When polar zincblende GaAs nucleates on nonpolar Si(001), the allocation of the Ga and As sublattices, the atomic structure that compensates the heterovalent interface charge, and the electronic quality of the near-interface region are all fixed during the first moments of growth. Kroemer's 1987 nucleation model for antiphase-domain suppression left five linked gaps: the electronic activity of antiphase boundaries (APBs), the neutralization of the large donorlike charge implied by a bulklike As-first plane, direct verification of the proposed Ga-assisted As–Si site-exchange pathway, quantitative kinetic parameters for step-doubling and site exchange, and the connection between nucleation stoichiometry and the near-interface deep levels seen as capacitance–voltage anomalies. Later work settled the first two in substance — APBs are measurable recombination and scattering centers whose removal improves transport and photoluminescence, and the polar-on-nonpolar interface compensates charge by cation–Si/anion–Si intermixing rather than by a charged bulklike plane, antisites, or free carriers — and antiphase-boundary-free GaAs on nominal Si(001) is now routinely obtained via double-step surface preparation. What remains open is the atomic-scale formation pathway of the GaAs/Si(001) interface under Ga-rich versus As-rich nucleation — in particular whether the proposed two-step As–Si site exchange actually operates, and with what activation barriers — together with the causal explanation of the nucleation-dependent deep-level densities and capacitance–voltage anomalies at the GaAs/Si interface.

Background

GaAs is a polar semiconductor: it crystallizes in the zincblende structure, in which two interpenetrating face-centered-cubic sublattices are occupied by different chemical species (Ga and As), so that (100) atomic planes alternate between all-Ga and all-As composition. Silicon crystallizes in the diamond structure, where both sublattices carry Si atoms and differ only in bond orientation; the surface is therefore nonpolar. When GaAs is grown epitaxially on Si(001) — the combination pursued since the 1980s for integrating III-V optoelectronics with silicon electronics — the growing polar film must choose which of its sublattices to place on which Si site (the sublattice-allocation problem). On an ideal atomically flat Si(001) surface this choice is inconsequential, but every real surface has atomic steps: crossing a single-height step switches the local sublattice assignment, so GaAs nucleated on adjacent terraces can carry opposite Ga/As registry. The film then breaks into antiphase domains (APDs) separated by antiphase boundaries (APBs) — planar defects across which the Ga and As sublattices are interchanged, so that the crystal bonds through Ga–Ga and As–As 'wrong bonds'. Growth is typically by molecular beam epitaxy (MBE, deposition from elemental beams in ultrahigh vacuum) or metalorganic chemical vapor deposition (MOCVD).

In the mid-1980s, as-polished Si(100) surfaces showed mostly single-height steps and GaAs grown on exact (100) Si contained copious APBs. Empirically, a small deliberate misorientation of the wafer combined with a high-temperature surface anneal was found to suppress APBs: the anneal allows Si surface atoms to diffuse and pair single-height steps into double-height steps, and since a double step advances the Si sublattice by one full period, all terraces then present the same Si sublattice to the film. Aspnes and Ihm supplied the energetic driver: a reconstruction of the type-A double-step edge forming a π-bonded chain (a row of atoms along the step edge bonded in the graphite-like sp² geometry rather than the usual tetrahedral geometry), lowering that step's edge energy by roughly 40 meV per edge atom. On Si(211) orientations APBs are geometrically forbidden, but (100) is strongly preferred for device processing.

Against this background two experimental puzzles motivated the source paper. First, the group of Fischer reported (in work including GaAs bipolar transistors grown on (100) Si by MBE, and companion device studies) that antiphase-disorder-free GaAs could be obtained on a given Si(100) surface with either of the two possible sublattice allocations, depending on the initial nucleation conditions: an As prelayer at low temperature versus a Ga-related prelayer at higher temperature. Naive chemical intuition (As–Si bonds are stronger than Ga–Si bonds) predicts an As-first layer, so the occurrence of the opposite, effectively Ga-first allocation demanded a mechanism. Second, electrostatics forbids a simple bulklike As-first interface: as Harrison, Kraut, Waldrop, and Grant had shown for polar heterojunctions, an abrupt polar-on-nonpolar interface with one species bonded uniformly across it carries a huge interface charge density (a large donorlike sheet charge for a bulklike As-first plane on Si), and the interface responds by atomic rearrangement — their proposed 'breakup' of the polar plane by intermixing, so that each interface bond is locally charge-neutral in the electron-counting sense (a bookkeeping rule requiring that all valence electrons are consumed in bonds, leaving no charged dangling bonds).

