Very Low Temperature Epitaxy of Heavily In-situ Phosphorous Doped Ge Layers and High Sn Content GeSn Layers

Abstract

Combining in-situ phosphorous doping and tensile strain yield a direct bandgap-like behavior in germanium. The electrons given by the n-type impurities to the strained Ge crystalline lattice then occupy the L valleys of the conduction band, exalting fast, direct transitions between the Γ valley of the conduction band and the valence band. Using low temperatures (e.g. 400°C) and relatively high pressures (100 Torr, typically) enabled us to promote P incorporation in Ge:P layers [1]. With the conventional precursors used (GeH4 and PH3), we could not go to temperatures lower than 400°C, however. We have recently quantified the interest of using digermane (Ge2H6) in order to enhance the Ge growth rate at temperatures 425°C and lower [2] (by 5 to 10 orders of magnitude compared to GeH4). We have thus used digermane and phosphine (PH3) to grow Ge:P layers at 350°C, 100 Torr on top of slightly p-type {Ge Strain Relaxed Buffer / Si (001) substrate} stacks. The F(Ge2H6)/F(H2) Mass-Flow Ratio (MFR) and the growth duration were fixed (1.98x10-4 and 30 minutes, respectively). The process parameter varying was the PH3 mass-flow. We have first of all evidenced a Ge:P growth rate increase with the PH3 mass-flow (Figure 1), from 5 nm.min-1 for intrinsic Ge up to 11 nm.min-1 for the highest PH3 mass-flow probed. This could be due to (i) the catalysis of H desorption from the surface by P atoms, freeing sites for growth or/and (ii) a local surface temperature increase, caused by exothermal reactions [3]. All samples were golden and mirror-like under grazing light. Note that the situation was somewhat different at 400°C, 100 Torr with a GeH4 + PH3 chemistry. A transition from smooth/mirror-like to rough/milky surfaces and a sudden Ge:P growth rate drop were then evidenced for F(PH3)/F(H2) above 10-3 [4]. Conventional ω-2θ scans around the (004) XRD order were performed on those Ge:P layers. The resulting profiles are provided in Figure 2. For the highest two PH3 mass-flows probed, we obtained well-defined, intense Ge:P layer peaks (together with Pendellösung fringes on each side for the highest doping). We were thus able, from their angular position, to calculate the associated substitutional P concentrations, 2.7x1020 and 6.5x1020 cm-3. We have then performed SIMS depth profiling and four points probe measurements in order to gain access to the atomic and electrically active P concentrations in our Ge:P samples (Figure 3). As expected, the atomic P concentration monotonically increases with the PH3 mass-flow, from 9x1018 cm-3 for the lowest PH3 flow, up to 5x1020 cm-3 for the highest one. This increase slows down for F(PH3)/F(H2) MFRs above 2x10-6, however (in-line with Fig. 1 Ge:P growth rate evolution). The substitutional P concentrations (from XRD) measurements, are close to the atomic values. The vast majority of the P atoms thus seem to be substitutional. The evolution of the P+ ion concentration with the PH3 flow is somewhat peculiar. At first, it increases with the phosphine flow. The highest P+ concentration, 7.5x1019 cm-3, was obtained for F(PH3)/F(H2) = 1.25x10-6. This value is nearly 4 times higher than the value obtained at 400°C with GeH4 (~2x1019 cm-3). For higher PH3 flows, we then have a definite decrease of the P+ ions concentration. This could be due to an increase of point defects in the Ge:P crystal lattice.Finally, we have used AFM (Figure 4) in order to confirm that the Ge:P layers were smooth and featureless; the surface root mean square (rms) roughness and Z range associated with 20×20 μm² images of the most highly doped layer of the series were equal to 0.8 and 7.3 nm only, respectively. Small islands (density close to 1.6x107 cm−2) were by contrast present on the surface of intrinsic Ge layers.The high phosphorous-doping obtained here is very interesting to boost the optical performances of Ge. Another method in order to obtain a direct bandgap in group-IV semiconductors is to alloy Ge with Sn. GeSn layers have recently been shown to have a direct bandgap provided that the Sn content is high enough (12%) and the built-in strain low. This in mind, we will present some data on high Sn content (in the 6% to 10% range), compressively-strained GeSn layers we have grown at 325°C, 100 Torr on Ge SRBs (thanks to Ge2H6 and SnCl4). [1] J.M. Hartmann et al., J. Cryst. Growth 347 (2012) 37. [2] J. Aubin et al., J. Cryst. Growth, accepted (April 2016); doi: 10.1016/j.jcrysgro.2016.04.018. [3] D. Grützmacher et al., ECS Trans. 64 (11) (2014) 85. [4] J.M. Hartmann et al., Thin Solid Films 602 (2016) 13. Figure 1