Sapphire can form the bedrock for the growth of relatively thick, single crystal epitaxial films of rhombohedral SiGe. This material has much higher carrier mobility than single crystal silicon, and could spur the development of ultra-fast chipsets, say Yeonjoo Park and Sang Choi from NASA Langley Research Center.

Band-gap engineering is one of the most important guidelines to design new semiconductor alloys and devices. For the last 60 years, two important semiconductor alloy engineering models were developed for band-gap engineering; the cubic crystal structure model shown in Figure 1-(a) and hexagonal crystal structure model shown in Figure 1-(b). The first cubic crystal model includes group IV semiconductors (Si,Ge,C) of diamond structure, and group III-V (GaAs, InP, etc.) and group II-VI (ZnSe, CdTe, etc.) semiconductors of cubic zinc-blende structure. The second hexagonal crystal model includes III-Nitride semiconductors (GaN, AlN, InGaN, etc.) and hexagonal SiC semiconductors.

Our team at NASA Langley Research Center proposed and proved that a third alloy engineering model - “rhombohedral-trigonal crystal model” - can be established between the cubic one and hexagonal one. A simple diagram of rhombohedral-trigonal crystal model is shown in Figure 1-(c). In details of the new model, a general cubic crystal is not only a special case of tetragonal crystal but also a special case of a rhombohedral crystal with inter-planar angle of 90°. When a cubic crystal is strained in the [111] direction, it becomes a rhombohedron. Thus, a cubic crystal belongs to a rhombohedral crystal group.

Additional mathematical transformation equation transforms any rhombohedral crystal into a trigonal crystal in hexagonal frame. Therefore, cubic crystals belong to a more general trigonal crystal group and the epitaxy between cubic crystals and trigonal crystals can be established not as an accidental coincidence-latticematching but as a fundamental crystal symmetry relation. Figure 1-(d) shows such an example of rhombohedrally aligned cubic SiGe on trigonal c-plane sapphire. The problem of this epitaxy is that two crystal structures which are twin to each other can be formed as shown in Figure 1-(e), the top view. This twin defect was a major problem in the rhombohedral epitaxy and has hindered further applications so far.

However, we found that optimized growth under new Xray diffraction (XRD) characterization can eliminate twin defect and form single crystalline rhombohedral SiGe layer on c-plane sapphire in one of the crystal alignment of Figure 1-(e). This is because threefold symmetry of a trigonal crystal prefers one rhombohedrally aligned cubic crystal to the other. Thus, a symmetry breaking occurs between two cubic crystals that are rotated by 60° from each other, i.e. one cubic crystal becomes dominant and the other cubic crystal diminishes.



Figure 1. Crystal structure of (a) cubic zinc-blende or diamond structure, (b) hexagonal Wurtzite structure, (c) crystal symmetry group relation, (d) Rhombohedrally aligned SiGe on c-plane Sapphire, (e) two possible alignments of SiGe on trigonal c-plane Sapphire, twin to each other

 



Figure 2. Innovative patented XRD methods characterize integral density and spatial distribution of twin defects

 

The discovery of super-hetero-epitaxy growth technology for rhombohedral single crystalline SiGe on c-plane sapphire was confirmed by the NASA-invented new XRD methods: (1) Total defect density measurement and (2) Spatial wafer mapping method (see further reading, patents pending 2, 4, and 5). Figure 2 shows the characterization results by XRD. Innovative XRD wafer mapping method shows spatial distribution of major single crystalline SiGe (99.8%) (Left image) and twin defect SiGe (0.1%) (Right image) that exist on the same sapphire wafer clipped by three plastic jaws outside. Twin crystal defect is reduced to below 0.1% and it exists only at the edge of a wafer. The successful development of rhombohedrally aligned SiGe has also led the inventors to construct a new hybrid bandgap engineering diagram with transformed lattice constants. In conventional cubic bandgap engineering diagram, the distance of [100] vector, i.e. lattice constant of a cube is used as coincidence lattice distance. On the other hand, conventional hexagonal bandgap engineering diagram uses the distance of basal plane basis vector as coincidence lattice distance. Direct comparison of the two lattice constants, one from the cubic crystal and the other from the hexagonal crystal is not very meaningful because it is an apples-to-oranges comparison although they are often plotted in one diagram. To the contrary, in rhombohedral super-hetero-epitaxy, three vectors of cubic crystal that are perpendicular to [111] vector is making coincidence lattice matching with the combination of basal basis vectors of the trigonal crystal. Also, by adding another fact that hexagonal crystals can be epitaxially grown on trigonal crystals such as GaN on Sapphire with in-plane rotation, a combined ab-initio hybrid bandgap engineering diagram was developed and under test now.