Kroemer's 1987 model, the source of this problem, proposed for the Ga-rich branch a concrete two-step atomistic sequence: an arriving As atom exchanges sites with a Si atom of the top Si plane (plane #0), the displaced Si is relocated to a site in the next plane down (plane #1), and concurrently arriving Ga atoms bond so that, after two stages, Ga occupies plane #1 sites and As occupies planes #0 and #2 — a Ga-dominated registry opposite to the naive As-first picture, with lateral spreading of the nuclei preserving the initially established sublattice order. In this indexing plane #0, plane #1, and plane #2 are successive Si(001) atomic planes counted downward from the surface; adjacent Si(001) planes alternate between the two interpenetrating face-centered-cubic sublattices, so planes #0 and #2 lie on the sublattice of the top Si plane and plane #1 on the other. Sublattice allocation here means which chemical species occupies which of the two Si sublattices, and this mechanism assigns the Ga and As sublattices to them in the sense opposite to the naive As-first layer. For the As-rich branch the model instead assumes no exchange and an As-terminated first plane, leaving explicitly open how the implied electrostatic imbalance is neutralized — candidates being a Harrison–Kraut–Waldrop–Grant-type breakup of the top Si plane with Ga replacement, very high concentrations of Ga-on-As antisite defects (a Ga atom sitting on an As site, a doubly charged defect), or screening solely by mobile carriers. The paper acknowledged five linked open points: (i) the electronic properties and device impact of APBs themselves (recombination, trapping, transport) were unknown; (ii) the neutralization mechanism in the As-rich branch was unresolved; (iii) the site-exchange pathway lacked direct atomic-scale verification and activation barriers; (iv) quantitative energetic and kinetic parameters (activation energies, diffusion coefficients, rate models predicting the anneal temperature, time, and misorientation needed for step-doubling and site exchange with statistical reliability) were absent; and (v) the relation between nucleation pathway and the observed near-interface deep levels and capacitance–voltage (C–V) anomalies — diodes nucleated As-rich reportedly showing anomalous intercepts in 1/C21/C^2-versus-VV plots, consistent with a semi-insulating near-interface layer, while Ga-rich-nucleated diodes behaved as expected — was unexplained.

The field subsequently grew enormously: monolithic III-V-on-Si integration (lasers, solar cells, photodiodes) is a standing technological driver, and GaP on Si(001) (nearly lattice-matched) became the model polar-on-nonpolar system. The modern toolkit includes scanning transmission electron microscopy in high-angle annular dark-field mode (HAADF-STEM, whose atomic-column contrast scales with atomic number so Ga, As, and Si columns can in principle be distinguished), cross-sectional scanning tunneling microscopy (XSTM, imaging a cleaved interface with atomic resolution), synchrotron photoelectron spectroscopy of buried interfaces, deep-level transient spectroscopy (DLTS, which measures capacitance transients to extract trap activation energies and concentrations), and density-functional theory (DFT, first-principles total-energy computation). Using these, later work established that APBs are electrically active — non-radiative recombination centers and sources of interface conduction channels — and that the equilibrium III-V/Si(001) interface is charge-compensated by intermixing (coexisting cation–Si and anion–Si bonds in roughly half-and-half proportions) rather than being an abrupt charged plane; reproducible antiphase-boundary-free GaAs on nominal 300 mm Si(001) wafers is now achieved through double-step surface preparation plus two-step growth. What was never directly performed is the atomic-scale test of the specific nucleation pathway the 1987 model proposed for GaAs, nor the mechanistic explanation of the nucleation-dependent deep-level and C–V signatures.

