Ge Precursors for Strained Si and Compound Semiconductors
Egbert Woelk and Deo Shenai, Rohm and Haas Electronic Materials, North Andover, Mass.; Peter Storck, Siltronic AG, Burghausen, Germany; Isabelle Sagnes, LPN-CNRS, Marcoussis, France -- Semiconductor International, 4/1/2006
|
As the silicon industry looks for new ways and materials to continue the rapid pace of increasing the speed of microelectronic devices, strain and crystal lattice engineering become crucially important. Silicon can be strained by several methods, many of which involve the use of epitaxially grown silicon germanium (SiGe).1 The epitaxial processes employed to generate strain can be quite different from each other depending mostly on the restrictions that are imposed on the grower by the thermal budget still available to the wafer to be processed. Thus, the main difference is the growth temperature.
For local application of an epitaxial film, areal selectivity is desired, and can be greatly enhanced by a halogen-containing precursor at moderate growth temperatures. For a blanket deposition, high growth rates and high crystal perfection are desired, and are commonly obtained at high growth temperatures. So far, germane has been used to grow all germanium-containing films. This is mainly due to its reasonably good temperature fit for deposition of the base layer of a SiGe HBT for BiCMOS, which is by far the oldest application of SiGe. With new SiGe films requiring new processes, germane is not the optimal match, and better suited germanium precursors are required (Fig. 1 ).
In addition to strained silicon, the International Technology Roadmap for Semiconductors (ITRS) anticipates that, in the future, III-V compound semiconductor devices will be integrated with silicon. To integrate these semiconductors on a silicon substrate, the transition from the lattice constant of silicon to the lattice constant of the III-V semiconductor has to be engineered. Here again, germanium is expected to play a key role.2 To open up new process regimes, we developed and tested several new germanium precursors that either improve film growth or enable the growth of new films. By careful selection of different ligands, the new germanium precursors balance deposition and removal rates for a given process, analogous to the chain of chlorosilanes that are most widely used in silicon epi. Chlorogermanes are different from chlorosilanes in the way that only germane (GeH4) and germaniumtetrachloride (GeCl4) are stable and useful as precursors for epi. This confronts the grower with a gap in the temperature range available for his process. GeH4 generally works well from 400 to 700°C, while GeCl4 cannot be efficiently activated below 850°C. To close the temperature gap, we have so far added two new germanium precursors to the lineup.
![]() |
| 1. With new SiGe films requiring new processes, germane is not the optimal match, and better suited germanium precursors are required. |
One of the new precursors is monomethylgermaniumchloride, (CH3)GeCl3, for the high-temperature growth of a SiGe template as a substrate for strained silicon. The molecule is activated at temperatures above 750°C. Figure 2 shows the SIMS results of an epitaxial film that was grown at 1000°C under atmospheric pressure using (CH3)GeCl3 and TCS. Compared with GeH4, we observed a sharper interface and smoother grading slope. Depositions on reactor walls are greatly reduced, most likely as a result of the increased presence of chlorine. Less wall deposition leads to less maintenance and lower costs. Pile-up and dislocation densities were slightly reduced compared with films grown by an optimized process using GeH4.
Another new precursor is isobutylgermane, (H9C4)GeH3, for low-temperature growth of pure germanium films or germanium-containing films. Isobutylgermane is liquid and much less toxic than germane. Figure 3 shows the SIMS results of a film grown at 500°C under reduced pressure. Oxygen and carbon concentrations, although high at the interface between substrate and epi film, fall back to their detection limit in the film.
As integration of more materials into silicon-based electronic devices progresses, germanium will gain more importance as a pivotal element. Its main function will be bandgap and strain engineering in order to provide the required functionality. Chemical vapor deposition (CVD) will be used for the manufacture of the various types of films that are needed. Every combination of a new material with silicon will require an optimized process that may only be available with a specialized germanium precursor.
| Author Information |
| Egbert Woelk is director of marketing for the CVD precursor business at
Rohm and Haas Electronic Materials . He has spent more than 15 years in CVD tool and precursor sales for compound semiconductors and other electronic films. He has a Ph.D. in physics. E-mail: ewoelk@rohmhaas.com |
| Deo Shenai is a technical manager at Metalorganics, Rohm and Haas Electronic Materials. He is responsible for directing the R&D and intellectual property for MOCVD and ALD precursors, processes and technologies. He has a Ph.D. in organometallic chemistry, and an MBA in finance. E-mail: dshenai@rohmhaas.com |
| Peter Storck is senior manager of the Advanced Products Group at Siltronic AG . He has a Ph.D. in physical chemistry from the Technical University Darmstadt. E-mail: peter.storck@siltronic.com |
| Isabelle Sagnes is a research engineer in semiconductor III-V compounds in
the LPN-CNRS laboratory . The main fields of her research are MOCVD growth of vertical cavity systems and of quantum dots on GaAs and InP substrates. She has a Ph.D. in physics. E-mail: isabelle.sagnes@lpn.cnrs.fr |
| References |
|


