Library

Notes: Deep-level bulk and interface defect states in InP metal/insulator/semiconductor diodes have been investigated using capacitance-voltage measurements and deep-level transient spectroscopy. The InP surfaces were cleaned with an ultraviolet light/ozone/HF process followed by an optional polysulfide passivation and then capped with a layer of silicon nitride produced by remote plasma-enhanced chemical-vapor deposition. Polysulfide treatment reduced the bulk trap concentrations produced during contact annealing, reducing a bulk trap at 0.33 eV below the measurement limits and reducing a 0.45 eV trap to a concentration of 5×1013 cm−3. The density of interface states close to the conduction-band edge also decreased with the polysulfide treatment.

Notes: Diamondlike carbon films were deposited on germanium crystals with a mass-separated C+ ion beam in ultrahigh vacuum over the energy range 20–275 eV, and the interfaces were characterized with x-ray photoelectron spectroscopy. It was found that ion bombardment induced a carbide phase on the germanium surface. Further carbon accumulation then led to the growth of an amorphous carbon overlayer. The carbide phase was identified by a rather low C 1s binding energy (at about 283.8 eV) and small positive shift of the Ge 3p peak (about 0.4 eV). The valence-band spectra of these samples also suggested that germanium carbide formed with a pure carbon beam for the bombardment energy range considered has a band gap between germanium and diamondlike carbon.

Notes: The surface defect structures on diamond (100) surfaces induced by 500 eV neon ion bombardment and by subsequent annealing were studied in situ with x-ray absorption near-edge structure (XANES) spectroscopy using 250–800 eV synchrotron radiation and with low energy electron diffraction. Ex situ x-ray photoemission spectroscopy (XPS) was also used to characterize the defective layer. Significant changes in the XANES spectra were identified for the defects induced by ion bombardment and subsequent annealing. The diamond discrete exciton absorption at 289.0 eV was clearly suppressed even at the lowest ion fluence used in this study, i.e., 3×1014/cm2, and no such exciton could be observed at 7×1014/cm2. However, the changes in the multi-maxima shape-resonance absorption structure in the range of 290–310 eV indicated that a loss of the diamond long range order required a fluence of 1×1015/cm2. The structural changes were also manifested by the transformation of gap state absorption typical of clean 2×1 surfaces to the π* absorption typical of amorphous carbon. XPS showed that the defective layer was about 2 nm thick. For all samples prepared with the bombardment conditions in the study, both the XANES and XPS data also indicated no phase transformation from defective layers to graphite even after annealing to a temperature of 1100 °C. © 1994 American Institute of Physics.

Notes: Metal-insulator-semiconductor capacitors were fabricated on cleaved n-GaAs (110) facets using remote plasma-deposited silicon nitride as gate insulators. The interface properties of the capacitors made on this surface were analyzed by capacitance-voltage (C-V) measurements. X-ray photoemission spectroscopy was also used to investigate the chemical structure of the interface. Prior to the insulator deposition, the cleaved facets were processed with different surface treatments including HF etch of native oxide, passivation with an ammonium sulfide solution, passivation with hydrogen polysulfide, and passivation with a silicon interface control layer. It was found that while the passivation procedures with the sulfur compounds did improve the C-V data when compared with the HF oxide etch, the silicon interface control layer technique led to the best C-V results. By comparing the quasistatic and high-frequency (1 MHz) C-V data, it was found that the minimum interface state density of the fabricated capacitors was about 1012 eV−1 cm−2. © 1994 American Institute of Physics.

Notes: Argon and carbon ion bombardment of p-diamond at 500–5000 eV in ultrahigh vacuum were studied by in situ x-ray photoelectron spectroscopy (XPS) and low energy electron diffraction analysis. Both argon and carbon ion bombardment at room temperature in the present energy range created a defective surface layer. The radiation damage was manifested by the introduction of a distinct C 1s peak (referred to as the "defect'' peak later) with a binding energy about 1 eV less than that of the bulklike diamond peak, and by the introduction of some additional filled states (referred to as the "filled states'') near the valence band edge of diamond. It was found that in comparison to argon bombardment, carbon bombardment was more efficient in producing the filled states but less efficient in raising the C 1s defect peak. While the filled states disappeared by annealing at about 500 °C, the C 1s defect peak did not change much even with a 1000 °C anneal. These results suggest that the C 1s defect peak, which has also been observed on reconstructed diamond surfaces after hydrogen desorption [see, e.g., B. B. Pate, Surf. Sci. 165, 83(1986)], is associated with vacancy formation and aggregation which give some "internal surfaces'' with a behavior like a reconstructed atomically clean diamond surface.The filled states introduced by ion bombardment are associated with interstitials or interstitial clusters. The amount of residual defects was found to increase with both an increasing bombardment dose and energy. For an argon bombardment at 1000 eV to a dose of 5×1014/cm2, the defective layer was estimated to be about 1.5 nm. Further, it was found that the radiation damage, particularly the "vacancy defects'', could only be annealed (at 1000 °C) when the dose was below 5×1014/cm2 at a bombardment energy of 500 eV. XPS band bending analyses also showed that room temperature bombardment induced a small reduction (0.2 eV) of the surface Fermi level position (EFs) on the p-diamond. However, subsequent vacuum annealing caused a rather large increase of EFs. But the EFs data from about 20 bombarded and annealed samples were always less than 2.2 eV. Thus the formation of an n-type diamond was not observed.