Problem Statement

Determine, at the atomic scale and with predictive kinetic content, how the first moments of GaAs nucleation on Si(001) build the buried polar-on-nonpolar interface, and how that structure controls the electronic quality of the near-interface region — for GaAs nucleated under Ga-rich conditions (excess Ga / Ga prelayer at higher temperature) versus As-rich conditions (As prelayer at low temperature), the two regimes in which both sublattice allocations were observed. Specifically: (1) identify the interfacial atomic structure and its formation pathway for each nucleation regime — in particular, establish or refute the proposed two-step As–Si site exchange, in which an arriving As atom exchanges with a Si atom of the top Si plane, the displaced Si occupies a site in the next plane down, and Ga atoms bond into that second plane so that Ga occupies plane #1 sites and As occupies plane #0 and #2 sites — and quantify the corresponding activation energies and their dependence on temperature and Ga/As flux, comparing against the competing mechanisms proposed by later work (sublattice registry set by the As-terminated surface reconstruction and step structure; charge-compensated intermixed interfaces; polarity selection by coalescence of monophase islands); and (2) establish the causal connection between the nucleation route and interfacial electronic quality: whether Ga-rich nucleation intrinsically yields lower densities of near-interface deep levels, and whether the reported anomalous 1/C21/C^2-versus-VV intercepts of As-rich-nucleated GaAs/Si diodes are reproducible and, if so, which microscopic defect structure produces the semi-insulating near-interface layer that they manifest. Completion requires both numbered parts, each resolved to the following standard: part (1) must resolve the per-regime interfacial structure and its formation pathway, deliver an explicit verdict on the proposed two-step As–Si site exchange, and quantify kinetic parameters (activation energies and their dependence on temperature and Ga/As flux) for whichever pathway the evidence supports — the operative alternative's barriers if the exchange is refuted, rather than barriers for the refuted sequence — and must compare the result against the competing mechanisms; part (2) must give the causal account of the nucleation-route dependence of near-interface deep-level densities and resolve the reported 1/C21/C^2-versus-VV anomaly, i.e. its reproducibility and the microscopic defect structure responsible for the semi-insulating near-interface layer. Two of the source paper's five original open points have been settled in substance by later work (the general electrical activity of APBs as recombination and scattering centers, and the generic charge-neutralization mode of the III-V/Si(001) interface via compensated intermixing) and are treated as background.

