Clean GaAs Research Articles

Electrical study of Schottky barriers on atomically clean GaAs(110) surfaces

We report here a systematic study of the electrical properties of a large number of metal/n-type GaAs (Cr, Mn, Sn, Ni, Al, Pd, Cu, Ag, Au) diodes. Diodes were fabricated on cleaved GaAs(110) surfaces under ultrahigh-vacuum conditions with in situ metal deposition. Using current-voltage (I-V) and capacitance-voltage (C-V) measuring techniques, we were able to obtain very reliable and consistent determinations of the barrier height ${\ensuremath{\varphi}}_{b}$ and ideality factor n. All of the metal-semiconductor systems formed on lightly doped (5\ifmmode\times\else\texttimes\fi{}${10}^{16}$/${\mathrm{cm}}^{3}$) n-type GaAs substrates were characterized by near-unity (1.05) ideality factors. A decrease in the effective I-V barrier height, an increase in the ideality factor in forward bias and a strong voltage dependence on the thermionic emission currents in reverse bias were found for diodes formed on the more heavily doped samples.These changes are essentially metal independent, but depend strongly on the doping of the substrate. The characterization (and elimination in some cases) of peripheral leakage currents from the thermionic emission current is found to be essential in obtaining consistent results in our work and in reinterpreting some of the prior work in the literature. The dominant leakage current flows through a small area, low barrier at the periphery of the device and can be eliminated by mesa etching. The consistent and reproducible barrier-height determinations reported in this study, when combined with the results of recent surface-sensitive studies, are a particularly critical test of models of Schottky-barrier formation.The barrier heights measured from the electrical characteristics of thick-metal-film diodes were found to be essentially identical to those reported during the initial stages (submonolayer to several monolayers of metal) of Schottky-barrier formation by photoemission spectroscopy. This agreement indicates that the physical mechanism responsible for Fermi-level pinning in the thick-metal Schottky diodes is first established at submonolayer to several-monolayer coverages of adatoms and an atomic scale model is therefore necessary to account for the available experimental data. No strong correlation between the barrier heights and the work function of the metal or chemistry at the interface was found. Also, diodes formed on clean GaAs(110) surfaces were found to have essentially identical barrier heights to those formed on clean GaAs(100) surfaces and on contaminated (i.e., chemically prepared) GaAs(100) and GaAs(110) surfaces. Several currently popular models concerned with the physical mechanism responsible for the formation of the Schottky barrier are discussed, and the unified defect model is found to be most consistent with the experimental data.

Reactions of Pd on (100) and (110) GaAs surfaces

The reactions of Pd on atomically clean or air-exposed (100) and (110) GaAs surfaces at temperatures between 20 to 500 °C in different ambients were investigated by transmission electron microscopy. Interfacial reactions quite different from previous x-ray results were observed and two new Pd-Ga-As ternary phases were identified for the first time. At lower temperatures (T≲250 °C) the formation of a ternary phase PdGa∼0.3As∼0.2, which has a hexagonal structure very similar to that of Pd2Ge or Pd2Si with a0=b0=0.672 nm and c0=0.340 nm, was observed. This ternary phase is epitaxially oriented with (12̄0)ternary∥(100)GaAs and [001]ternary∥[011]GaAs on (100) GaAs substrates, and with (11̄0)ternary∥(110)GaAs and [001]ternary∥[11̄0]GaAs on (110) GaAs substrates. At temperatures between 350 and 500 °C only one phase, PdGa, was observed to form in a high vacuum environment, whereas in a forming gas ambient, either a mixture of PdAs2 and another ternary phase PdGa∼0.6As∼0.4 (at 350 °C) or a mixture of PdAs2 and PdGa (at 500 °C) was observed. The ternary phase PdGa∼0.6As∼0.4 is also hexagonal in structure with a0=b0=0.947 nm and c0=0.374 nm. The PdGa phase formed at high temperature is epitaxially oriented on (100) substrates with (110) PdGa∥(100)GaAs and [1̄11]PdGa∥[011]GaAs, but is randomly oriented on (110) substrates. All these observations indicate that the Pd-GaAs reactions at T≳350 °C are very sensitive to the ambient conditions but not as sensitive to the GaAs surface cleanliness or substrate orientation. Correlation of these structural observations to ultraviolet and x-ray photoelectron spectroscopy data obtained from the same reacted interfaces are also discussed.

