P. H. Mahowald

The advanced unified defect model for Schottky barrier formation

The advanced unified defect model (AUDM) for GaAs proposed in this paper can be looked upon as a refinement of the unified defect model (UDM) proposed in 1979 to explain Fermi level pinning on 3–5 compounds due to metals or nonmetals. The refinement lies in identifying the defect producing pinning at 0.75 and 0.5 eV above the valence band maximum as the AsGaantisite. Since the AsGaantisite is a double donor, a minority compensating acceptor is necessary. This is tentatively identified as the GaAsantisite. The concentration of As excess or deficiency due to processing or reactions at interfaces is particularly emphasized in this model. A wide range of experimental data is discussed in terms of this model and found to be in agreement with it. This includes the original data on which the UDM was based as well as more recent data including Fermi level pinning on the free-GaAs(100) molecular-beam epitaxy surface, Schottky barrier height for thick (∼ 1000 A) Ga films on GaAs, and the LaB6Schottky barrier height on GaAs(including thermal annealing effects). Of particular importance is the ability of this model to explain the changes in Schottky barrier height for Al and Au on GaAs due to thermal annealing and to relate these changes to interfacial chemistry.

Metallic and atomic approximations at the Schottky barrier interfaces

Stimulated by the work of Zur, McGill, and Smith (ZMS), we have undertaken a broad examination of the ‘‘potential normalization’’ conditions at the metal–semiconductor interface. We have made calculations using a metallic approximation and the model of Bardeen. Strong overall agreement is found with the results of ZMS; however, the details depend on the specific parameters used in the calculations. In particular, parameters were identified which give different pinning positions on n‐ and p‐type semiconductor materials. Experimentally, Cs is found to be in strong disagreement with any ‘‘metallic’’ approximation at the interface; rather, it is shown that the experimental and theoretical understanding of Cs on solids indicates that an atomic model for the last atomic layer of Cs (with some fixed charge due to the induced Cs atomic dipole at the interface) is a more appropriate approximation than the metallic approximation used by ZMS and ourselves. Experimental results for thick noble metals (Au, Cu, and Ag)...

Comparative uptake kinetics of N2O and O2 chemisorption on GaAs(110)

Using surface sensitive photoemission we have measured oxygen uptake versus gas exposure to N2O and O2 on cleaved GaAs(110) surfaces in the coverage range from 0 to 1.5 monolayers, focusing on the reaction variables of surface temperature and visible light illumination. As would be expected given the greater ease of dissociation for N2O compared with O2 (bond strengths of 1.7 and 5.1 eV, respectively), we found that N2O reacts much more readily than O2 at room temperature in the dark, with sticking coefficients at 0.4 ML of 3×10−7 and 7×10−10, respectively. This observation is also supported by a truer measure of reactivity, the chemisorption activation energy, which at 0.15±0.08 eV for N2O is only one-third of the value 0.47±0.02 eV for O2 when corrected for the molecular physisorption energies. These results confirm the fact that the breakup of oxygen molecules physisorbed on the GaAs surface is a major rate-limiting step in the chemisorption reaction. The similarity which we have observed between the kinetics of N2O and optically enhanced O2 uptake furthermore suggests strongly that it is that step—the breakup of O2—which is assisted by visible light.

Chemical and electrical properties at the annealed Ti/GaAs(110) interface

We present the results of a photoemission study of the annealed Ti/GaAs(110) interface, using soft x‐ray (SXPS), x‐ray (XPS), and ultraviolet (UPS) photoelectron spectroscopies to monitor the substrate and overlayer core levels and the valence band for Ti thicknesses between 6.7 and 20 monolayers and annealing temperatures between 200 and 425 °C. The SXPS data imply the formation of a stable Ti–As region near the Ti/GaAs interface that increases in width with temperature. Concomitant to this, the binding energy of the Ga changes from its room‐temperature position towards elemental Ga. The XPS data suggest the presence of elemental Ga at or near the overlayer surface. The UPS data suggest the possibility of cluster formation on the overlayer surface. In situ current–voltage (I–V) measurements were also made on thick‐film systems fabricated using similar methods to those used in the photoemission studies [i.e., cleaving, depositing metal (∼1000 A), and annealing in UHV]. The movement of Ti–As reaction produ...

Aluminum Schottky barrier formation on arsenic capped and heat cleaned MBE GaAs(100)

The fabrication and characterization of an aluminum Schottky barrier on an arsenic capped and heat cleaned GaAs(100) surface is described. The GaAs surface was prepared by molecular beam epitaxy (MBE), and covered with an arsenic cap for protection during transfer through air. The arsenic cap was sublimed by heating, and its removal was monitored by photoemission spectroscopy utilizing synchrotron radiation. After the formation of an aluminum Schottky barrier on this surface, the interfacial position of the Fermi level was measured by photoemission spectroscopy, and the barrier height was measured by I–V and C–V techniques.