References

  1. H. Kroemer, 'Sublattice allocation and antiphase domain suppression in polar-on-nonpolar nucleation', J. Vac. Sci. Technol. B 5(4), 1150–1154 (1987). DOI: 10.1116/1.583702, https://doi.org/10.1116/1.583702
  2. W. A. Harrison, E. A. Kraut, J. R. Waldrop, R. W. Grant, 'Polar heterojunction interfaces', Phys. Rev. B 18, 4402–4410 (1978). DOI: 10.1103/PhysRevB.18.4402, https://doi.org/10.1103/PhysRevB.18.4402
  3. D. E. Aspnes, J. Ihm, 'Biatomic steps on (001) silicon surfaces', Phys. Rev. Lett. 57, 3054–3057 (1986). DOI: 10.1103/PhysRevLett.57.3054, https://doi.org/10.1103/PhysRevLett.57.3054
  4. R. Fischer, N. Chand, W. Kopp, H. Morkoç, L. P. Erickson, R. Youngman, 'GaAs bipolar transistors grown on (100) Si substrates by molecular beam epitaxy', Appl. Phys. Lett. 47, 397–399 (1985). DOI: 10.1063/1.96179, https://doi.org/10.1063/1.96179
  5. R. Hull, A. Fischer-Colbrie, S. J. Rosner, S. M. Koch, J. S. Harris, 'Effect of substrate surface structure on nucleation of GaAs on Si(100)', Appl. Phys. Lett. 51, 1723–1725 (1987). DOI: 10.1063/1.98556, https://doi.org/10.1063/1.98556
  6. M. Kawabe, T. Ueda, 'Self-annihilation of antiphase boundary in GaAs on Si(100) grown by molecular beam epitaxy', Jpn. J. Appl. Phys. 26, L944 (1987). DOI: 10.1143/jjap.26.l944, https://doi.org/10.1143/jjap.26.l944
  7. J. E. Northrup, 'Theoretical studies of GaAs on Si', MRS Symp. Proc. 159, 3 (1989). DOI: 10.1557/PROC-159-3, https://doi.org/10.1557/PROC-159-3
  8. J. Varrio, H. Asonen, J. Lammasniemi, K. Rakennus, M. Pessa, 'Model of growth of single-domain GaAs layers on double-domain Si substrates by molecular beam epitaxy', Appl. Phys. Lett. 55, 1987–1989 (1989). DOI: 10.1063/1.102141, https://doi.org/10.1063/1.102141
  9. K. Adomi, S. Strite, H. Morkoç, 'Antiphase-domain-free GaAs grown on pseudomorphic Si (100) surfaces by molecular beam epitaxy', Appl. Phys. Lett. 56, 469–471 (1990). DOI: 10.1063/1.102768, https://doi.org/10.1063/1.102768
  10. D. K. Biegelsen, R. D. Bringans, J. E. Northrup, L. E. Swartz, 'Early stages of growth of GaAs on Si observed by scanning tunneling microscopy', Appl. Phys. Lett. 57, 2419–2421 (1990). DOI: 10.1063/1.103864, https://doi.org/10.1063/1.103864
  11. Y. González, L. González, F. Briones, 'In-situ monitoring of antiphase domain evolution during atomic layer MBE (ALMBE) and MBE growth of GaAs/Si(001) by reflectance difference', J. Cryst. Growth 111, 120–124 (1991). DOI: 10.1016/0022-0248(91)90958-8, https://doi.org/10.1016/0022-0248(91)90958-8
  12. V. Narayanan, S. Mahajan, K. J. Bachmann, V. Woods, N. Dietz, 'Antiphase boundaries in GaP layers grown on (001) Si by chemical beam epitaxy', Acta Mater. 50, 1275–1287 (2002). DOI: 10.1016/S1359-6454(01)00408-6, https://doi.org/10.1016/S1359-6454(01)00408-6
  13. D. B. Holt, B. G. Yacobi, 'Extended Defects in Semiconductors: Electronic Properties, Device Effects and Structures', Cambridge University Press (2007). DOI: 10.1017/CBO9780511534850, https://doi.org/10.1017/CBO9780511534850
  14. W. Guo, A. Bondi, C. Cornet, A. Létoublon, O. Durand, T. Rohel, S. Boyer-Richard, N. Bertru, S. Loualiche, J. Even, A. Le Corre, 'Thermodynamic evolution of antiphase boundaries in GaP/Si epilayers evidenced by advanced X-ray scattering', Appl. Surf. Sci. 258, 2808–2815 (2012). DOI: 10.1016/j.apsusc.2011.10.139, https://doi.org/10.1016/j.apsusc.2011.10.139
  15. S. Hosseini Vajargah, S. Y. Woo, S. Ghanad-Tavakoli, R. N. Kleiman, J. S. Preston, G. A. Botton, 'Atomic-resolution study of polarity reversal in GaSb grown on Si by scanning transmission electron microscopy', J. Appl. Phys. 112, 093101 (2012). DOI: 10.1063/1.4759160, https://doi.org/10.1063/1.4759160
