Abstract
The application of surface science spectroscopies has made an invaluable contribution to our understanding of metal-semiconductor interfaces. They have clearly shown that interfaces formed by the deposition of metals on atomically clean surfaces are rarely abrupt and ordered on an atomic scale, even when the substrate is at room temperature during deposition. These techniques have been used to probe Schottky barriers in conjunction with conventional ones, such as current-voltage (l- V), capacitance-voltage (C- V), and photoresponse, and various attempts have been made to relate barrier heights (tP b ) to the microscopic chemical reactions at the interface. So far these have met with only limited success.
The most thoroughly studied metal-semiconductor system is NiSi2 on Si and here it has been shown that the Schottky barrier height is dependent on the precise bonding at the interface. The NiSi2 layer is epitaxial and ordered and the barrier height may vary by 0.13 e V depending on how the interface is processed. For most metal-semiconductor systems the interface is neither abrupt nor ordered and one might again expect significant variations in tPb' with slightly different processing procedures.
Currently there is a great deal of controversy regarding the most appropriate model to describe Schottky barrier formation at metal-semiconductor interfaces. States generated by a change in semiconductor structure at the interface, by metal induced gap states (MIGS), and states due to defects in the semiconductor generated by the metal have been proposed as primarily responsible for Fermi level pinning. Evidence for or against these models, and others, is generally derived from experimental results of metals on clean cleaved III-V semiconductors, particularly GaAs and InP, and it is thus essential to closely scrutinize the available experimental data for these materials. For the metal-GaAs system one school presents strong evidence that the defect model is most important and that the Fermi level pinning is dominated by a defect donor level 0.9 ± 0.1 e V below the conduction band minimum (Ec ) and a defect acceptor level 0.7 ± 0.1 e V below Ec. This conclusion is derived from photoemission data which indicate that, for coverages of the order ofa monolayer of many metals on clean cleaved (110) GaAs, the Fermi level on n andp-type material is pinned at the acceptor and donor energies, respectively. Furthermore it is reported that a metal coverage less than one atomic layer is sufficient to cause the barrier to form in full. However, measurements of Schottky barrier heights by the conventional 1-V, C-V, and photoresponse techniques do not provide conclusive evidence for this two level mode1. In addition there is now some question as to whether or not the values of tPb appropriate for thick contacts are fully generated when the metal is only a fraction of an atomic layer thick. Likewise a two level model based on defects has been proposed for metal-InP interfaces despite the fact that tPb in this case is known to be highly sensitive to the precise way the contact is prepared and shows large variations for different metals. As a result of this it has been suggested that several processes are important in determining the Schottky barrier heights at metal-lnP interfaces and indeed at metal-semiconductor interfaces in general.
In view of the importance of knowing accurate values of Schottky barriers on GaAs and InP crystals we have examined these systems further both by surface science techniques and by conventional transport techniques. In this paper, therefore, we discuss the interaction of three metals, namely Mn, AI, and Ag with clean and oxidized GaAs and InP. Mn, AI, and Ag were chosen because with these semiconductors they represent highly reactive, moderately reactive, and unreactive metals, respectively. In these studies we have concentrated in particular on probing the reproducibility of Schottky barrier heights and whether or not this may be related to chemical reactivities at the interfaces.
The most thoroughly studied metal-semiconductor system is NiSi2 on Si and here it has been shown that the Schottky barrier height is dependent on the precise bonding at the interface. The NiSi2 layer is epitaxial and ordered and the barrier height may vary by 0.13 e V depending on how the interface is processed. For most metal-semiconductor systems the interface is neither abrupt nor ordered and one might again expect significant variations in tPb' with slightly different processing procedures.
Currently there is a great deal of controversy regarding the most appropriate model to describe Schottky barrier formation at metal-semiconductor interfaces. States generated by a change in semiconductor structure at the interface, by metal induced gap states (MIGS), and states due to defects in the semiconductor generated by the metal have been proposed as primarily responsible for Fermi level pinning. Evidence for or against these models, and others, is generally derived from experimental results of metals on clean cleaved III-V semiconductors, particularly GaAs and InP, and it is thus essential to closely scrutinize the available experimental data for these materials. For the metal-GaAs system one school presents strong evidence that the defect model is most important and that the Fermi level pinning is dominated by a defect donor level 0.9 ± 0.1 e V below the conduction band minimum (Ec ) and a defect acceptor level 0.7 ± 0.1 e V below Ec. This conclusion is derived from photoemission data which indicate that, for coverages of the order ofa monolayer of many metals on clean cleaved (110) GaAs, the Fermi level on n andp-type material is pinned at the acceptor and donor energies, respectively. Furthermore it is reported that a metal coverage less than one atomic layer is sufficient to cause the barrier to form in full. However, measurements of Schottky barrier heights by the conventional 1-V, C-V, and photoresponse techniques do not provide conclusive evidence for this two level mode1. In addition there is now some question as to whether or not the values of tPb appropriate for thick contacts are fully generated when the metal is only a fraction of an atomic layer thick. Likewise a two level model based on defects has been proposed for metal-InP interfaces despite the fact that tPb in this case is known to be highly sensitive to the precise way the contact is prepared and shows large variations for different metals. As a result of this it has been suggested that several processes are important in determining the Schottky barrier heights at metal-lnP interfaces and indeed at metal-semiconductor interfaces in general.
In view of the importance of knowing accurate values of Schottky barriers on GaAs and InP crystals we have examined these systems further both by surface science techniques and by conventional transport techniques. In this paper, therefore, we discuss the interaction of three metals, namely Mn, AI, and Ag with clean and oxidized GaAs and InP. Mn, AI, and Ag were chosen because with these semiconductors they represent highly reactive, moderately reactive, and unreactive metals, respectively. In these studies we have concentrated in particular on probing the reproducibility of Schottky barrier heights and whether or not this may be related to chemical reactivities at the interfaces.
Original language | English |
---|---|
Pages (from-to) | 966-973 |
Number of pages | 8 |
Journal | Journal of Vacuum Science & Technology B |
Volume | 4 |
Issue number | 4 |
Early online date | 10 Mar 1986 |
DOIs | |
Publication status | Published - 01 Jul 1986 |
Externally published | Yes |