Mazes and meso-islands: Impact of Ag preadsorption on Ge growth on Si(111)

The preadsorption of Ag on Si(111) drastically changes the growth of Ge. In a temperature range from $400{}^{\circ }\mathrm{C}$ to $650{}^{\circ }\mathrm{C}$, Ag adsorption on Si leads to the formation of a $\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ reconstruction that exhibits a maze-like morphology on the mesoscopic scale, as observed by low-energy electron diffraction (LEED) and low-energy electron microscopy. This maze morphology can be attributed to a surface roughening on an atomic scale, induced by the re-arrangement of top layer atoms during the $7\times 7$ to $\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ transition. The subsequent deposition of Ge results in the formation of a wetting layer, the evolution of which has been found to be governed by the Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ template's maze structure, as the latter offers a high density of heterogeneous nucleation sites. Upon further Ge growth, three-dimensional islands with diameters in the micrometer range are formed, which exhibit a large and flat (111) top facet. X-ray photoemission electron microscopy reveals that during Ge growth, Ag is segregating to the surface very efficiently. Grazing-incidence x-ray diffraction and transmission electron microscopy have been used to study the composition, strain state and defect structure of the Ge islands in dependence of the growth temperature. The strain induced by lattice mismatch is found to be largely relaxed (80–90% relaxation) in the investigated growth temperature range from 400 to $600{}^{\circ }\mathrm{C}$, which is confirmed by high-resolution LEED measurements. As a main relaxation mechanism, the formation of interfacial misfit dislocations has been identified. Interdiffusion of Si into the Ge islands becomes more and more pronounced for increasing growth temperature, whereas the formation of twinned Ge regions can drastically be suppressed at higher temperature.

Figure 1

Bright-field LEEM images obtained during adsorption of Ag on Si(111) at $550{}^{\circ }\mathrm{C}$. The Ag coverage is indicated in each frame.

Figure 2

(a) Bright-field LEEM image after Ag saturation of the Si(111) surface at $550{}^{\circ }\mathrm{C}$. The contrast leading to the “maze” pattern is most likely due to an atomic step height surface roughness. (b) LEED image (logarithmic grayscale, integer order spots indexed) obtained from the same surface, showing a single-phase $\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ diffraction pattern, proving that the contrast in (a) does not arise from different superstructures.

Figure 3

(a) Top view of the Si(111)-$7\times 7$ surface according to the dimer-adatom-stacking-fault (DAS) model by Takayanagi et al. [86]. (b) Cross section from the lower left to the upper right corner hole of the reconstruction. Within the topmost four monolayers and including the adatoms, the $7\times 7$ reconstruction comprises of 200 silicon atoms, as compared to 196 atoms in case of 2 bilayers of the bulk structure.

Figure 4

(a) Top view of the Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ surface according to the honeycomb-chained-trimer (HCT) model by Ding et al. [62]. The small and the large shaded areas mark a $1\times 1$ and a $\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ unit mesh, respectively. (b) Cross section from the bottom left towards the top right of the slab shown in (a). The top bilayer of this reconstruction is a mixed bilayer and consists of one monolayer of Si and one of Ag.

Figure 5

Bright-field LEEM images during Ge growth on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }\mathrm{at} 500{}^{\circ }\mathrm{C}$. The Ge deposit is indicated in each frame. After the wetting layer formation (a)–(f), a large Ge island comes into the field of view in (g) and (h). Note that the contrast in frame (a) has been specially enhanced to make the initial “maze” structure better visible at this electron energy (9.0 eV). The display contrast of the images has been adjusted for each frame separately (see Fig. 6 for reference).

Figure 6

Field-of-view averaged LEEM intensity for the experiment shown in Fig. 5, i.e., for bright-field LEEM at 9.0 eV, as a function of Ge deposit during Ge growth on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ at $500{}^{\circ }\mathrm{C}$. The labels ‘a’ to ‘g’ refer to the image frames in Fig. 5.

Figure 7

Bright-field LEEM images after growth of 5.2 BL Ge on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ at $500{}^{\circ }\mathrm{C}$. (a) Overview of the morphology. The Ge islands appear dark at this energy (25.0 eV). (b) Zoom into the edge of an island, marked with a white circle in frame (a). At this energy (9.3 eV), the Ge islands appear bright. The part of the wetting layer (WL) marked with a white square is shown at larger scale and with enhanced grayscale contrast in the inset at the bottom left of frame (b), in order to make the maze structure on the WL better visible. Owing to the change of magnification, the image in frame (b) is rotated with respect to the one in frame (a).

Figure 8

XPEEM images, (a) and (b), of the same surface area as depicted in Fig. 7, using (a) Si $2p$ and (b) Ge $3d$ photoelectrons for imaging. The incident photon energy was 500 eV. The local spectra in (c) were obtained from XPEEM image series, integrating the intensity in the squares drawn for reference in (a) and (b).

