1202
CMOS Compatible in-Situ N-Type Doping of Ge Using New Generation Doping Agents P(MH3)3 and As(MH3)3 (M=Si, Ge)

Monday, October 12, 2015: 09:40
Curtis A (Hyatt Regency)
C. Xu (Department of Physics, Arizona State Univ.), J. D. Gallagher (Department of Physics, Arizona State Univ.), C. Senaratne, P. Sims (Dept. of Chemistry and Biochemistry, Arizona State Univ.), J. Kouvetakis (Dept. of Chemistry and Biochemistry, Arizona State Univ.), and J. Menendez (Department of Physics, Arizona State Univ.)
N-type doping of Ge is essential for the realization of Ge based n-MOSFETs, which are envisioned to take advantage of the much higher electron mobility in this material relative to Si. Unfortunately, the low solubility and fast diffusion of n-type dopants in Ge make this task very challenging. Dopant-atom implantation followed by annealing to activate carriers is a standard approach in Si but is not straightforward in Ge due to difficulties in controlling the doping level and doping depth at the same time, while the high temperature used in annealing is not fully compatible with CMOS processing conditions. Alternatively, in-situ doping should in principle enable independent control of doping depth and doping level. PH3, a standard precursor for in situ doping of Si, has in fact been used to dope Ge by several groups, while AsH3 has only been studied by a few researches. However, the quality of the in situ doped Ge layers has not been definitively demonstrated from a device aspect, and no convincing data has shown doping levels approaching the n-MOSFET requirement of n ~1×1020 cm-3.

In this study, higher order germanes (Ge3H8, Ge4H10) are chosen as Ge sources, and a series of new generation molecules P(MH3)3 and As(MH3)3 (M=Si, Ge) are employed as doping sources. These highly active precursors make it possible to grow highly doped, device-quality Ge films at low temperatures (T<350°C). Major findings include: 1), flat doping profiles with well-defined n-i interface and carrier concentrations up to 8.5×1019 cm-3 according to IR ellipsometry; 2), high mobility and low resistivity that are comparable to bulk Ge; 3), Ge based pin photodiodes with performance comparable to state-of-the-art devices; 4), essentially full ionization is seen even at the highest doping levels beyond the solubility limit of P in Ge; 5), the diffusion mechanism of P in Ge-rich GeSi alloys grown with P(SiH3)3 seems to be different from that of P in pure Ge.

For P-doped Ge, the films are grown on Ge-buffered Si(100) at 340°C using two methods. The first approach is based on gas-source molecular epitaxy (GSME) using Ge4H10 as the Ge-source to grow films with carrier densities up to 3.5×1019cm-3. The second approach uses an ultra-high vacuum chemical vapor deposition (UHV-CVD) technique and Ge3H8 as the Ge-source to achieve even higher carrier concentrations up to 6.2×1019cm-3. For the P(SiH3)3 case, the amount of Si incorporated in GSME growths equals or exceeds the 3:1 ratio in the P(SiH3)3 compound, whereas the Si:P ratios in CVD growths are well below the 3:1 value. 

The As-doped Ge films are grown on Ge-buffered Si(100) at 330°C using the UHV-CVD approach. Higher carrier concentrations up to 8.5×1019 cm-3 have been achieved in this case. The resistivities of As doped Ge films are slightly higher than those measured in the P-doped Ge counterparts, which agree with literature results for bulk doped Ge that date back to the 1950s. The XRD 004 diffraction scan shows a shoulder next to the Ge buffer peak, indicating a higher lattice constant c relative to pure Ge. By contrast, no lattice contracting effects are seen with P doping. These observations are consistent with a strong electronic contribution— in addition to standard atomic size effects—to the doping dependence of the lattice constant.  The Si incorporation amount in the Ge lattice when using As(SiH3)3 is only at doping levels (0-9×1019cm-3), indicating that the growth mechanism is the same as when using P(SiH3)3 as the doping source.