Chemical and electronic properties of the Pt/GaAs(110) interface

The interaction between evaporated Pt and the GaAs (110) surface has been studied at room temperature in ultrahigh vacuum utilizing surface‐sensitive soft x‐ray and ultraviolet photoemission spectroscopies. Detailed curve‐fitting analysis of the substrate and metal core levels indicates dissociation of the GaAs with a reaction between the Pt and the Ga and a strong reaction with the As as well. Evidence of Pt–As compound formation is observed. Above 6.67 monolayers (ML) (10 A), a second reacted component dominates and the As‐core energy shifts smoothly to a final position ∼0.3‐eV‐higher binding energy than As in GaAs. The reacted Ga has a well‐defined energy ∼0.4 eV less bound than the Ga in GaAs. This correlates well with the calculated core level binding energy shift reported by Nogami et al. for dilute Ga in Pt. The Schottky barrier height is established by 0.67 ML and is observed to be 0.83 (0.48)±0.1 eV for n‐type (p‐type) substrates, in agreement with electrical measurements on much thicker diodes. ...

Photoenhancement mechanism for oxygen chemisorption on GaAs(110) using visible light

Visible light from an argon ion laser (514.5 nm, 3 W/cm2) is seen to increase oxygen chemisorption on cleaved GaAs(110) surfaces up to a final coverage between one and two monolayers. Using photoemission spectroscopy to measure the oxygen coverage after simultaneous exposure of the surface to oxygen and light, we have determined that oxygen uptake for photoenhanced exposures is independent of sample temperature and doping type. In addition, significantly less enhancement is seen for weakly bound oxidizing molecules (N2O) relative to the effects with molecular oxygen. These results are explained by a photoenhancement mechanism in which energy is released in a surface recombination event, possibly in the form of nonthermal phonons, causing physisorbed gas molecules to dissociate and thereby overcoming a major rate limiting step of the reaction in the dark. This reaction mechanism is supported by calculations of the surface recombination rates and free carrier densities at the surface which show that only the recombination rate is correlated with enhanced oxygen uptake. Other mechanisms and experimental data are also discussed.

The Si/GaAs(110) heterojunction discontinuity: Amorphous versus crystalline overlayers

We have studied both the electrical and structural properties of the Si/GaAs(110) interface in an attempt to understand the factors affecting the valence‐band discontinuity, ΔEv, and to gain insight into the fundamental properties of the amorphous silicon overlayers. In order to differentiate between effects due to intrinsic material properties and those due to extrinsic defects, we grew the interfaces at three different temperatures and performed postgrowth anneals in ultrahigh vacuum. Such growths resulted in overlayers with varied crystalline perfection. In situ ultraviolet and x‐ray photoemission spectroscopies, low‐energy electron diffraction, and surface extended x‐ray absorption fine‐structure measurements were obtained for both amorphous and crystalline overlayers. We found a valence‐band offset of 0.23±0.1 eV for both amorphous and crystalline Si on GaAs provided that the amorphous Si had either been annealed to or grown at sufficiently high temperatures, i.e., 375 and 300 °C, respectively. For t...

Chemical reaction and Schottky barrier formation at the Ti/InP(110) and Sn/InP(110) interfaces: Reactive versus nonreactive cases

We present a soft x‐ray photoemission study of the in situ grown Ti/InP(110) and Sn/InP(110) interfaces. We find that the Ti overlayer deposited onto the cleaved InP(110) surface readily reacts even at submonolayer coverages. A Ti–In alloy (In dispersed in the Ti overlayer) of changing composition and two Ti–P phases are formed. While In outdiffuses at room temperature (RT) through more than 40 A of Ti, the P‐containing reaction phases are more confined to the interface. In contrast, the Sn overlayers do not react with the InP(110) surface and the overlayer is growing in the Stranski–Krastanow fashion with at least one uniform Sn layer. Independent of large differences in reactivity, the surface Fermi level pinning energies are for both interfaces studied similar (0.3 eV for Ti and 0.4 eV for Sn). This indicates, in agreement with our previous studies of InP interfaces, that the RT pinning level is insensitive to the microscopic chemistry and calls for a simple (in the first approximation) mechanism of th...

Valence‐band discontinuity at the Ge/InP(110) interface

The valence band of the Ge overlayer is 0.65±0.1 eV higher than that of the InP(110) substrate when the interface is grown at 280 °C. The In 4d core level was curve fit to obtain a shift of −0.3 eV for surface atoms compared to the bulk at the lowest coverages and to separate the chemically reacted component at higher coverages to obtain an accurate band‐bending determination. The interface is slightly intermixed and after the first two laminar layers, islands of Ge may be forming. Also, islands of In are formed at the surface of the overlayer. The interface was grown at 20 °C also, where less indium is intermixed in the overlayer, and more is surface segregated. The bulk component can still be obtained for the core levels, but the metallic indium emission obscures the valence‐band maximum and a valence‐band discontinuity value is not available for 20 °C.