Elsevier

Surface Science

Volume 458, Issues 1–3, 20 June 2000, Pages 147-154
Surface Science

Hydrogen-induced domain-wall structure on Si(113)

https://doi.org/10.1016/S0039-6028(00)00434-9Get rights and content

Abstract

In-situ high resolution electron diffraction (SPA-LEED) and scanning tunneling microscopy (STM) measurements during the chemical vapor deposition growth from disilane (Si2H6) on Si(113) show a variety of different surface reconstructions, depending on the equilibrium hydrogen coverage on the surface adjustable by substrate temperature. At high H coverage, a regularly formed missing row structure with light domain walls is observed. At low H coverage, STM images reveal large uncovered areas between several hydrogen-terminated rows, changing the light domain wall structure into a heavy domain wall structure. The domain wall formation depends only on the equilibrium H coverage on the surface and is fully reversible with desorption of the hydrogen. The relation between the observed surface structures and thermal desorption of H is shown in a phase diagram. The activation energy for the transitions was found to be 2.1 eV and agrees well with the desorption energy for atomic hydrogen determined by growth oscillations.

Introduction

The use of chemical vapor deposition (CVD) techniques is of large technological relevance for the growth of thin epitaxial semiconductor layers. However, for structured films, application of optical lithography in combination with wet etching processes is necessary for patterning the films after growth. The use of selective epitaxial growth (SEG) techniques reduces the total number of technological steps to be taken. In the case of disilane (Si2H6) CVD on an oxide patterned surface, deposition of Si from the precursor molecules only takes place in oxide-free regions on the surface. However, the formation of a (113) facet plane at the interface between the silicon substrate and the oxide window instead of a smooth film in connection with the oxide was reported [1]. This (113) type faceting could be explained by an atomistic model of step formation energies [2], even though the proposed model completely neglects the influence of hydrogen, which is an essential element of the disilane precursor. For the case of the cited SEG experiment [1], [2], this approximation is correct, because the chosen growth temperature was much higher than 550°C, which is the desorption temperature of hydrogen (β1 peak) from the (111) [3] as well as from the (001) plane.

However, during disilane CVD on Si(111) at temperatures much below the β1 peak, the formation of (113) facets could be observed, too [4]. This is surprising as the (111) plane is expected to be energetically more favourable than (113) [5], [6]. The change in surface morphology must therefore be due to the presence of hydrogen. During Si growth under the presence of adsorbed hydrogen, the formation of a flat film with (111) orientation is kinetically hindered [7]. In the temperature regime of layer-by-layer growth, the surface roughness increases [the electron diffraction (LEED) intensity oscillation amplitude decreases exponentially], and a (113) facet as well as a much steeper (334) facet could be observed [7], [8]. For both reasons, the technological interest in CVD techniques as well as a basic interest in high index planes, it would be interesting to take a closer look at the Si(113) surface and its behaviour after exposure to adsorbates. Some authors have already reported on the stability [6], [9] and structure [10], [11], [12], [13], [14] of the clean (113) surface as well as the (113) surface with adsorbates [15], [16], and recently, a scanning tunneling microscopy (STM) CVD growth study of disilane on Si(113) has been performed [17]. The cited work confirms the structure of the completely hydrogenated surface to be (2×2), and in addition, (2×7) and (2×5) structures have been observed.

Here, we present a high resolution electron diffraction (SPA-LEED) [18] and STM study on the CVD growth from disilane on the Si(113) surface. Depending on the disilane flux and sample temperature, we observe several different superstructures. The appearance of all these is attributed to the influence of hydrogen: variations in sample temperature at a given disilane flux only change the equilibrium hydrogen coverage on the surface. All observed structures are explained by a simple domain wall model. In the layer-by-layer growth regime, it is possible to determine the hydrogen desorption energy from the growth rate and to compare the energy necessary for hydrogen desorption with the energy necessary for structural changes on the surface.