Hundreds of new alloys and thousands of new device structures can be fabricated with rhombohedral-trigonal model because this epitaxial scheme is not only limited to SiGe on sapphire but can be extended to other cubic semiconductors on thousands of trigonal crystals in nature. It is expected that not all trigonal crystals can accommodate rhombohedrally aligned cubic crystals, but selected trigonal crystals that have enough difference of formation energies between two cubic crystals rotated by 60° can yield a single crystal epitaxial layer.

To identify other new materials within the rhombohedral super-hetero-epitaxy category, NASA scientists have selected a few candidate materials and they are developing the growth methods. Another benefit of rhombohedral super-hetero-epitaxy in addition to new hybrid crystal structure is that it has unprecedented lattice matching conditions that are different from cubic lattice matching. These new opportunities to create lattice matched and strained semiconductors are under study now. It is also interesting that a cubic semiconductor on trigonal substrate is strained from directions and elongated or compressed along [111] direction so that it deforms into a rhombohedron shape while conventional cubic epitaxy creates tetragonal deformation by strains in direction.

Many trigonal crystals are insulators like sapphire. Therefore, it is also possible to create SiGe on Insulator (SGOI) with a possible lattice matching condition. Tables 1 and 2 show how the key features of currently developing lattice matched SiGe on Insulator (LM-SGOI) under our research compare to existing products or technologies: The far right column of table 1 shows the NASA Langley developed SiGe material that is compatible with the conventional insulator silicon oxide. The compatibility of SiGe with the silicon oxide is a very important factor for wafer-based mass production. Table 2 shows the attainable speed of SiGe chipsets based on the gate length and the charge mobility. Lattice-matched SiGe widely opens a possibility of chipset speed improvement, while the single crystal silicon itself has its own intrinsic limit on speed even by miniaturized feature size. From this table, one can easily imagine the great impact of NASA Langley’s rhombohedral lattice-matched silicon-germanium material on the new generation ultrafast chipset development.



Table 1. Comparison of Si, SOI, SiGe on Si, and LM-SGOI technologies



Table 2. Expected operation speed of a transistor with various mobilities and gate lengths

 

NASA’s rhombohedrally single crystal SiGe is the first of its kind ever achieved in the world. Therefore, there is no competition. The highly anticipated increase in charge mobility of the proposed materials technology is unique for the development of ultra fast chipsets. The latticematched SiGe is also complemented by silicon oxide as an insulator, unlike the arsenide, antimonide, or other compound semiconductors.

A proper insulation material like SiO2 for lattice-matched SiGe allows fabrication of several hundreds of chips on a wafer basis. Compound semiconductors, such as zincblendes and Wurtzites, do not have proper insulators to allow mass production, instead of a single chip.

Another challenge is to incorporate higher germanium content into SiGe layer to raise carrier mobility. For example, the electron mobility of germanium is 4,000 cm2/V·s while that of silicon is only 1,400 cm2/V·s. By providing a suitable substrate for SiGe layer, transistors of higher operation frequencies can be fabricated as shown in Table 2.

Ultra-CMOS

Rhombohedral semiconductors on trigonal substrates can improve the following products: Ultra fast Complementary Metal Oxide Semiconductor (CMOS) chipsets; Heterojunction Bipolar Transistors (HBT); Thermo-Electric (TE) device; photo-voltaic solar cell device; advanced detectors; high frequency high power transmitters; and others as shown in Figure 3.



