Wright and Kroemer (Applied Physics Letters 36, 210 (1980)) showed that an atomic gallium molecular beam at about 800 °C reduces nonvolatile SiO2 on silicon to volatile SiO and Ga2O — via SiO2+4 Ga→Si+2 Ga2O\mathrm{SiO}_2+4\,\mathrm{Ga}\rightarrow\mathrm{Si}+2\,\mathrm{Ga}_2\mathrm{O} and SiO2+2 Ga→SiO+Ga2O\mathrm{SiO}_2+2\,\mathrm{Ga}\rightarrow\mathrm{SiO}+\mathrm{Ga}_2\mathrm{O} — leaving a clean, damage-free surface suitable for molecular-beam epitaxy. As reported in [1], it left open the method's applicability to high-quality heteroepitaxy such as GaP-on-Si, the necessity of deliberately preoxidizing the silicon, the stoichiometry of the very thin near-surface oxides, and the quantitative kinetics and full parameter window of the reduction. The growth-trial question was answered at proof-of-principle level by the same authors in 1984 (ref 2); suboxide (SiOx) chemistry is now a standard input; and the reaction scheme has been corroborated and extended on Si, SiC, GaAs and GaSb. What no later work supplies is the Si-specific quantitative core: rate laws and activation energies of the Ga-assisted reduction as functions of Ga flux, temperature, oxide thickness and stoichiometry; the complete process window including contaminant loads; and a controlled comparison of preoxidized versus native-oxide surfaces. The open problem is this quantitative, mechanism-based completion of the gallium-beam ('galliation') surface-preparation method, whose purpose is the epitaxial interface quality needed for III-V/Si growth.

Background

Molecular-beam epitaxy (MBE) grows crystalline semiconductor films in ultrahigh vacuum (UHV) from beams of atoms or molecules impinging on a heated substrate; it requires an atomically clean substrate surface, since residual oxide or carbon blocks epitaxial ordering. A silicon wafer exposed to air is always covered by a native oxide (amorphous SiO2, with sub-stoichiometric SiOx near the interface) that is nonvolatile: thermal removal in UHV requires roughly 900 °C for ordinary native oxides — high enough to disturb shallow dopant profiles — while the historical alternative, Ar-ion sputtering, damages the near-surface region. Epitaxy of Ga-containing III-V compounds (GaP, GaAs, GaSb) on Si, motivated since the late 1970s by combining III-V optoelectronics with silicon technology, adds a polar-on-nonpolar problem: zinc-blende III-V material grown on diamond-structure Si can nucleate with either of two atomic orderings, forming antiphase domains (APDs), regions of inverted cation/anion order separated by defective boundaries.

In this context Wright and Kroemer (1980) demonstrated an in-situ, low-damage oxide-removal route: exposing oxidized Si to an atomic Ga beam at 800 °C reduces the oxide by the apparent reactions SiO2+4 Ga→Si+2 Ga2O\mathrm{SiO}_2+4\,\mathrm{Ga}\rightarrow\mathrm{Si}+2\,\mathrm{Ga}_2\mathrm{O} and SiO2+2 Ga→SiO+Ga2O\mathrm{SiO}_2+2\,\mathrm{Ga}\rightarrow\mathrm{SiO}+\mathrm{Ga}_2\mathrm{O}, where SiO (silicon monoxide) and Ga2O (gallium suboxide) are volatile and desorb, elemental Si is restored, and excess Ga re-evaporates (its vapor pressure at 800 °C is about 10−510^{-5} Torr, corresponding to roughly one monolayer (a single atomic layer) per second). The surface, verified by Auger electron spectroscopy (AES, a surface-sensitive elemental-analysis technique), shows no detectable oxygen, carbon, or residue. The method — later called 'galliation', 'Ga polishing', or, in flash form on GaSb, 'FLAG' (flashes of gallium) — works on deliberately grown chemical oxides 20–30 Å thick as well as on thin native oxides, with a sharp temperature threshold: at 750 °C reduction is negligible, at 800 °C it completes within minutes at a Ga flux of about 0.1 monolayer per second. A purely thermal alternative — a thin, clean protective oxide grown after wet-chemical carbon removal and desorbed below 800 °C — was established for Si MBE by Ishizaka and Shiraki (1986) and became the mainstream Si cleaning standard; the Ga-beam route remained attractive whenever a Ga source is available and lower thermal budgets or thicker, contaminated oxides are involved.