  16. O. Supplie, M. M. May, P. Kleinschmidt, A. Nägelein, A. Paszuk, S. Brückner, T. Hannappel, 'In situ controlled heteroepitaxy of single-domain GaP on As-modified Si(100)', APL Mater. 3, 126110 (2015). DOI: 10.1063/1.4939005, https://doi.org/10.1063/1.4939005
  17. R. Alcotte, M. Martin, J. Moeyaert, R. Cipro, S. David, F. Bassani, F. Ducroquet, Y. Bogumilowicz, E. Sanchez, Z. Ye, X. Y. Bao, J. B. Pin, T. Baron, 'Epitaxial growth of antiphase boundary free GaAs layer on 300 mm Si(001) substrate by metalorganic chemical vapour deposition with high mobility', APL Mater. 4, 046101 (2016). DOI: 10.1063/1.4945586, https://doi.org/10.1063/1.4945586
  18. M. Martin, D. Caliste, R. Cipro, R. Alcotte, J. Moeyaert, S. David, F. Bassani, T. Cerba, Y. Bogumilowicz, E. Sanchez, Z. Ye, X. Y. Bao, J. B. Pin, T. Baron, P. Pochet, 'Toward the III–V/Si co-integration by controlling the biatomic steps on hydrogenated Si(001)', Appl. Phys. Lett. 109, 253103 (2016). DOI: 10.1063/1.4972394, https://doi.org/10.1063/1.4972394
  19. C. Prohl, H. Döscher, P. Kleinschmidt, T. Hannappel, A. Lenz, 'Cross-sectional scanning tunneling microscopy of antiphase boundaries in epitaxially grown GaP layers on Si(001)', J. Vac. Sci. Technol. A 34, 031102 (2016). DOI: 10.1116/1.4945992, https://doi.org/10.1116/1.4945992
  20. M. Feifel, J. Ohlmann, J. Benick, T. Rachow, S. Janz, M. Hermle, F. Dimroth, J. Belz, A. Beyer, K. Volz, D. Lackner, 'MOVPE grown gallium phosphide–silicon heterojunction solar cells', IEEE J. Photovolt. 7, 502–507 (2017). DOI: 10.1109/JPHOTOV.2016.2642645, https://doi.org/10.1109/JPHOTOV.2016.2642645
  21. A. Lenz, O. Supplie, E. Lenz, P. Kleinschmidt, T. Hannappel, 'Interface of GaP/Si(001) and antiphase boundary facet-type determination', J. Appl. Phys. 125, 045304 (2019). DOI: 10.1063/1.5080547, https://doi.org/10.1063/1.5080547
  22. L. Ostheim, P. J. Klar, Y. Moryson, M. Rohnke, A. Beyer, M. Volk, M. Munde, W. Stolz, K. Volz, 'Effect of the interface morphology on the lateral electron transport in (001) GaP/Si heterostructures', J. Appl. Phys. 126, 215704 (2019). DOI: 10.1063/1.5124049, https://doi.org/10.1063/1.5124049
  23. A. Boley, E. Luna, C. Zhang, N. Faleev, C. B. Honsberg, D. J. Smith, 'Interfacial intermixing and anti-phase boundaries in GaP/Si(0 0 1) heterostructures', J. Cryst. Growth 562, 126059 (2021). DOI: 10.1016/j.jcrysgro.2021.126059, https://doi.org/10.1016/j.jcrysgro.2021.126059
  24. O. Romanyuk, A. Paszuk, I. Gordeev, R. G. Wilks, S. Ueda, C. Hartmann, R. Félix, M. Bär, C. Schlueter, A. Gloskovskii, I. Bartoš, M. Nandy, J. Houdková, P. Jiříček, W. Jaegermann, J. P. Hofmann, T. Hannappel, 'Combining advanced photoelectron spectroscopy approaches to analyse deeply buried GaP(As)/Si(1 0 0) interfaces: Interfacial chemical states and complete band energy diagrams', Appl. Surf. Sci. 605, 154630 (2022). DOI: 10.1016/j.apsusc.2022.154630, https://doi.org/10.1016/j.apsusc.2022.154630
  25. D. Gupta, S. Pallikkara Chandrasekharan, S. Thébaud, C. Cornet, L. Pedesseau, 'Stability of monodomain III-V crystals and antiphase boundaries over a Si monoatomic step', Appl. Surf. Sci. 678, 161076 (2024). DOI: 10.1016/j.apsusc.2024.161076 (preprint: arXiv:2406.09476), https://doi.org/10.1016/j.apsusc.2024.161076
  26. P. Kleinschmidt, D. Bratek, M. Nandy, J. Koch, V. S. Chejarla, A. Beyer, K. D. Hanke, K. Volz, A. Paszuk, T. Hannappel, 'Atomic surface structure of antiphase domains in heteroepitaxial GaP films grown on arsenic-terminated Si(100)', J. Cryst. Growth 677, 128450 (2026). DOI: 10.1016/j.jcrysgro.2025.128450, https://doi.org/10.1016/j.jcrysgro.2025.128450