Growth kinetics of Ge islands during Ga-surfactant-mediated ultrahigh vacuum chemical vapor deposition on Si(001)

The surfactant effect of Ga pre-deposited prior to ultrahigh vacuum (UHV) chemical vapor deposition (CVD) of Ge on Si(001) is investigated using in situ transmission electron microscopy. Island shape, nucleation and growth kinetics, and the influence of growth temperature and pressure are studied. When 1 monolayer of Ga covers the surface, Ge island nucleation occurs after a $5–10\times$ longer growth time than without Ga for the same growth temperature and pressure, and the island density then varies with time in a more complex way than for Ga-free growth. The islands formed have a different shape and aspect ratio compared to conventional hut and dome clusters, and the critical lateral size for dislocation nucleation is reduced. Ex situ medium energy ion scattering and x-ray photoelectron scattering measurements of the total amount of Ge deposited before island nucleation demonstrate that the Ga layer decreases the surface reactivity and inhibits Ge growth. We discuss a simple model in which the surfactant-mediated growth kinetics are principally driven by the balance between Ge deposition and Ga desorption. We show that Ga surfactant-mediated UHV CVD can be used to produce Ge quantum dots with a higher surface density and a thinner wetting layer than is possible using conventional CVD.

Figure 1

Desorption of Ga (initial layer thickness several ML) during UHV TEM in situ annealing. The excess Ga forms islands before desorbing at $600°\mathrm{C}$ or above, leaving $1\mathrm{ML}$ on the surface. Images were taken using bright field conditions and are shown after $\sim 2\min$ at the temperatures indicated. All images in this paper [except Figs. 4b and 10] were obtained by averaging 2–4 single frames from videos recorded, using an image intensifier, onto VHS tapes at 30 frames per second during growth. The scale bar corresponds to $50\mathrm{nm}$.

Figure 2

Plan view TEM images taken after the in situ UHV-CVD growth of Ge without Ga: (a) several huts and a larger dome, formed after growth at $T_{\mathrm{G}}=600°\mathrm{C}$, $P_{\mathrm{G}}=1.1\times {10}^{-7}\text{Torr}$ and $t_{\mathrm{G}}=88\mathrm{s}$. (b) One very large dislocated dome island, formed after growth at $T_{\mathrm{G}}=530°\mathrm{C}$, $P_{\mathrm{G}}=1.0\times {10}^{-7}\mathrm{Torr}$ and $t_{\mathrm{G}}=5\min$. The scale bar corresponds to $250\mathrm{nm}$.

Figure 3

Variation of the product of island nucleation time $\left(t_{\mathrm{N}}\right)$ and growth pressure $\left(P\right)$ with temperature $\left({\mathit{T}}_{\mathrm{G}}\right)$, for $P=1(\pm 0.2)\times {10}^{-7}\text{Torr}$ (triangles) and $1(\pm 0.2)\times {10}^{-6}\text{Torr}$ (circles). Solid symbols correspond to Ga-mediated growth and open symbols correspond to Ge growth without Ga. The dotted and solid lines correspond to the best exponential fit of the Ga-mediated data at $P={10}^{-7}$ and ${10}^{-6}\text{Torr}$, respectively. The arrow indicates a sample on which MEIS and XPS were performed. 1 langmuir $\left(\mathrm{L}\right)={10}^{-6}\text{Torr}\mathrm{s}$.

Figure 4

Plan view TEM images of Ga-mediated grown Ge layers recorded in situ: (a) $T_{\mathrm{G}}=650°\mathrm{C}$, $P_{\mathrm{G}}=1.0\times {10}^{-7}\text{Torr}$ and $t_{\mathrm{G}}=317\mathrm{s}$, where the small, undislocated islands appear with very faint dark and bright contrast; (b) $T_{\mathrm{G}}=530°\mathrm{C}$, $P_{\mathrm{G}}=1.0\times {10}^{-7}\text{Torr}$ and $t_{\mathrm{G}}=2304\mathrm{s}$, where much larger square dislocated islands are visible. The scale bars correspond to $100\mathrm{nm}$.