Influence of S and Se on the Schottky-barrier height and interface chemistry of Au contacts to GaAs

The electrical properties and interface chemistry of Au Schottky-barrier contacts formed on n-type GaAs (100) surfaces that had prior exposure (at room temperature) to elemental S or Se are compared to those of Au contacts to clean surfaces (ideal contacts). The interface Fermi-level energy EiF and chemistry during contact formation were investigated by x-ray photoemission spectroscopy (XPS). The Schottky-barrier height φB was measured by current-voltage and capacitance-voltage methods on the same interfaces characterized by XPS. The exposure of clean GaAs to elemental S or Se results in formation of a ∼5 Å surface layer of Ga and As chalcogenides via a limited chemical reaction. The Schottky-barrier height for Au contacts to the S or Se exposed surfaces had a 0.2 eV variation (0.8–1.0 eV) centered about the φB=0.89 eV value for the ideal Au contact. Changes in EiF over a ∼0.4 eV range occur in response to changes in interface chemistry during surface treatment sequences that include S and Se exposure and Au deposition.

Cr on GaAs (110): The effect of electronegativity on the Schottky barrier height

Cr deposited in sequential steps on atomically clean GaAs(110) surfaces shows strong chemical reactivity at the interface with Cr‐arsenide formation. Soft x‐ray and ultraviolet photoemission spectroscopies have been employed to elucidate the room temperature chemistry and the Fermi level pinning behavior of this system. It was recently reported that the Pd/n‐GaAs(110) system has a Schottky barrier height of ∼0.9 eV, is highly reactive, and is characterized by similar adatom arsenide formation. The distinction here is that Cr/n‐GaAs(110) has a barrier height of ∼0.7 eV. Thus the Schottky barrier height appears to be independent of the nature of the reaction. An examination of the trends in the electronegativity (Pauling’s scale) shows a significant correlation to the Fermi level pinning. We found that low electronegativity (∼1.6) metals like Cr pin at the 0.7 eV level (below the conduction band minimum) which normally acts as an acceptor. The Fermi level of higher electronegativity (∼2.0) materials, e.g., Pd, Cu, Ag, and Au, is observed to pin at the 0.9 eV level which normally acts as a donor. This is exactly what one expects for charge transfer between the semiconductor and the metal using electronegativity as a basis. Further confirmation is found in the examination of changes in the ionization potential threshold of the Cr/GaAs system with metal deposition.

Optically enhanced low temperature oxygen chemisorption on GaAs(110)

Optically enhanced oxidation of GaAs is promising both for insight into fundamental mechanisms of oxygen chemisorption and for possible application in GaAs direct write. For atomically clean GaAs(110), we have found that sample temperature is a key parameter in determining the increase of oxygen uptake under low-intensity argon ion laser illumination (1 W/cm2 at 2.41 eV) relative to the ‘‘dark’’ uptake. Although cooling to 185 K suppresses the dark oxygen adsorption rate, the oxygen uptake with optical irradiation is the same at both 185 and 400 K. The ratio of laser-assisted to dark uptake for 108 L exposure then approaches the enhancement necessary for spatially selective oxidation, showing an increase from 2.5 at room temperature to well over 10 at 185 K. After ruling out sample heating and direct oxygen gas excitation, we suggest that the light-induced increase in uptake is due to electronic excitations in the GaAs. To explain the enhancement, one also needs to assume a weakly adsorbed precursor state that converts to the chemisorption state when thermally or optically activated; the implications for previously suggested oxidation mechanisms will be discussed.

Microscopic effects at GaAs/Ge(100) molecular-beam-epitaxy interfaces: Synchrotron-radiation photoemission study.

We have measured the evolution of abrupt GaAs/Ge(100) interfaces to study the relationship between the valence-band discontinuity (\ensuremath{\Delta}${E}_{v}$) and growth properties on an atomic scale. We used soft x-ray photoemission, low-energy electron diffraction, and Auger electron spectroscopy. The interface was obtained by evaporating Ge (at ${10}^{\mathrm{\ensuremath{-}}9}$ Torr and intentionally coevaporating at an overpressure of ${\mathrm{As}}_{4}$) in situ onto a molecular-beam-epitaxy- (MBE-) grown GaAs sample under epitaxial growth conditions. The MBE-grown GaAs(100) provided different surface reconstructions with controllable starting anion-to-cation ratio. We present a new core-intensity-analysis method for determining thin-film growth independent of the adlayer thickness calibration. For all interfaces, a monolayer of segregated As was observed at the free Ge surface. This suggests that the As-stabilized Ge surface phase formation plays an important role in determining the interface growth. The value we obtain for \ensuremath{\Delta}${E}_{v}$ is (0.47\ifmmode\pm\else\textpm\fi{}0.05) eV, independent of both the initial clean GaAs(100) surface properties and the evolution of the Fermi level. We propose that the driving force for MBE interface formation always yields a unique, low-energy equilibrium structure at the materials transition region, although its extent may vary by as much as a couple of atomic layers. However, the observed constancy of \ensuremath{\Delta}${E}_{v}$ suggests that the contribution of any perturbation caused by variations in the local microscopic properties of the interface (e.g., dipoles) is very small compared to the intrinsic (perhaps bulk) mechanism that determines \ensuremath{\Delta}${E}_{v}$.

