ALBERT

Notes: CuPt type ordering, which consists of a monolayer compositional modulation along one of the 4 〈111〉 directions in the lattice, was studied using transmission electron microscopy for GaAs1−xPx with values of x extending from 0.25 to 0.85. The samples were grown by organometallic vapor phase epitaxy on nominal (001) GaAs substrates that were misoriented by varying amounts in three directions. No CuPt type ordering was observed for GaAs1−xPx with x ≤0.35, while ordering was found to occur for 0.4≤x≤0.85. The direction of substrate misorientation has a major effect on the determination of which of the four possible CuPt variants are formed for 0.4≤x≤0.85. Two variants, with ordering on the (1¯11) and (11¯1) planes, appear for epilayers grown on substrates oriented exactly on the (001) plane and for substrates misoriented by 6° towards the [110] direction. Only one variant, with ordering on the (1¯11) plane, appears for epilayers grown on substrates misoriented by 6° towards [1¯10]. These ordering-induced spots observed in transmission electron diffraction (TED) patterns for GaAsP occur only for the [110] cross section. From TED studies of GaInP grown on similar substrates, we conclude that the CuPt variants in GaAsP are exactly the same as for GaInP. Further evidence supporting this conclusion was obtained by growing first a layer of GaInP followed by a layer of GaAsP. High-resolution dark field electron micrographs show domains of the same variants in both layers. A mechanism describing the formation of the specific ordered variant for both GaAsP and GaInP is proposed. From studies of ordering in a strain-layer superlattice, the strain due to lattice mismatch was found to play no significant role in the propagation of ordered domains. Microtwins, also generated due to lattice mismatch, can act as domain boundaries and prevent the propagation of the ordered domains.

Notes: GaInAs/InP quantum wells have been grown on (001) InP substrates by atmospheric pressure organometallic vapor-phase epitaxy with well widths ranging from a few monolayers to 100 A(ring). The interface quality and the epilayer thickness were examined using x-ray diffraction spectroscopy and transmission electron microscopy. The layers were found to be very uniform with both interfaces apparently free of defects. In addition, for a nominal 8 A(ring) quantum well, a (110) high-resolution cross-sectional lattice image clearly shows the well thickness to be 2–3 monolayers, confirming that the growth rate obtained from thick layers is accurate for very short growth times. These results demonstrate that atmospheric pressure organometallic vapor-phase epitaxy can produce extremely thin GaInAs/InP quantum wells of a few monolayers.

Notes: High-quality Ga0.51In0.49P, lattice-matched to GaAs, has been grown by atmospheric pressure organometallic vapor-phase epitaxy. The growth was performed at a temperature of 680 °C and a growth rate of about 12 μm/h. The indium distribution coefficient was found to be unity at this growth temperature. At a V/III ratio of 148, the Ga0.51In0.49P epilayers had photoluminescence (PL) half-widths of 35 and 7.2 meV at 300 and 10 K, respectively, the best reported results to date. As the V/III ratio was changed from 94 to 240, the 300-K energy band gap measured by PL varied only from 1.897 to 1.912 eV, values close to that of Ga0.51In0.49P grown by liquid-phase epitaxy. An ordered arrangement of gallium and indium atoms on the column III sublattice with a Cu-Pt structure was observed in transmission electron diffraction patterns for all samples. Our results show that the phenomenon of changing energy band gap with varying V/III ratio does not occur at these high growth rates.

Notes: In this study we apply a series of focused and overfocused electron probes with current densities ranging from 105 to 109 A m−2 to irradiate electron-gun deposited thin films of amorphous AlF3 (a-AlF3)and amorphous SiO2(a-SiO2). Statistical distributions of the time deemed necessary to produce a given amount of mass loss from the two beam-irradiated materials are measured as functions of beam current density and sample thickness. According to those results, a-AlF3 is damaged in parallel throughout the irradiated volume of the sample as indicated by no detectable thickness dependence, whereas a-SiO2 displays a distinct scaling of characteristic drilling times with thickness indicative of a combination of surface and volume mass loss processes. © 1999 American Institute of Physics.