As reported in [1], the 1980 letter closed by acknowledging what it did not settle: whether the method would enable high-quality heteroepitaxial growth such as GaP-on-Si heterojunctions (growth trials were then under way); whether deliberately preoxidizing the Si before insertion is necessary or advantageous — the oxide was used as a sacrificial protective layer that may keep carbon and heavy metals away from the Si until it is removed, but the authors saw no reason to consider preoxidation required; the precise stoichiometry and structure of the very thin oxides and oxide/Si interfaces, then subjects of contemporary study; and the detailed kinetics (e.g., activation energies) and full parameter window in flux, time, and temperature for different oxide thicknesses and contaminant loads.

Later literature answered part of this. The same authors, with Inada, reported in 1984 the Ga-primed MBE growth of GaP on Si, APD-free on (211)-oriented substrates, with heterojunction bipolar transistors (HBTs, transistors whose emitter is the III-V side of the heterojunction) reaching 90% emitter injection efficiency — proof of principle that the Ga-beam preparation supports heteroepitaxy, self-described as approaching device quality. The reaction chemistry was corroborated and extended: Ga-beam reduction of SiC surface oxides (Kaplan and Parrill, 1986); mass-spectrometric and systematic studies of Ga-flux-driven oxide conversion on GaAs (Sasaki and Yoshida, 1994; Asaoka, 2003; Wasilewski et al., 2004, Ga2O3-to-Ga2O conversion); SiO2-to-Ga2O conversion under Ga flux on Si (Kumar et al., 2011); X-ray photoelectron spectroscopy of Ga nanodroplets drilling through SiOx on Si (Fouquat et al., 2019); and optimized Ga-flash oxide removal on GaSb (Mathews et al., 2017). The technique is still used and cited — in self-catalyzed III-V nanowire growth on oxidized Si (Madsen et al., 2011, who measured an effective Ga-deoxidization rate and modeled the native oxide as sub-stoichiometric SiOx), in antimonide MBE on Si (Madiomanana et al., 2015; the GaSb-lasers-on-Si line of Castellano et al. 2017 and Tournié et al. 2022), and in a 2025 discussion of in-situ Si cleaning alternatives (Zhang et al., 2025). Modern device-grade III-V-on-Si, however, is typically achieved with other surface preparations — optimized thermal deoxidation and As-modified or specially prepared Si(100) surfaces (Kunert et al., 2008; Supplie et al., 2015) — so the Ga route is one option among several rather than the key to the field.

What was never published, for the Si system itself, is the quantitative layer: no rate law, no activation energies, no systematic mapping of the process window versus oxide thickness/stoichiometry and contaminant load, and no controlled study of the preoxidation question. Practitioners still cite the 1980 letter for empirical parameters (flux, time, the ~800 °C threshold). The unresolved question is therefore whether the Ga-beam reduction of silicon oxide on Si can be completed as a quantitative, mechanism-based surface-preparation method.