An Investigation Of The Cause Of Initial Band Bending Of A Cleaved Clean n-GaAs(llO) Surface

ABSTRACTThe cause of the initial band bending on a clean n-GaAs(llO) surface of so-called “poor quality cleavage” has been controversial. To study this problem, a new type of geometric factor multi-metal evaporator has been designed, which enables us to obtain reproducible metal coverages as low as 10−4 monolayers. By following the early stages of band bending induced by the deposition of noble metals Cu, Ag and Au on n-GaAs (110) surfaces of different cleavage quality, it is found that a certain amount of initial band bending of a clean GaAs(llO) surface of poor quality cleavage corresponds to a certain amount of noble metal coverage (usually lower than 0.01 ML). This phenomenon can be explained in the following way. Any initial band bending on a clean n-GaAs(HO) surface is due to surface defect states created by the cleaving. These cleavage induced defects may be of the same nature as the interface states created by deposited noble metal atoms. Our experimental results will be discussed in the framework of the unified defect model of the Schottky barrier formation on III-V semiconductors.

Chemisorption of oxygen at cleaved GaAs(110) surfaces: Photon stimulation and chemisorption states

The adsorption of oxygen on clean GaAs(110) surfaces cleaved in situ was studied by using Auger electron and electron energy loss spectroscopy. The surfaces were exposed to unexcited molecular oxygen in the dark and under simultaneous illumination with light from a xenon arc lamp. The oxygen uptake was quantitatively determined from the AES signals by using a layer model. Without illumination the oxygen uptake tends to saturate below 0.5 and below a full monolayer with the samples held at 120 and 300 K, respectively, while simultaneous exposure to light with a photon energy above the band-gap energy of GaAs and to oxygen molecules results in uptakes well beyond a monolayer ( θ = 1 ML). The photon stimulation of the oxygen adsorption is due to the photocreated electron-hole pairs, and the uptake is to be characterized by the “combined” exposure ptn (oxygen pressure p, exposure time t, electron-hole pair density n). Four different chemisorption states of oxygen were detected. State S is related to cleavage-introduced defects. At 300 K and in the dark, it saturates at pt ≃ 10 5 L, which results in θ ≃ 0.1 ML. It is followed by the main adsorption states T 1 and T 2. State T 1 is an intermediate one at RT but the only one observed at 120 K in the dark. At 300 K and for exposures above 3 × 10 6 L without illumination, it converts to state T 2, which has been studied in the past only. Above an uptake equivalent to one monolayer which can only be obtained during simultaneous exposures to oxygen molecules and photons a further phase was found. Therefore, it is concluded that the initial monolayer of oxygen is chemisorbed at the top GaAs layer while additional uptake of oxygen then results in the formation of an oxide layer.

Chemisorption of oxygen at cleaved GaAs(110) surfaces: Photon stimulation and chemisorption states

Abstract The adsorption of oxygen on clean GaAs(110) surfaces cleaved in situ was studied by using Auger electron and electron energy loss spectroscopy. The surfaces were exposed to unexcited molecular oxygen in the dark and under simultaneous illumination with light from a xenon arc lamp. The oxygen uptake was quantitatively determined from the AES signals by using a layer model. Without illumination the oxygen uptake tends to saturate below 0.5 and below a full monolayer with the samples held at 120 and 300 K, respectively, while simultaneous exposure to light with a photon energy above the band-gap energy of GaAs and to oxygen molecules results in uptakes well beyond a monolayer ( θ = 1 ML ). The photon stimulation of the oxygen adsorption is due to the photocreated electron-hole pairs, and the uptake is to be characterized by the “combined” exposure ptn (oxygen pressure p , exposure time t , electron-hole pair density n ). Four different chemisorption states of oxygen were detected. State S is related to cleavage-introduced defects. At 300 K and in the dark, it saturates at pt ≃ 10 5 L, which results in θ ≃ 0.1 ML. It is followed by the main adsorption states T 1 and T 2 . State T 1 is an intermediate one at RT but the only one observed at 120 K in the dark. At 300 K and for exposures above 3 × 10 6 L without illumination, it converts to state T 2 , which has been studied in the past only. Above an uptake equivalent to one monolayer which can only be obtained during simultaneous exposures to oxygen molecules and photons a further phase was found. Therefore, it is concluded that the initial monolayer of oxygen is chemisorbed at the top GaAs layer while additional uptake of oxygen then results in the formation of an oxide layer.

