024 Huy Tran and Huck Beng Chew

Owning Collections