Notes: Tantalum-related thin films containing different amounts of nitrogen are sputter deposited at different argon-to-nitrogen flow rate ratios on (100) silicon substrates. Using x-ray diffractometry, transmission electron microscopy, composition and resistivity analyses, and bending-beam stress measurement technique, this work examines the impact of varying the nitrogen flow rate, particularly on the crystal structure, composition, resistivity, and residual intrinsic stress of the deposited Ta2N thin films. With an adequate amount of controlled, reactive nitrogen in the sputtering gas, thin films of the tantalum nitride of nominal formula Ta2N are predominantly amorphous and can exist over a range of nitrogen concentrations slightly deviated from stoichiometry. The single-layered quasi-amorphous Ta2N (a-Ta2N) thin films yield intrinsic compressive stresses in the range 3–5 GPa. In addition, the use of the 40-nm-thick a-Ta2N thin films with different nitrogen atomic concentrations (33% and 36%) and layering designs as diffusion barriers between silicon and copper are also evaluated. When subjected to high-temperature annealing, the single-layered a-Ta2N barrier layers degrade primarily by an amorphous-to-crystalline transition of the barrier layers. Crystallization of the single-layered stoichiometric a-Ta2N (Ta67N33) diffusion barriers occurs at temperatures as low as 450 °C. Doing so allows copper to preferentially penetrate through the grain boundaries or thermal-induced microcracks of the crystallized barriers and react with silicon, sequentially forming {111}-facetted pyramidal Cu3Si precipitates and TaSi2 Overdoping nitrogen into the amorphous matrix can dramatically increase the crystallization temperature to 600 °C. This temperature increase slows down the inward diffusion of copper and delays the formation of both silicides. The nitrogen overdoped Ta2N (Ta64N36) diffusion barriers can thus be significantly enhanced so as to yield a failure temperature 100 °C greater than that of the Ta67N33 diffusion barriers. Moreover, multilayered films, formed by alternately stacking the Ta67N33 and Ta64N36 layers with an optimized bilayer thickness (λ) of 10 nm, can dramatically reduce the intrinsic compressive stress to only 0.7 GPa and undergo high-temperature annealing without crystallization. Therefore, the Ta67N33/Ta64N36 multilayered films exhibit a much better barrier performance than the highly crystallization-resistant Ta64N36 single-layered films. © 2000 American Institute of Physics.

Notes: Single and multiple Ga0.4In0.6P/(Al0.4Ga0.6)0.5In0.5P quantum wells have been grown using atmospheric pressure organometallic vapor phase epitaxy. The Ga0.4In0.6P well layers are coherently strained to match the lattice parameter of the GaAs substrate. Transmission electron microscopic results showed that the quantum-well layers are very uniform in thickness and the interface is abrupt and free of misfit dislocations. The photoluminescence peak energy increases as the well width decreases, due to carrier confinement in the quantum well. Growth interruptions do not change the photoluminescence peak energy of the quantum well. However, the photoluminescence intensity is drastically reduced for longer growth interruption times. Higher-order x-ray diffraction satellite peaks and a narrow photoluminescence halfwidth are observed in a 20-layer multiple-quantum-well sample, indicative of high structural uniformity and precise control of the composition and thickness during the growth process. Considering the effect of strain on the heterojunction band offsets, the photoluminescence peak energy of the strained quantum well can be described by a simple theory as a function of the well width.

Notes: X-ray diffraction and transmission electron microscopy, along with electrical and film stress measurements, were used to evaluate the effectiveness of 40-nm-thick amorphous Ta2N and microcrystalline TaN diffusion barriers, both single and multilayered, against Cu penetration. Failure of the single-layered Ta2N diffusion barrier upon annealing is initialized by crystallization/grain growth, mainly helped by frozen-in compressive stress (3–4 GPa) to transform itself into a columnar structure with a comparable grain size to the thickness of the barrier. However, when subjected to annealing, the Ta2N/TaN alternately layered diffusion barrier with an optimum bilayer thickness (10 nm) remains almost stress-free (0–0.7 GPa) and transforms itself into an equiaxed structure with grain sizes of only ≤3 nm. Such quasisuperlattice films can present lengthening and complex grained structures to effectively block Cu diffusion, thus acting as much more effective barriers than Ta2N (and TaN) single-layered films. © 2000 American Institute of Physics.