Preparation of a clean GaAs

We describe a procedure for producing a clean surface of GaAs

Investigation of the surface structure of GaAs(110) by high energy ion channeling

The first application of high energy ion channeling for the structure determination of the atomically clean GaAs(110) surface is reported. It is found that the surface Ga and As atoms have small lateral displacements (≤0.1 Å) from ideal bulklike sites, whereas a normal component of the first layer shear vector as large as 0.7 Å is compatible with the experimental data. The implications of these results with respect to current LEED models are discussed.

Application of high energy ion channeling to GaAs(110), Au-GaAs(110) and Pd-GaAs(110)

The first application of high energy ion channeling to the atomically clean GaAs(110) surface and metal-GaAs interfaces is reported. Questions of sample preparation, background correction and computer simulation are addressed. It is found that the Ga and As atoms at the clean surface are laterally displaced ⩽ 0.1 Å from the ideal bulk-like sites. The implications of this result to current LEED models are discussed. Au overlayers, deposited at room temperature, do not seem to produce lateral displacements of the substrate for coverages below ≈ 5 monolayer (ML). However, ≈ 0.9 ML of the substrate are expanded or contracted upon Au deposition; this process is completed at a coverage of 0.5 ML. Neither an indication of any order in the Au film is found, nor seems a significant (≳ 5%) fraction of Au atoms to occupy substitutional sites. In contrast, room-temperature deposition of Pd disorders the substrate substantially, without threshold coverage, even at very small film thicknesses.

Chemical bonding, adatom-adatom interaction, and replacement reaction of column-3 metals on GaAs(110)

Studies of the electronic states associated with column-3 (Al, Ga, and In) overlayers on cleaved GaAs(110) surfaces by photoemission electron spectroscopy (PES) and low-energy electron diffraction are reported. Deposition of In or Ga leaves the valence-band spectra devoid of structure other than that characteristic of clean GaAs(110). Comparison with theory shows that the models assumed previously in calculations are in error and that a fundamentally new understanding of the metal-semiconductor bonding is needed. The most significant result is that a strong interaction between the column-3 adatoms leads to cluster formation on the surface. It is shown that at submonolayer coverage Ga most likely forms flat (one- or two-atom-thick) clusters on the surface, which are not in registry with the GaAs surface lattice, while In shows a greater tendency toward three-dimensional cluster formation. As a result, the electronic structure of the overlayer is free-electron-like even at submonolayer coverage. As a contrasting example, Sb (a column-5 element), which forms an ordered overlayer, produces quite significant changes in the surface valence band. The room-temperature reaction between Al adsorbate atoms and a GaAs(110) surface is observed by means of soft-x-ray photoemission techniques using core-level spectroscopy. Evidence of two states of Al which are distinct from the bulk Al metal is seen at submonolayer coverages. The sequential appearance of these states suggests that Al, as deposited at room temperature, tends to replace Ga below the surface layer first; but, as monolayer coverage of Al is reached, a large degree of replacement in the top or surface layer may take place. This interpretation is consistent with the finding by Kahn, Duke, and co-workers that (in thermodynamic equilibrium) Al tends to replace Ga below the GaAs(110) surface layer before much replacement in the first layer occurs. It is concluded that the differences between the room-temperature data of various workers in both the type (surface versus subsurface) and extent of the Al-Ga replacement reaction (for a given Al coverage without annealing) is due to the situation that kinetically, at room temperature, nonequilibrium states of the Al adatoms can dominate, and that the relative proportion of Al in nonequilibrium states to Al in the equilibrium state is a sensitive function of experimental conditions. Smearing of the upper-valence-band features (as seen in photoemission data) occurs as a result of the replacement reaction, suggesting that the replacement reaction is too random at room temperature to produce sharp well-defined new valence-band structure. The Al overlayers do not have structure as free-electron-like as those for Ga and In; this is most likely due to a partial replacement of Ga by Al atoms at the surface and a different type and/or size of clusters formed by Al overlayers compared to Ga or In.

The dependence of Al Schottky barrier height on surface conditions of GaAs and AlAs grown by molecular beam epitaxy

A systematic study of the Al Schottky barrier height on clean GaAs(100) and AlAs(100) as a function of initial arsenic surface coverage has been performed. Two new surface phases on AlAs(100) have been observed. Al Schottky barrier height showed a steady decrease with increasing arsenic coverage on GaAs, and showed steady increase with increasing arsenic coverage on AlAs. The results are compared with current models of Schottky barrier formation.

