Monolayer chemical beam etching: Reverse molecular beam epitaxy

Abstract

We have developed an etching process with real-time counting of each monolayer removed, thus achieving etching with monolayer precision and control. This is an exact reversal of molecular beam epitaxy or more specifically in this case, chemical beam epitaxy (CBE). This new etching capability which we refer to as monolayer chemical beam etching (ML-CBET) is achieved by employing in situ reflection high-energy electron diffraction (RHEED) intensity oscillation monitoring during etching. Etching is accomplished in high vacuum by injecting AsCl3 directly into a CBE growth chamber impinging on a heated GaAs substrate surface. Having both epitaxial growth and etching integrated in the same process and both capable of ultimate control down to the atomic layer precision represents a very powerful combination. This permits instant switching from growth to etching and vice versa, clean regrown interfaces critical for device applications, direct modification of surface chemistries during etching or growth, and high temperature etching (500–570 °C for InP and 500–650 °C for GaAs) unachievable in conventional etching processes. The temperature and flux dependence of etching rates are also studied using RHEED oscillations. Results indicate that ML-CBET is studied predominantly via a layer-by-layer mechanism under the present etching conditions.

Challenges of Tailoring Surface Chemistry and Plasma/Surface Interactions to Advance Atomic Layer Etching

“…1c. While the approach varies in detail, in general, species 'R' (this could be ions, 3,28 energetic neutrals 44,46 or others 4,29 ) interact with the adsorbed monolayer and the top layer of the substrate to form volatile species. Neutral species not participating in the reaction (purple) may be present in case of plasma exposure.…”

“…This could then mean that under certain conditions, SiCl and SiCl 2 liberated from the surface could undergo gas-phase collisions and/or recombination events that produce additional atomic chlorine, potentially leading to unintended reactions on the substrate surface. Alternatively, SiCl and SiCl 2 could N5057 scavenge other Cl species at adjacent surface sites to form more stable SiCl 4 , potentially leading to areas of unetched silicon. This is further complicated in the case of more complex etch systems such as SiO 2 and Si 3 N 4 , where more than one atomic species need to be volatilized from the surface to remove a monolayer of material.…”

“…For Si-based etch systems, Matsuura demonstrated feasibility for ALE using a 4-step process including alternating cycles of chlorine (gas only), followed by exposure to an Ar + ion beam with an ion energy chosen to ensure self-limiting mono-layer removal. 5 Similarly, Sakaue et al also demonstrated ALE capability for silicon, however their approach utilized alternating cycles of a fluorine containing downstream plasma (CF 4 /O 2 , NF 3 /N 2 or F 2 /He) at cryogenically cooled substrate temperatures followed by Ar + ion irradiation. 6 While each of these approaches demonstrates merit and significance for atomic layer etching, the adoption of ALE had not garnered z E-mail: suengelm@us.ibm.com significant interest until now.…”

“…3 In a different approach (derived from molecular beam epitaxy), Tsang et al also showed sub-monolayer precision GaAs removal by injecting AsCl 3 directly onto a heated sample surface in a chemical beam epitaxy chamber. 4 This approach demonstrated not only the ability to control a flux driven process down to the Angstrom level, but also the ability to modify interface properties and/or change surface morphology. For Si-based etch systems, Matsuura demonstrated feasibility for ALE using a 4-step process including alternating cycles of chlorine (gas only), followed by exposure to an Ar + ion beam with an ion energy chosen to ensure self-limiting mono-layer removal.…”

Self-assembling nanostructures and atomic layer precise etching in molecular beam epitaxy

“…Under pulsed operadimethylaminoarsenic (TDMAAs) [22][23][24][25][26][27]. PC1 3 tion we obtain the typical output power versus was used by Tsang et al [28] and Gentner et al [29] injection current density characteristic at 150 K for the etching of InP in CBE. The in-situ etching which is shown in Fig.…”