Notes: Large area aligned carbon nanotube (CNT) arrays have been successfully synthesized from C2H2 and H2 mixture by rf plasma-enhanced chemical vapor deposition (without hot filament) on iron-coated silicon substrates. H2 plasma (not H2 gas) was confirmed to play the role of reducing iron oxide to metallic iron and promoting the formation of evenly separated particles, as well as being the primary factor in synthesizing aligned CNTs. The addition of H2 gas with no plasma during the growth resulted in randomly oriented CNTs. Meanwhile, without the addition of H2, the C2H2 plasma resulted in the growth of very fine worm-like carbon fibers. Using substrates with a thicker catalyst layer (〉90 nm) reduced the CNT density significantly. © 2001 American Institute of Physics.

Notes: Ga0.5In0.5P is observed to form the CuPt ordered structure during organometallic vapor phase epitaxy (OMVPE). Of the four possible {111} planes on which CuPt ordering could occur, only two are observed for growth on (001)-oriented substrates, giving the (1¯11) and (11¯1) variants. The mechanism by which ordering occurs is not completely understood. Recent total energy calculations indicate that the phenomenon can be explained on the basis of thermodynamic considerations. Indirect evidence indicates that kinetic factors, including processes occurring at steps propagating across the surface in the two-dimensional growth mode, affect ordering. In this letter, Ga0.5In0.5P layers have been grown on (001)GaAs substrates by OMVPE. In order to examine the effects of surface kinetic factors, the substrates were first patterned with [110] oriented grooves 5 μm wide and from 0.2 to 1 μm deep. This yields adjacent areas of epitaxial material within the grooves produced by growth via the motion of steps in opposite directions. Transmission electron diffraction reveals that the two directions of step motion produce two different variants. For exactly (001) oriented substrates, one half of the groove is filled with a single domain of the (1¯11) variant while the other half is also a single domain, but of the (11¯1) variant. For substrates misoriented by 6° to give [110] steps, the domains are asymmetric. The domains are very large, several square microns in cross section extending along the entire length of the groove. The strong intensities of the order-induced spots indicate a high degree of order in the material grown in the grooves. These results demonstrate directly, for the first time, that kinetic factors related to the motion of steps on the surface determine the ordered structure formed. They also demonstrate the possibility of producing very large domains of ordered material.

Notes: GaAs1−xPx with 0.4≤x≤0.85 forms the CuPt ordered structure during organometallic vapor phase epitaxy (OMVPE). Only the (1¯11) and (11¯1) variants are observed for growth on (001)-oriented substrates. The mechanism by which ordering occurs is only now being discovered. Total energy calculations, including the effects of surface reconstruction, indicate that the phenomenon can be explained purely on the basis of energy considerations. Indirect evidence indicates that kinetic factors, including processes occurring at steps propagating across the surface in the two-dimensional growth mode, control ordering. In this work, GaAs1−xPx layers have been grown by OMVPE on (001)-oriented GaAs0.6P0.4"substrates.'' In order to examine the effects of surface kinetic factors, the substrates were first patterned with [110]-oriented grooves 5 μm wide and a fraction of a micron deep. The groove is used to provide a source of steps moving in opposite directions from the two edges. Transmission electron diffraction reveals the formation of large domains of the two variants that meet in the center of the groove. A surprising feature is the presence of a region in the groove with absolutely no ordering. Tracing the surface shape during growth using a superlattice structure indicates that the disordered region is due to growth on {511} facets. The domains formed after the groove is filled are very large, several square microns in cross-sectional area and extending along the entire length of the groove. These results demonstrate that natural ordering in GaAsP, an alloy with mixing on the group V sublattice, can be controlled by regulating the propagation of steps during growth, exactly as for GaInP where mixing is on the group III sublattice.