Atomic layer etching of metals with anisotropy, specificity, and selectivity

Abstract

In this work, a special focus is given to atomic layer etching (ALE) of metals, since this is a relatively new field but is expected to grow rapidly given the major advancements potentially enabled via metal incorporation throughout the manufacturing process of integrated circuits. The impact of John Coburn’s work on the development of ALE processes is analyzed with a focus on ion energy and the neutral-to-ion ratio. To realize atomic precision in removing etch-resistant materials with complex compositions or structures, the surface reactivity would replace etch rate as the parameter of interest to control the chemical contrast needed for selectivity. The desirable etching anisotropy dictates the usage of directional ions. John Coburn’s work on ion-enhanced etching of Si serves as an example that a fine control of ion energy and the neutral-to-ion ratio could be the gateway of reactivity control, which is demonstrated with recent progress on thermal-plasma ALE of Ni. The effect of surface reactivity is studied from first-principle atomistic calculations and confirms the experimental findings.

Thermodynamics of Atomic Layer Etching Chemistry on Copper and Nickel Surfaces from First Principles

“…The calculated Δ G rm,act is 0.10 eV for the HFAmd etching of Ni 3 N, which is consistent with the reports of Ni 3 N deposition using amidinate complexes of nickel (which can be subsequently reduced by H 2 to give Ni films), although the ligands in the report are heavily substituted . We highlight the fact that etching O-activated Ni with formic acid is predicted to be unfavorable by 0.26 eV, inconsistent with the experimentally observed etching …”

“…Throughout the text, the units for reaction 40 We highlight the fact that etching O-activated Ni with formic acid is predicted to be unfavorable by 0.26 eV, inconsistent with the experimentally observed etching. 41 The bulk model results suggest that formamidine etching is preferable compared to formic acid by 0.24 and 0.07 eV, respectively, on Ni and Cu. The differences between oxygen and nitrogen activation (0.08 eV favoring oxygen on Ni, 0.09 eV favoring nitrogen on Cu) are not significant.…”

“…The influence of surface morphology is clearly indicated when the layer model (assuming a mononuclear nickel formate complex) predicts etching with formic acid to be unfavorable on Ni(111) but favorable when the subsurface sites become occupied. 41 The heterogeneity among different terminations must be treated carefully for thermodynamic predictions to guide the experiments. The approach taken in the present work relies on the surface energies to summarize the results from various terminations (higher surface energy structures are suppressed).…”

“…Following the general trend of more favorable etching at higher coverages, it is expected that the inclusion of subsurface sites into the configuration search scheme will continue this trend and yield more favorable sites. As our published results show, this is indeed the case. The results from our previous work on the nickel oxide systems with nickel formate monomer as the etching product are marked in Figure S12 for comparison.…”

“…Such effects cannot be determined a priori without a detailed study. The influence of surface morphology is clearly indicated when the layer model (assuming a mononuclear nickel formate complex) predicts etching with formic acid to be unfavorable on Ni (111), but favorable when the subsurface sites becomes occupied 41 . The heterogeneity among different terminations must be treated carefully for thermodynamic predictions to guide the experiments.…”