Section snippets

Experimental

Disilane CVD experiments were performed in a standard ultra-high-vacuum (UHV) chamber, and a base pressure of 10−10 mbar was routinely achieved. The system was equipped with a quadrupole mass-spectrometer for residual gas analysis and a SPA-LEED for the high-resolution electron diffraction measurements. The disilane flux was adjusted by a leak valve and calibrated by monitoring the background pressure of the chamber. The use of a CVD process for hydrogen adsorption has the advantage of not being

SPA-LEED results

Deposition of hydrogen, due to disilane CVD, transforms the reconstruction of the clean surface into various H-covered superstructures, depending on the sample temperature and disilane partial pressure (Fig. 1). As already known from previous work [15], [17], a high disilane pressure induces a (2×2) structure (Fig. 1a), indicating a saturation of the whole surface with H. Reduction of disilane pressure leads to the formation of a (2×7) structure (shown in Fig. 1b). The transition between these

Transformation in detail

For a detailed understanding of the phase transitions from a saturation-covered surface to an uncovered surface, subsequent scans analog to Fig. 2 were taken at different temperatures and fixed disilane pressure and afterwards, assembled to gray-scale images. The top of both images in Fig. 5 shows the integer-order spots of the (113) surface and, additionally, (2×2) spots. Increasing the temperature (downwards in Fig. 5a) induces a splitting of the (2×2) spots, and additional spots become

STM results

In-situ STM measurements were achieved in addition to the SPA-LEED investigation. Several of the domain wall structures are clearly resolved. The hydrogen-saturated surface in Fig. 6a shows the ‘bean-like’ (2×2) reconstruction already reported by Dorna et al. [17]. A small defect exists in the periodic arrangement of the (2×2) beans (feature ‘A’) where H is missing, but the remainig part of the surface is completely covered with hydrogen. With the sample temperature being fixed, a reduction in

Conclusions

We have performed SPA-LEED and STM studies of the hydrogen-induced domain wall structures formed during disilane CVD on Si(113) substrates. Depending on the equilibrium hydrogen coverage, we report on various different reconstructions depending on the sample temperature and disilane partial pressure. An overview of the observed reconstructions is given in Fig. 7. The H coverage decreases from top to bottom. At saturation coverage, a (2×2) structure is observed by LEED and STM, in agreement with

Acknowledgements

The authors would like to thank M. Kammler, Ch. Tegenkamp and H. Pfnür for helpful discussions.

References (19)

  • J.M. Bermond et al.

    Surf. Sci.

    (1995)
  • B.Z. Olshanetsky et al.

    Surf. Sci.

    (1981)
  • U. Myler et al.

    Surf. Sci.

    (1989)
  • K. Jacobi et al.

    Surf. Sci.

    (1993)
  • U. Scheithauer et al.

    Surf. Sci.

    (1986)
  • H. Hirayama et al.

    Phys. Rev. B

    (1993)
  • A. Oshiyama

    Phys. Rev. Lett.

    (1995)
  • R. Imbihl et al.

    Phys. Rev. B

    (1989)
  • U. Köhler et al.

    Phys. Stat. Sol. A

    (1997)
There are more references available in the full text version of this article.

Cited by (6)

  • Time-evolution of thermal oxidation on high-index silicon surfaces: Real-time photoemission spectroscopic study with synchrotron radiation

    2012, Surface Science
    Citation Excerpt :

    This model has been supported by investigations using scanning tunneling microscopy (STM) [6,7]. The mechanisms of various processes on Si(113) have been investigated: oxidation [8–10], hydrogenation [11–14], heteroepitaxial growth of Ga [15,16], In [15,16], Pb [15], Al [16], Ge [17,18], and Sb [19], homoepitaxial growth involving disilane (Si2H6) species [20] and one-dimensional Ge nanowire formation [21,22]. Growth of alkali metal [23] and alkali metal-promoted oxidation [24–26] have also drawn considerable interest.

1

Present address: Institut für Laser und Plasmaphysik, Universität Essen, Universitätsstraße 5, 45117 Essen, Germany.

View full text