Dislocation-free Substrates Research Articles

ВПЛИВ ТЕХНОЛОГІЧНИХ ЧИННИКІВ НА ЯКІСТЬ ПРИЛАДОВИХ СТРУКТУР

n the modern production of semiconductor devices and integrated circuits, epitaxial compositions are widely used: sili-con single-layer epitaxial structures, silicon inverted epitaxial structures and silicon structures with dielectric insulation. An urgent task is a thorough study of the defects of such structures and technological factors that significantly affect their quality at various stages of the manufacturing process.The purpose of the work is to study the dependence of the density of defects in the substrate and the built-up layer of silicon epitaxial compositions on technological factors and to develop a system that has increased resistance to electromigra-tion and at the same time prevents erosion of silicon in the contact windows.Substrates with a thickness of 260 μm with a crystallographic surface orientation made of single crystals of dislocation-free silicon with a resistivity of 10-50 Оm∙m were taken for the study. Defects in the structure were detected by selective etching and investigated using metallographic and scanning electron microscopes. Processing of the working side of the surface was also carried out: chemical-mechanical polishing with removal of a layer 1-2 microns and 20 microns thick; mechanical pol-ishing with diamond paste with a grain size of 1.0 μm and 5 μm. The processing of the non-working side of the substrates was also different: chemical-mechanical polishing, grinding, hetering -grinding with a free abrasive followed by shallow mechan-ical polishing. After growth, the epitaxial compositions were polished and grinded by chemical-mechanical polishing on both sides to a thickness of 80 μm on the substrate side and 170-180 μm on the side of the built-up layer.During research, the substrates were exposed to various technological factors. The obtained experimental results allow us to conclude that to produce semiconductor device structures based on epitaxial compositions with a low density of disloca-tions, it is necessary to use dislocation-free substrates that do not contain bands of A-type microdefects. All other things being equal, the best quality of epitaxial compositions is achieved by applying thorough chemical-mechanical polishing of the sub-strate on its working side and heterogenization on the reverse side.

Nonius approach for Si1-xGex/Si(0 0 1) epitaxy characterization

In Si(001)\\(Si1-xGex 6–8 nm + Si 100–120 nm) × 2 (0.07 ≤ x ≤ 0.21) structures, grown by MBE on dislocation-free substrates and considered as nonius systems, X-ray reflectometry and diffraction on the reflection (004) revealed second period thickness decreasing compared to the first one due to misfit energy increasing. Electron microscopy of the same samples confirms the smaller thickness of the second period. Three-waves diffraction on the forbidden reflection (002) was applied to compare the structural perfection in two azimuthal directions 〈110〉: along the maximum substrate misorientation from the plane (001) and in the perpendicular direction. In the same sample with reduced photoluminescence quantum yield, it was found that the average period value determined by reflectometry (109.5 nm) was larger than that determined by diffraction on the reflection (004) (97.5 nm), and the angular distance between satellites on curves (002) corresponded to two different period values: 109.5 nm and 102 nm. Misfit dislocations are absent in the studied samples. This suggests both possible appearance of noncrystallographic defects in strained epitaxial layers and only effective thickness values corresponding to perfect structure fragments which are determined by X-ray diffraction.

Reduction of grown-in dislocation density in Ge Czochralski-grown from the B 2O 3-partially-covered melt

Germanium (Ge) single crystals with an extremely low density of grown-in dislocations were grown by the Czochralski (CZ) technique with boron oxide (B 2O 3) liquid. Because attachment of particles floating on the melt surface to a growing Ge crystal leads to generation of dislocations during the growth, partial covering of the Ge melt surface with B 2O 3 liquid was attempted. Such attachment of particles was drastically suppressed or the particles were caught by the introduction of B 2O 3 liquid, and a particle-free Ge melt was realized in the central region of the melt surface. Ge single crystals were successfully grown from such melt, the grown-in dislocation density being 0–1×10 3 cm −2, which was remarkably lower than that in Ge crystals grown by a conventional CZ technique. The contaminations by B and O atoms of the grown crystal detected by SIMS analysis were very low. These Ge crystals have the potential for application to be applied as high-quality, dislocation-free substrates of GaAs solar cells for various usages including in space.