Problem Statement

Establish a quantitative, mechanism-based characterization of the Ga-beam reduction of silicon oxide on silicon, sufficient to make the method predictable and transferable as an MBE surface preparation. Concretely: (1) Kinetics — determine the rate law and activation parameters (e.g., apparent activation energies) of the Ga-assisted reduction of SiOx layers on Si as functions of atomic Ga flux, substrate temperature, and oxide thickness/stoichiometry, with purely thermal oxide loss separated, and with the volatile products (SiO, Ga2O) identified and their desorption confirmed against the reactions SiO2+4 Ga→Si+2 Ga2O\mathrm{SiO}_2+4\,\mathrm{Ga}\rightarrow\mathrm{Si}+2\,\mathrm{Ga}_2\mathrm{O} and SiO2+2 Ga→SiO+Ga2O\mathrm{SiO}_2+2\,\mathrm{Ga}\rightarrow\mathrm{SiO}+\mathrm{Ga}_2\mathrm{O} and the near-stoichiometric Ga consumption observed originally. (2) Process window — establish the complete window of Ga flux, exposure time, and temperature (bracketing the sharp 750–800 °C sensitivity reported in 1980) for oxide thicknesses ranging from thin native oxides to deliberately grown chemical oxides about 20–30 Å thick, and for carbon/heavy-metal contaminant loads typical of MBE practice, including the conditions under which excess Ga re-evaporates without residue, droplet formation, or Ga–Si surface damage. (3) Preoxidation — provide a controlled comparison of deliberately preoxidized versus native-oxide (or HF-passivated) Si surfaces carried through the same reduction process, settling whether the sacrificial oxide sequesters contaminants or otherwise improves the resulting surface and subsequent epitaxy, or is unnecessary. The letter's heteroepitaxy question — whether the method enables GaP-on-Si heterojunctions — was answered at proof-of-principle level by the same authors in 1984 (APD-free layers on (211)Si, heterojunction bipolar transistors with emitter injection efficiencies up to 90%, self-described as approaching device-quality interfaces), while the quantitative completion stated in items (1)–(3) remains open.

