A Benchmark of Germane and Digermane for the Low Temperature Growth of Intrinsic and Heavily in-situ Boron-Doped SiGe

Abstract

We have recently demonstrated the interest of using Si2H6 + GeH4 for the 450°C-500°C epitaxy of (i) intrinsic SiGe layers and (ii) SiGe:B raised sources and drains [1-4]. One way of further increasing performances (in terms of growth rates at really low temperatures and Ge contents reachable) could be to use a higher order Ge precursor, such as digermane (Ge2H6). We have thus benchmarked GeH4 and Ge2H6 for the Reduced Pressure–Chemical Vapor Deposition (RPCVD) of SiGe and SiGe:B in an Applied Materials 200 mm Epi Centura 5200 tool. We have first of all compared the impact of the GeH4 and Ge2H6 mass-flows on the SiGe growth kinetics (at 500°C, 20 Torr and with a fixed Si2H6 flow). Whatever the source of Ge atoms, we had a linear increase of the SiGe growth rate with the Ge flow (cf. Figure 1, top). This was due to the catalysis of H desorption by Ge surface atoms, freeing sites for incorporation. Because of its higher reactivity, the SiGe growth rate was 8 to 10 times higher with Ge2H6 than with GeH4, however. We were also faced with a sub-linear increase of the Ge content x with F(Ge)/F(Si) (bottom of Figure 1). Using Ge2H6 instead of GeH4 otherwise yielded much higher Ge contents for a given disilane flow: in the 35-48% range instead of 20-33% only. The evolution of x with the F(Ge) flow was well accounted for by x2/(1-x)=nF(Ge)/F(Si) relationships (F(Ge) = F(GeH4) or 2F(Ge2H6)), with n = 0.47 for germane and n = 1.74 for digermane. We have then studied the evolution of the SiGe growth rate with temperature (Figure 2), for both types of Ge precursors. The use of GeH4 led to an exponential increase of the 20 Torr SiGe growth rate with temperature (with a 42 kcal.mol.-1 activation energy), together with a slight linear decrease of the Ge concentration. The situation was very different with Ge2H6. As the growth temperature increases, we were faced with a slow exponential increase of the SiGe growth rate (Ea = 15 kcal.mol.-1), together with a huge linear decrease of the Ge content. We have also quantified the impact HCl, an etchant gas, had on the growth kinetics of intrinsic SiGe with fixed amounts of disilane and digermane. Adding HCl led at 500°C, 20 Torr to a significant decrease of the growth rate (from 7.8 down to 4.2 nm min.-1), together with a significant Ge concentration increase (from 35% up to 55%). We have otherwise studied the impact of the B2H6 flow on the SiGe:B growth kinetics at 500°C, 20 Torr. The SiGe:B growth rate and the real or “apparent” Ge contents associated with such SiGe:B layers are plotted in Figure 3 as a function of the B2H6 mass-flow. Adding small amounts of diborane to Si2H6 + Ge2H6 + HCl led to a dramatic increase of the SiGe:B growth rate, from 5.8 up to 13.2 nm.min.-1. H desorption was indeed catalysed by the presence of B surface atoms. The “apparent” Ge concentration is coming from X-Ray Diffraction when supposing that an in-situ boron-doped SiGe layer behaves as an intrinsic one. This is however not true for high doping levels. Boron atoms, which are much smaller than Si or Ge atoms, then compensate the compressive strain in SiGe layers, yielding Ge concentrations artificially lower than the real ones. The concentration difference thus reflects the compressive strain compensation by substitutional B atoms. There is in Fig. 3 a definite decrease of the real and “apparent” Ge concentration with the B2H6 flow, from 47% down to 15% and 20%, respectively. We have also performed SIMS depth profiling of the B atoms in those SiGe:B layers. Their concentration increased linearly with the B2H6 mass-flow, reaching at most 3.6x1020 cm-3 in single crystalline layers (Figure 4, left) The associated resistivity (Figure 4, right) was as low as 0.43 mΩ.cm. For higher B2H6 flows, we had a resistivity increase, which was likely due to the crystalline quality degradation then the transformation into poly-SiGe:B evidenced in X-Ray Diffraction. Small islands were otherwise present in limited numbers on the surface of those SiGe:B layers, with no clear dependence on the B2H6 flow. [1]J.M. Hartmann et al. ECS J. Sol. State Sci. and Technol. 3 (11) P382 (2014). [2] J.M. Hartmann et al. Thin Sol. Films 557, 19 (2014). [3] J. Aubin et al. Thin Sol. Films 602, 36 (2016). [4] J. Aubin et al. Semicond. Sci. Technol. 30, 115006 (2015). Figure 1