We have evaluated the feasability of selectively growing (at 600°C-700°C, 20 Torr and with a SiH2Cl2 + GeH4 + B2H6 + HCl chemistry) SiGe:B Raised Sources and Drains on each side of PMOS FD-SOI devices with a vertical Ge concentration gradient. The reason was twofold : (i) inject meaningful amounts of uniaxial compressive strain and boost hole mobility thanks to really high Ge concentrations (50%) in layers on each side of the channel and (ii) benefit from SiGe:B 20% to 30% layers on top, that are easier to germano-salicide, for instance with Ni or NiPt. As the diborane flow increased, we had, whatever the Ge concentration (20%, 30% or 50%), (i) sub-linear SiGe:B growth rate increases (from +14% up to + 23%) and (ii) linear decreases of the “apparent” Ge concentration from X-Ray Diffraction (XRD), as shown in Figures 1a and 1b. They were due to (i) a catalysis of H/Cl desorption by B adatoms and (ii) compressive strain compensation by small size B atoms, respectively. The resulting substitutional B concentrations increased almost linearly with the B2H6 mass-flow. The slopes of those increases were steeper and B concentrations systematically higher for higher Ge contents, however (at most 2.2x1020 cm-3 for 50% of Ge), as illustrated in Figure 1c. We otherwise showed, for Si0.7Ge0.3:B layers, that the Ge atomic concentration was steady as the B2H6 mass-flow increased. Atomic, substitutional and ionized B concentrations (from Secondary Ions Mass Spectrometry, XRD and Hall effect measurements) were the same for the lowest diborane flow probed (e.g. 7x1019 cm-3). For a medium flow, atomic and substitutional B concentrations were equal (e.g. 1.4x1020 cm-3), while the ionized B concentration was ~80% of it. Discrepancies were the highest for the highest diborane flow probed, with substitutional and ionized B concentrations ~80% and ~60% the atomic B concentration of 2.3x1020 cm-3, then (Figure 1d). The specifics of SiGe:B Selective Epitaxial Growth (SEG) in sources and drains regions on each side of FD-SOI devices (Figure 2) were then described, such as (i) loading effects, because of the presence of dielectrics, resulting in significant growth rate increases (x1.5 - x1.7), together with slight Ge and B concentrations increases that might impact the layer quality on patterned wafers, (ii) facet formation, (iii) the importance of proper surface preparation prior to epitaxy and of tight SiN hard masks on top of poly-Si gates, (iv) the strain loss happening upon germano-salicidation and so on. Growing a thin Si(:B) cap on SiGe:B is another way of obtaining uniform and stable metallic contacts. This is why we have studied the capping, with pure Si, of SiGe 20% or 30% layers. As growth had to be conducted, for the SEG of Si with a chlorinated chemistry, at temperatures significantly higher than that of SiGe (750°C, typically, to be compared with 700°C for SiGe 20% and 650°C only for SiGe 30%), we have quantified the impact of using temperature ramping-ups with SiH2Cl2 + HCl flowing into the growing chamber. The purpose of such active ramps was to avoid elastic strain relaxation, e.g. the formation of SiGe surface undulations that would happen with temperature ramping-ups under H2 only. Superior quality 2D SiGe / Si stacks were then obtained, as shown by X-Ray Reflectivity and XRD. Energy Dispersive X-ray maps of such stacks enabled us to quantify loading effects on patterned wafers. SiGe growth rates in small, recessed windows were 1.5 times (SiGe 30%) – 1.8 times higher (SiGe 20%) than on bulk, blanket Si. Meanwhile, Ge concentrations were 2% to 3% higher. Finally, Si growth rates were lower than bulk, most probably because of the presence of thick buried oxide layers on those patterned wafers that impacted the effective surface temperature during growth. Figure 1
Read full abstract