Figure 3. Applications of rhombohedral semiconductors on trigonal crystals

Before our research, trigonal crystal materials were not considered to be compatible with cubic semiconductors. The hybrid structure of rhombohedrally deformed cubic semiconductors and trigonal crystals create new opportunities to fabricate completely new single-crystal alloy structures for ultrafast semiconductor chip development beyond the silicon-based chip technology.

Since the NASA’s rhombohedral SiGe can allow faster electron motion with higher germanium contents than single crystal silicon has, it will offer the development of ultrafast chipsets for numerous applications. In addition, hexagonal space symmetry materials can be grown on trigonal space symmetry materials such as GaN on cplane sapphire.

We summarize the following inter-crystal-structure epitaxial relation between cubic [111] direction, trigonal [0001] direction, and hexagonal [0001] direction for further possibilities as shown in Figure 4. This diagram shows the possibility of epitaxial growth from underlying substrate material of one space symmetry group to an epitaxial layer of a different space symmetry group. A solid line means that it is possible to form a single crystal layer, and a dashed line means that double position defect creates huge difficulties to form a single crystal. A round green circle indicates that twin detection XRD methods can be applied and an empty circle means that twin detection XRD methods do not work.



Figure 4. Inter-crystal-structure epitaxy possibility relations with applicability of twin detection XRD methods

 

This research has won a R&D100 award in 2009 and it has double-edge impact as the world’s first development of single crystalline rhombohedral SiGe semiconductors on trigonal substrates and opening the first window to hybrid crystal structure alloy engineering, namely, “Rhombohedral Hybrid Band-gap Engineering” with innovative XRD methods.

Further reading

Yeonjoon Park, Sang H. Choi, Glen C. King, and James R. Elliott “Hybrid Bandgap Engineering for Rhombohedral Super-Hetero-Epitaxy”, NASA Case No. LAR 17405-1, June 28, 2006. [patent filed on April 23, 2007]

Yeonjoon Park, Sang H. Choi, Glen C. King, and James R. Elliott, “Supporting DATA (XRD and EBSD) for Hybrid Bandgap Engineering for Rhombohedral Super-Hetero-Epitaxy”, NASA Case No. LAR 17519-1, March 12, 2007. [patent filed on August 25, 2008] Yeonjoon Park, Sang H. Choi, Glen C. King, and James R. Elliott, “New Rhombohedral Alignment of Cubic Semiconductor On Trigonal Substrate At A High Temperature”, NASA Case No. LAR-17553-1, July 10, 2007. [patent filed on September 5, 2008] Yeonjoon Park, Sang H. Choi, and Glen C. King, “Epitaxial Growth of Group IV Semiconductor Alloys on Basal Plane of Trigonal or Hexagonal Crystal”, Invention Disclosure, NASA Case No. LAR 17185-1, June 13, 2005. [patent filed on July 13, 2006]

Yeonjoon Park, Sang H. Choi, Glen C. King, and James R. Elliott

“Supporting XRD DATA for rhombohedrally-grown diamond-structured group-IV alloys on basal plane of trigonal substrate”, NASA Case No. LAR 17392-1, June 28, 2006. [patent filed on September 5, 2008]

Yeonjoon Park, Sang H. Choi, and Glen C. King, “Lattice Matched SiGe Layer on Single Crystalline Sapphire Substrate”, Invention Disclosure, NASA Case No. LAR

16868-1, February 26, 2004. [patent filed on April 23, 2007]

Yeonjoon Park, Sang H. Choi, and Glen C. King, “Silicon Germanium Semiconductive Alloy And Method Of Fabricating Same”, U.S. Patent No. 7,341,883, March 11, 2008 Yeonjoon Park, Sang H. Choi, Glen C. King, James Elliott, Diane M. Stoakley, “Graded Indexed SiGe Layers on Lattice Matched SiGe Layers on Sapphire”, U.S. Patent No. 7,514,726 B2, April 7, 2009

Yeonjoon Park, Glen C. King, and Sang H. Choi, “Highly [111]-oriented SiGe layer on cplane sapphire”, Journal of Crystal Growth, Vol. 310, Issue 11, p2724~2731, May 15, 2008.