Notes: Effects of 35Cl+ bombardment of Si (100) surfaces were studied with a mass-separated low energy ion beam system operated under ultrahigh vacuum and with in situ x-ray photoelectron spectroscopy (XPS). Before ion bombardment, each silicon sample was etched with a hydrofluoric acid solution and rinsed in deionized water. XPS showed that the surface received no ion bombardment was effectively hydrogen passivated and had no silicon with an oxidation number higher than 2. However, oxygen was found which was probably present in the form of Si—OH or adsorbed water. Chlorine ion bombardment at room temperature initially drove the surface oxygen to the formation of silicon oxide which was subsequently etched off by further ion bombardment. The surface oxidation was initiated by the formation of Si—Cl bonds followed by the thermodynamically favorable replacement of Si—Cl with Si—O. The removal of the surface oxygen depended critically on the bombardment energy. At a bombardment energy of 1±0.6 eV, oxygen bonding was still observed even with a dose of 1018/cm2. The critical dose for oxygen depletion was found to be about 1×1018/cm2 for 40 eV bombardment, and about 1×1016/cm2 for 100 eV bombardment. Physical sputtering was the main oxide removal mechanism. Once the surface oxygen atoms were consumed, the silicon surface was etched by the chlorine ions with an enhancement by the chlorine chemistry. However, the absence of Si—Clx (x(approximately-greater-than)1) was clearly shown by XPS. Hence, bombardment must have promoted the desorption of the surface silicon chlorides.

Notes: A study of chemical etching and chemo-mechanical polishing of Hg0.8Cd0.2Te (MCT) with bromine-methanol has been carried out. It was found that the etch rate could be controlled down to 0.1 nm/s when 0.001% of bromine-methanol was used. Surface analysis using x-ray photoelectron spectroscopy indicated that differential etching of the constituents and accumulation of elemental tellurium occurred even when only a few monolayers were etched from a stoichiometric MCT surface. The relative etch rates were determined to be Cd(very-much-greater-than) Hg(approximately-greater-than)Te. Nevertheless, it was found that chemo-mechanical polishing could produce smooth surfaces with no significant accumulation of elemental tellurium. The production of such surfaces, however, required a balance of chemical etching and mechanical lapping. Furthermore, quick quenching of chemical etching was also required after chemo-mechanical polishing in order to prevent further surface degradation.

Notes: Raman spectroscopy has been used to characterize anodic films grown on n-Hg0.8Cd0.2Te from alkaline media. The spectral results, when compared with those from several reference chemicals, provide unambiguous identification of the anodic film as a 1:1 mixture of amorphous cadmium and mercury tellurites. Since the anodic film is transparent to visible light, Raman spectroscopy was also used to probe nondestructively the substrate surface under the film. The results indicate that the substrate surface was depleted in cadmium.

Notes: Argon incorporation in Si(100) by low energy ion bombardment has been studied by polar angle dependent x-ray photoelectron spectroscopy and Rutherford backscattering spectroscopy. The bombardment was performed at 15, 20, and 100 eV in an ultrahigh vacuum chamber where a mass-separated argon ion beam with an energy spread of less than 1 eV was directed to the target. Both the argon penetration depth and incorporation probability were found to increase with bombardment energy. With a fluence of 2×1017/cm2, most of the incorporated argon was located within 20 A(ring) of the target surface for the 100 eV bombardment and within 10 A(ring) for the 15 eV bombardment. In all cases, the argon depth distribution reached a maximum and then declined. At this fluence, the incorporation probabilities were 0.0015 and 0.0004 for the 100 and 15 eV bombardment, respectively. When the amount of incorporated argon was measured as a function of fluence, it increased with fluence at low fluences, reached a quasisaturation at about 1×1016/cm2, but became fluence dependent again above 1×1018/cm2. The retained argon was stable at room temperature but showed at least two stages of thermal desorption in the temperature range 25–500 °C.

Notes: The hysteresis in metal-nitride-silicon (MNS) capacitors can be reduced to nearly zero by using nitrogen-rich silicon nitride as the gate dielectric and treating the silicon substrate by an ammonia plasma before dielectric deposition. However, the ammonia plasma treatment step also causes an increase in interface state density, especially in the middle of the silicon band gap. Without the ammonia plasma treatment, the virgin flat-band voltage V*FB is always negative. With the ammonia plasma treatment, V*FB can be shifted from a negative value to zero for MNS capacitors on n-type silicon, whereas V*FB will be shifted from a negative value to a more negative value for MNS capacitors on p-type silicon. These effects can be explained by postulating that the interface states generated by the ammonia plasma treatment step are amphoteric defects similar to Pb centers at the oxide/silicon interface in metal-oxide-silicon capacitors.