Figure 5

The height $\left(h\right)$ of a population of Ga-induced Ge islands versus their lateral size $\left(L\right)$. $h$ and $L$ were measured by AFM in air for two different growth conditions: $T_{\mathrm{G}}=550°\mathrm{C}$, $P_{\mathrm{G}}=1.0\times {10}^{-7}\text{Torr}$, $t_{\mathrm{G}}=500\mathrm{s}$ and $T_{\mathrm{G}}=550°\mathrm{C}$, $P_{\mathrm{G}}=1.1\times {10}^{-6}\text{Torr}$, $t_{\mathrm{G}}=1920\mathrm{s}$. The two solid lines correspond to the relationship of $h$ and $L$ expected for hut $(h∕L=0.1)$ and dome $(h∕L=0.2)$ islands. Dotted lines corresponding to aspect ratios of 0.05 and 0.3 are also shown for comparison.

Figure 6

Plan view TEM images obtained during in situ annealing of Ga-mediated grown Ge layers: (a) the annealing of less than $3\mathrm{ML}$s of Ge deposited at $613°\mathrm{C}$ ($\sim 1\min 30\mathrm{s}$ between images). The small islands (weak contrast) which formed during Ga-mediated growth disappear leaving a flat Ge layer as the Ga desorbs. (b) The annealing of a thicker Ge layer deposited at $530°\mathrm{C}$ composed of larger square shaped dislocated islands, which transform into domes ($\sim 2\min 30\mathrm{s}$ between images). In each sequence the same area is shown in each image. The scale bar corresponds to a length of $50\mathrm{nm}$.

Figure 7

(a) Island density versus Ge dose (the product of growth time and pressure), for Ga-mediated growths performed on the same sample at a pressure of $P=1(\pm 0.2)\times {10}^{-6}\text{Torr}$ and different temperatures: $T=530°\mathrm{C}$ (∎), $550°\mathrm{C}$ (◻), and $600°\mathrm{C}$ (●). (b) Island density versus Ge dose, for Ga-mediated growths performed at a temperature of $T=600°\mathrm{C}$ and different pressures: $P=1.1\times {10}^{-8}\text{Torr}$ (∎) and $P=1.0\times {10}^{-6}\text{Torr}$ (●). 1 langmuir $\left(\mathrm{L}\right)={10}^{-6}\text{Torr}\mathrm{s}$.

Figure 8

Variation of the island density with growth time during Ga-mediated growth of Ge ($T_{\mathrm{G}}=600°\mathrm{C}$ and $P_{\mathrm{G}}=1.1\times {10}^{-8}\text{Torr}$) and plan view TEM images of the same area obtained in situ during the growth (scale bar $50\mathrm{nm}$ for all four pictures). The same area of the sample (of size $1.5\times {10}^5{\mathrm{nm}}^2$) was analyzed for each data point; thus 4–40 islands were counted per point but the error is not independent for each data point. The maximum error bar, calculated for the data point corresponding to the maximum island density, is shown. The arrow shows the weak contrast from a small Ga-mediated island, which disappears as the larger domes grow.

Figure 9

Variation of the island density with growth time during Ga-mediated growth of Ge ($T_{\mathrm{G}}=530°\mathrm{C}$ and $P_{\mathrm{G}}=1.0\times {10}^{-6}\text{Torr}$). The experiment was carried out on the same sample as that shown in Fig. 8. Plan view TEM images obtained in situ during the growth are also shown (scale bar $50\mathrm{nm}$ for the four pictures). The maximum error bar, calculated for the data point corresponding to the maximum island density, is the size of the squares used in the graph.

Figure 10

Plan view TEM image taken after the Ga-mediated growth of Ge. The growth was performed at $T_{\mathrm{G}}=530°\mathrm{C}$, $P_{\mathrm{G}}=1.0\times {10}^{-6}\text{Torr}$ and over a long time $(t_{\mathrm{G}}=428\mathrm{s})$ in order to emphasize the beam effect. The black circle indicates the approximate sample area exposed to the electron beam during growth. The scale bar corresponds to $250\mathrm{nm}$.

Figure 11

Plan view TEM image obtained after Ga-mediated growth of Ge at $T_{\mathrm{G}}=530°\mathrm{C}$, $P_{\mathrm{G}}=9.7\times {10}^{-7}\text{Torr}$ with $t_{\mathrm{G}}=525\mathrm{s}$: $12⩽L⩽24\mathrm{nm}$ and $d\sim 9.65\times {10}^{10}{\mathrm{cm}}^{-2}$. The scale bar corresponds to $50\mathrm{nm}$.