Germanium CMOS potential from material and process perspectives: Be more positive about germanium
Akira Toriumi,Tomonori Nishimura
DOI: https://doi.org/10.7567/jjap.57.010101
IF: 1.5
2017-10-30
Japanese Journal of Applied Physics
Abstract:CMOS miniaturization is now approaching the sub-10 nm level, and further downscaling is expected. This size scaling will end sooner or later, however, because the typical size is approaching the atomic distance level in crystalline Si. In addition, it is said that electron transport in FETs is ballistic or nearly ballistic, which means that the injection velocity at the virtual source is a physical parameter relevant for estimating the driving current. Channel-materials with higher carrier mobility than Si are nonetheless needed, and the carrier mobility in the channels is a parameter important with regard to increasing the injection velocity. Although the density of states in the channel has not been discussed often, it too is relevant for estimating the channel current. Both the mobility and the density of states are in principle related to the effective mass of the carrier. From this device physics viewpoint, we expect germanium (Ge) CMOS to be promising for scaling beyond the Si CMOS limit because the bulk mobility values of electrons and holes in Ge are much higher than those of electrons and holes in Si, and the electron effective mass in Ge is not much less than that in III–V compounds. There is a debate that Ge should be used for p-MOSFETs and III–V compounds for n-MOSFETs, but considering that the variability or nonuniformity of the FET performance in today’s CMOS LSIs is a big challenge, it seems that much more attention should be paid to the simplicity of the material design and of the processing steps. Nevertheless, Ge faces a number of challenges even in case that only the FET level is concerned. One of the big problems with Ge CMOS technology has been its poor performance in n-MOSFETs. While the hole mobility in p-FETs has been improved, the electron mobility in the inversion layer of Ge FETs remains a serious concern. If this is due to the inherent properties of Ge, only p-MOSFETs might be used for device applications. To make Ge CMOS devices practically viable, we need to understand why electron mobility is severely degraded in the inversion layer in Ge n-channel MOSFETs and to find out how it can be increased. In the Si CMOS technology, the SiO2/Si interface has long been investigated and cannot be ignored even after the introduction of high-k gate stack technology. In that sense, the GeO2/Ge interface should be intensively studied to make the best of Ge’s advantages. Therefore we first discuss the GeO2/Ge interface with regard to its physical and electrical characteristics. When we regard Ge as a channel material beyond Si for high performance ULSIs, we also have to seriously consider the gate stack scalability and reliability. The source/drain engineering, as well as the gate stack formation, is another challenge in Ge MOSFET design. Both the higher metal/Ge contact resistance and the larger p/n junction leakage current may be the consequences of Ge’s intrinsic properties because they are derived from the strong Fermi-level pinning and the narrow energy band gap, respectively. Even if the carrier transport in the channel may be ideally ballistic, these properties should degrade FET properties. The narrower energy band gap of Ge is often addressed, but the higher dielectric constant of Ge is rarely discussed. This is also the case for most of the other high-mobility materials. The dielectric constant is directly and negatively related to short-channel effects, and we have not been able to provide a substantial solution to overcome this hardship. We have to keep this in mind for the short-channel FET operation. Although a number of problems remain to be solved, in this paper, we view the current status of Ge FET technology positively. A number of (but not all) Ge-related challenges have been overcome in the past 10 years, which seems to be a good time to summarize the status of Ge technology, particularly materials engineering aspects rather than device integration issues. Since we cannot cover all of the results published to date, we mainly discuss fundamental aspects based on our experimental results. Remaining challenges are also addressed but not comprehensively. Integration issues are not discussed in this review. Finally, new types of electron devices utilizing Ge’s advantages are briefly introduced on the basis of our experimental results.
physics, applied