References

  1. Steve Wright and Herbert Kroemer, "Reduction of oxides on silicon by heating in a gallium molecular beam at 800 °Ca)", Applied Physics Letters 36, 210–211 (1980). DOI: 10.1063/1.91428, https://doi.org/10.1063/1.91428
  2. Steven L. Wright, Herbert Kroemer, and Masanori Inada, "Molecular beam epitaxial growth of GaP on Si", Journal of Applied Physics 55, 2916–2927 (1984). DOI: 10.1063/1.333333, https://doi.org/10.1063/1.333333
  3. Parvez N. Uppal and Herbert Kroemer, "Molecular beam epitaxial growth of GaAs on Si(211)", Journal of Applied Physics 58, 2195–2203 (1985). DOI: 10.1063/1.335987, https://doi.org/10.1063/1.335987
  4. Akitoshi Ishizaka and Yasuhiro Shiraki, "Low Temperature Surface Cleaning of Silicon and Its Application to Silicon MBE", Journal of the Electrochemical Society 133, 666–671 (1986). DOI: 10.1149/1.2108651, https://doi.org/10.1149/1.2108651
  5. R. Kaplan and T. M. Parrill, "Reduction of SiC surface oxides by a Ga molecular beam: LEED and electron spectroscopy studies", Surface Science 165, L45–L52 (1986). DOI: 10.1016/0039-6028(86)90799-5, https://doi.org/10.1016/0039-6028(86)90799-5
  6. Masahiro Sasaki and Seikoh Yoshida, "Mass spectrometric study of the reaction of photo-oxidized GaAs with Ga", Journal of Applied Physics 75, 4214–4219 (1994). DOI: 10.1063/1.356008, https://doi.org/10.1063/1.356008
  7. Yasushi Asaoka, "Desorption process of GaAs surface native oxide controlled by direct Ga-beam irradiation", Journal of Crystal Growth 251, 40–45 (2003). DOI: 10.1016/S0022-0248(02)02492-2, https://doi.org/10.1016/S0022-0248(02)02492-2
  8. Z. R. Wasilewski, J.-M. Baribeau, M. Beaulieu, X. Wu, and G. I. Sproule, "Studies of oxide desorption from GaAs substrates via Ga2O3 to Ga2O conversion by exposure to Ga flux", Journal of Vacuum Science & Technology B 22, 1534–1538 (2004). DOI: 10.1116/1.1752913, https://doi.org/10.1116/1.1752913
  9. B. Kunert, I. Németh, S. Reinhard, K. Volz, and W. Stolz, "Si (001) surface preparation for the antiphase domain free heteroepitaxial growth of GaP on Si substrate", Thin Solid Films 517, 140–143 (2008). DOI: 10.1016/j.tsf.2008.08.077, https://doi.org/10.1016/j.tsf.2008.08.077
  10. Mahesh Kumar, Mohana K. Rajpalke, Basanta Roul, Thirumaleshwara N. Bhat, S. Dash, A. K. Tyagi, A. T. Kalghatgi, and S. B. Krupanidhi, "Reduction of oxygen impurity at GaN/β-Si3N4/Si interface via SiO2 to Ga2O conversion by exposing of Si surface under Ga flux", Journal of Crystal Growth 327, 272–275 (2011). DOI: 10.1016/j.jcrysgro.2011.06.010, https://doi.org/10.1016/j.jcrysgro.2011.06.010
  11. Morten Hannibal Madsen, Martin Aagesen, Peter Krogstrup, C. B. Sørensen, and Jesper Nygård, "Influence of the oxide layer for growth of self-assisted InAs nanowires on Si(111)", Nanoscale Research Letters 6, 516 (2011). DOI: 10.1186/1556-276X-6-516, https://doi.org/10.1186/1556-276X-6-516
  12. K. Madiomanana, M. Bahri, J.-B. Rodriguez, L. Largeau, L. Cerutti, O. Mauguin, A. Castellano, G. Patriarche, and E. Tournié, "Silicon surface preparation for III-V molecular beam epitaxy", Journal of Crystal Growth 413, 17–24 (2015). DOI: 10.1016/j.jcrysgro.2014.12.004, https://doi.org/10.1016/j.jcrysgro.2014.12.004
  13. Oliver Supplie, Matthias M. May, Peter Kleinschmidt, Andreas Nägelein, Agnieszka Paszuk, Sebastian Brückner, and Thomas Hannappel, "In situ controlled heteroepitaxy of single-domain GaP on As-modified Si(100)", APL Materials 3, 126110 (2015). DOI: 10.1063/1.4939005, https://doi.org/10.1063/1.4939005
  14. Sen Mathews, Theodore Schuler-Sandy, Jong Su Kim, Clark Kadlec, Alireza Kazemi, Vinita Dahiya, David A. Ramirez, Stephen A. Myers, Yuliya V. Kuznetsova, and Sanjay Krishna, "In situ flashes of gallium technique for oxide-free epiready GaSb (100) surface", Journal of Vacuum Science & Technology B 35, 02B114 (2017). DOI: 10.1116/1.4978604, https://doi.org/10.1116/1.4978604
  15. A. Castellano, L. Cerutti, J. B. Rodriguez, G. Narcy, A. Garreau, F. Lelarge, and E. Tournié, "Room-temperature continuous-wave operation in the telecom wavelength range of GaSb-based lasers monolithically grown on Si", APL Photonics 2, 061301 (2017). DOI: 10.1063/1.4983389, https://doi.org/10.1063/1.4983389
  16. Louise Fouquat, Marco Vettori, C. Botella, Aziz Benamrouche, José Peñuelas, and G. Grenet, "X-ray photoelectron spectroscopy study of Ga nanodroplet on silica-terminated silicon surface for nanowire growth", Journal of Crystal Growth 514, 83–88 (2019). DOI: 10.1016/j.jcrysgro.2019.03.003, https://doi.org/10.1016/j.jcrysgro.2019.03.003
  17. E. Tournié, Laura Monge Bartolomé, Marta Rio Calvo, Zeineb Loghmari, D. A. Díaz-Thomas, R. Teissier, A. N. Baranov, L. Cerutti, and Jean-Baptiste Rodriguez, "Mid-infrared III–V semiconductor lasers epitaxially grown on Si substrates", Light: Science & Applications 11, 165 (2022). DOI: 10.1038/s41377-022-00850-4, https://doi.org/10.1038/s41377-022-00850-4
  18. Xuanchang Zhang, Haotian Zeng, Hui Jia, Xueying Yu, Suguo Huo, Makhayeni Mtunzi, Huiwen Deng, Jae-Seong Park, Hexing Wang, A. J. Seeds, Huiyun Liu, and Mingchu Tang, "Optimisation of argon plasma pre-treatment for enhanced silicon surface preparation for germanium epitaxy", Journal of Physics D: Applied Physics 58, 425103 (2025). DOI: 10.1088/1361-6463/ae0fa6, https://doi.org/10.1088/1361-6463/ae0fa6