Titanium nitride Schottky-barrier contacts to GaAs

The electrical properties and interface chemistry of ideal titanium nitride Schottky-barrier contacts to GaAs are investigated by using x-ray photoemission spectroscopy (XPS), current-voltage (I-V), and capacitance-voltage (C-V) techniques. The TiN films were grown on clean n-type GaAs (100) surfaces by reactive evaporation of Ti in a NH3 atmosphere within the XPS system. The TiN-GaAs Schottky-barrier height is 0.83 eV. I-V and C-V measurements were made for a sequence of anneal temperatures up to 700 °C; the electrical properties are nearly ideal up to 600 °C. TiN is found to be a relatively inert nonmetallic alternative Schottky-barrier contact to GaAs that is suitable for high-temperature applications.

Chemical deposition of TiO 2 layers on GaAs

TiO 2 layers were produced on clean etched GaAs by a chemical deposition technique in which a GaAs coupon is immersed in a non-aqueous solution of a titanium alkoxide and then removed to allow the alkoxide layer adhering to the surface to hydrolyze to TiO 2. Depth profiles of the films were measured by X-ray photoelectron spectroscopy and the scanning Auger microprobe technique, using both ion beam sputter etching and chemical solution etching. The results indicate that the composition and chemical structure of the films are determined by the amount of water present in the coating solution, coating atmosphere and hydrolysis atmosphere. Films grown in high humidity exhibit much intermixing of semiconductor and oxide components, and oxidation of the semiconductor; films grown in low humidity exhibit neither mixing nor oxidation of GaAs unless they are stored in this atmosphere for a prolonged period.

Band bending on MBE Al-GaAs(001) as studied by ARUPS

MBE-grown n-type GaAs(001) surfaces have been investigated by angle-resolved UV photo-electron spectroscopy (ARUPS). As-rich c(4 × 4) and Ga-rich (4 × 6) reconstructed surfaces were studied before and after aluminium deposition. The band bending was found to be fully developed already on the clean GaAs surface and had the same value for both reconstructions. Deposition of 0.2 monolayer (ML) of Al did not produce any additional shift of the spectra.

Refractory metal contacts to GaAs: Interface chemistry and Schottky-barrier formation

A survey of the interface chemistry of the refractory metals W, Ta, Re, Ir, and Mo during room- temperature Schottky-barrier contact formation to GaAs by using x-ray photoemission spectroscopy (XPS) is reported. The metals were deposited onto clean n-type GaAs (100) surfaces within the XPS system by two methods: Evaporation and plasma sputtering. For each metal a distinct interfacial reaction which produced GaAs dissociation and formation of a new metal arsenide is observed. Refractory metals are in general not chemically inert in contact to GaAs and nonabrupt interfaces ∠10 Å in width are formed. XPS was also used to correlate interface chemistry with measurement of Schottky-barrier height during contact formation. XPS measured barrier heights ranged from 0.9–0.7 eV, in the order W, Ir, Mo, Ta, and Re.

High Schottky barriers on and thermally induced processes at the Au– GaAs(110) interface

New photoemission measurements show higher Schottky barrier heights (≳1.3 eV) on atomically clean GaAs(110) surfaces at a Au coverage of about 25 monolayers. It is suggested that this effect is due to the movement of Au into the semiconductor; at room temperature it creates acceptor states near the valence band maximum (VBM) that cause the Fermi level at the surface (Efs ) to move close to the VBM. We found that heating of the GaAs(110) surface (above 100 °C) covered with a small amount of Au (0.2 monolayer) causes Efs to move back to its original position (on the clean surface before Au deposition). The heating process is found to greatly inhibit the formation of large barrier heights due to the removal of defect states from the surface region.

Low-energy electron energy-loss spectroscopy with Ge:GaAs (110) heterostructures

Electronic properties and chemical composition of Ge films grown by molecular beam epitaxy on GaAs (110) surfaces were studied in situ by electron energy-loss spectroscopy. The loss peaks involving core-level excitations proved that As atoms segregate at the surface of the growing film. The well known 20 eV loss peak of the clean GaAs (110) surface, being attributed to transitions from Ga(3d) to Frenkel-type excitons of dangling-bond surface states, was found to persist, slightly shifted to 19.7 eV, with the growing Ge (110) film. Since the Ga coverage amounts to below approximately 0.05 of a monolayer this transition seems to contain a strong intraatom contribution.