Figure 9

Cross-sectional HAADF-STEM image of Ge islands (bright) grown at $550{}^{\circ }\mathrm{C}$ on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$. The Ge deposit is 18 BL.

Figure 10

SPA-LEED patterns (inverse logarithmic grayscale) (a) from a clean Si(111)-$7\times 7$ surface, and (b) after growth of about 18 BL Ge on Ag:Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ at $500{}^{\circ }\mathrm{C}$. Integer order spots are indexed in (a). Note the splitting of the spots in (b), best to be seen near the corners of the viewgraph. Line scans (c) and (d) were taken through (c) an integer order spot and (d) a third order spot, marked by a circle and a square in (b), respectively.

Figure 11

Bright-field LEEM images after Ge growth at $480{}^{\circ }\mathrm{C}$ on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$, with a Ge deposit of 20.7 BL. The image in frame (b) is displayed at increased grayscale contrast.

Figure 12

(a) Evolution of the integral intensity of the Si(10) and the Ge(10) LEED spots at 80 eV during Ge deposition on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ at $440{}^{\circ }\mathrm{C}$. Both curves have been scaled differently for display reasons. The inset shows the details of the Si(10) intensity evolution in the initial stage, on logarithmic scale. Frames (b) and (c) show the LEED pattern after growth in the vicinity of the (00) spot at 80 eV and 84 eV, respectively.

Figure 13

Reciprocal space maps in a $Q_{\parallel }$ plane in the vicinity of the $\left(\overline{2}\overline{2}4\right)$ reflection [i.e., near the $\left(\overline{3}0\right)$-CTR] at $l\approx 0.07$, for Ge islands grown on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$at (a) $400{}^{\circ }\mathrm{C}$, (b) $500{}^{\circ }\mathrm{C}$, and (c) $600{}^{\circ }\mathrm{C}$. Satellite spots are indicated with small black dots in (a). In frame (d), the temperature dependence of the lateral reciprocal lattice parameter $a_{\parallel ,\mathrm{Ge}}^{*}$ of the Ge islands is shown (in units of the Si bulk value $a_{\parallel ,\mathrm{Si}}^{*}$), as determined from GIXRD ($\circ$) and SPA-LEED ($▵$).

Figure 14

Schematic of the reciprocal space of the Si(111) surface in the (111)-($11\overline{2}$) plane. $h, k$, and $l$ refer to LEED coordinates.

Figure 15

Reciprocal space maps in a $Q_{\parallel }-Q_{\perp }$ plane, recorded on the Ge(20) CTR, for Ge islands grown on Ag/Si(111)-$\sqrt{3}\times \sqrt{3}$-R ${30}^{\circ }$ at (a) $400{}^{\circ }\mathrm{C}$ and (b) $550{}^{\circ }\mathrm{C}$. In frame (c), the temperature dependence of the vertical reciprocal lattice parameter $a_{\perp ,\mathrm{Ge}}^{*}$ of the Ge islands is shown (in units of the Si bulk value $a_{\perp ,\mathrm{Si}}^{*}$) for both regular Ge ($\circ$) and twinned Ge ($\square$).

Figure 16

Growth temperature dependence (a) of the strain parameter and (b) of the average Si content of the Ge islands, as determined from GIXRD ($○$) and high-resolution TEM ($\blacksquare$).

Figure 17

Growth temperature dependence of the volume fraction of twinned Ge.

Figure 18

(a) High-resolution transmission electron microscopy (HRTEM) image of a Ge island on a sample prepared at $550{}^{\circ }\mathrm{C}$, recorded in $\left[1\overline{1}0\right]$ zone axis. (b) Diffractogram (Fourier transform) of the substrate region of the HRTEM image in (a). (c) Diffractogram of the island region in (a). (d) Fourier filtered image of the region marked with dashed square in (a), emphasizing A-type and B-type orientation. (e) Zoom into region marked with solid square in (a), exposing a dislocation.

Figure 19

Local (a) vertical and (b) lateral lattice parameters in units of the bulk Si lattice parameters, as determined from HRTEM, and (c) corresponding composition map, for the sample area shown in Fig. 18, i.e., for a growth temperature of $550{}^{\circ }\mathrm{C}$; (d) comparison of local vertical lattice parameter [normalized to the bulk Si(111) layer spacing] as determined by HRTEM (dots) and by SANBED (solid line).

Figure 20

Ge concentration profiles along growth direction for growth temperatures of (a) $400{}^{\circ }\mathrm{C}$, (b) $450{}^{\circ }\mathrm{C}$, and (c) $550{}^{\circ }\mathrm{C}$. Dots represent the local Ge concentration as determined from HRTEM data, the solid lines are a numerical fit to the Muraki model [104] [cf. Eq. (6)]. The values of $R, x_0$, and of the strain parameter $\gamma$ [cf. Eq. (4)] are indicated in each frame.