A Thermodynamic Approach to Selecting Alternative Gate Dielectrics
- PDF / 3,880,302 Bytes
- 7 Pages / 612 x 792 pts (letter) Page_size
- 64 Downloads / 211 Views
A Thermodynamic
Approach to Selecting Alternative Gate Dielectrics
Darrell G. Schlom and Jeffrey H. Haeni Abstract As a first step in the identification of suitable alternative gate dielectrics for metal oxide semiconductor field-effect transistors (MOSFETs), we have used tabulated thermodynamic data to comprehensively assess the thermodynamic stability of binary oxides and nitrides in contact with silicon at temperatures from 300 K to 1600 K. Sufficient data exist to conclude that the vast majority of binary oxides and nitrides are thermodynamically unstable in contact with silicon. The dielectrics that remain are candidate materials for alternative gate dielectrics. Of these remaining candidates, the oxides have a significantly higher dielectric constant () than the nitrides. We then extend this thermodynamic approach to multicomponent oxides comprising the candidate binary oxides. The result is a relatively small number of silicon-compatible gate dielectric materials with values substantially greater than that of SiO2 and optical bandgaps 5 eV. Keywords: bandgap, high-dielectric-constant materials, high- dielectrics, thermodynamics, thermal stability.
Introduction An “immediate grand challenge” for the semiconductor industry is the identification and development of an alternative gate dielectric for future metal oxide semiconductor field-effect transistors (MOSFETs) on silicon.1 This alternative gate dielectric, which needs to be fully implemented in production by 2005 in order for integrated circuits to continue to follow Moore’s law of rapid performance-doubling, needs (1) to have a significantly higher dielectric constant () than amorphous SiO2 ( 3.9), (2) to be stable in contact with silicon (capable of withstanding a 900C annealing step),2 (3) to have a bandgap high enough (4–5 eV) to provide sufficiently low gate leakage, (4) to have a low density of electrically active defects at the oxide/ silicon interface (Dit), and (5) to be the electrical equivalent (with a much lower
198
leakage current) of an SiO2 layer with a physical thickness tox of 10 Å.1,3,4 Many dielectrics are known with 3.9, so at first glance this grand challenge might seem trivial to solve. However, as we describe in this article, there is a significant driving force for most dielectrics to react with silicon—that is, most dielectrics are not thermodynamically stable in contact with silicon. While it is possible for a kinetic barrier to limit the extent of reaction or even prevent a thermodynamically unstable interface from reacting at all, the temperature that the dielectric/silicon interface must withstand in the processing of a MOSFET is high. Annealing temperatures of 1000C are used today to activate the implanted dopants in MOSFETs;3,4 by 2005, annealing temperatures are anticipated to decrease to 900C.2 At such high tempera-
tures, kinetic barriers must be significant to prevent reaction between silicon and a dielectric that is not thermodynamically stable in contact with it. Reactions between the dielectr
Data Loading...