Crystallization of GaN by HVPE on pressure grown seeds

Growth of GaN under pressure from solution in gallium results in almost dislocation free plate-like crystals but with size limited to approx. 1 cm (lateral) and 100 μm (thickness). On the other hand, deposition of GaN by HVPE on the pressure grown substrates allows stable crystallization with rates of a few hundreds μm/h. The crystals with a thickness exceeding 2 mm are grown in this way. However, in these thick GaN crystals grown directly on almost dislocation free substrates quite high number of dislocations appears if the crystal thickness exceeds certain critical value probably due to a small lattice mismatch between GaN grown by HVPE and the heavily doped pressure grown substrates Since the critical thickness for defect generation is of the order of 100 μm, almost dislocation free layers (density below 10 4 cm -2 ) thinner than 100 μm were grown. The pressure grown substrates were then removed by mechanical polishing, or conductivity sensitive electrochemical etching (for strongly n-type substrates). Then the HVPE low dislocation density GaN platelets were used as substrates for the growth of a few mm thick bulk GaN crystals.

Device advantage of the dislocation-free pressure grown GaN substrates

The process of GaN crystallization from solutions under high pressure will be described. The formation of point defects and their distribution in the crystals will be discussed. The results of the use of the pressure grown crystals for quantum epitaxial structures and blue laser devices will be used to demonstrate the advantage of dislocation-free substrates.

Native defect related inhomogeneity in characteristics of GaAs field-effect transistors fabricated on annealed dislocation-free substrates

The EL2 distribution showed a clear correlation to both the resistivity and the characteristics of field-effect transistors. It was observed that the nonuniformity of the characteristics of field-effect transistors was not improved by annealing, though those of both EL2 and the resistivity were improved by annealing. It is probable that the nonuniformity is related to the grown-in inhomogeneity in the distribution of native defects other than EL2. The importance of minimizing such an inhomogeneity in the growth process has been confirmed.

Homogeneity qualification of GaAs substrates for large scale integration applications

A novel procedure is used to qualify the homogeneity of GaAs substrates for large scale integration (LSI) circuit applications. This procedure is particularly well adapted for assessing the electrical variations of densely packed field-effect transistors (FET’s) used in GaAs LSI circuits. Threshold voltage nonuniformity due to crystal inhomogeneity and surface effects is measured for both conventional and In-alloyed, dislocation-free, slightly Cr-doped, semi-insulating liquid encapsulated Czochralski substrates. The average value of the threshold voltage standard deviation of about 300 dense rows over a 2-in.-diam wafer (30 FET’s over 300 μm for each row) is 20 mV using dislocation-free substrates which is much smaller than in conventional substrates (50–100 mV). These data show the importance of this qualification test with the highest ever reported resolution of 10 μm suitable for the evaluation of fabrication yield of LSI circuits.

Quantitative Analysis of In Density in Semi-Insulating GaAs by Photoluminescence

Photoluminescence (PL) measurements at 4.2 K were performed on In-doped semi-insulating GaAs (InxGa1-xAs) grown as a nearly dislocation-free substrate for GaAs large-scale integrated circuits. The In composition x of the sample wafers was calibrated by inductively-coupled-plasma emission spectrometry before the PL measurement. A small variation of In composition in the region 0 ≤x ≤0.0139 was detected by the peak-energy shift of exciton-recombination emissions Δh ν (eV). The value of x could be estimated from Δh ν using the equation x=- Δh ν/1.505 without destruction of the sample wafers. It was found that the emission of neutral-acceptor-bounded-exciton recombination observed in all samples was useful for the measurement of the x value.

Punch-through type InGaAs photodetector fabricated by vapor-phase epitaxy

Punch-throughact-Punch-throught ype InGaAs/InPp hotodiodes were made from high purity layers grown on (100) InP substrates by vapor-phase epitaxy. High-speed photoresponse to a mode-locked Nd:YAG laser was observed owing to as low carrier concentration as 1 X 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">15</sup> cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-3</sup> Full width at half maximum and rise time of the pulse response were 126 and 82 ps, respectively, at above 4 V. Dark current was affected by substrate etch pit densities and was 4.1 X 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-4</sup> A/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> at 0.5 V <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">B</sub> for a diode made from the layer on dislocation free substrate. A uniform photocurrent multiplication factor of 5 was obtained.

Effect of dislocation loops in macroscopically dislocation-free GaP substrates on the perfection of homo-epitactic deposits

Substrates and epitactic layers of GaP are etched under optimized conditions with the defect-revealing, preferential R-C etchant and subsequently examined using high-magnification (about 2500X) optical microscopy techniques. It is possible then to distinguish two new phenomena, viz. etch pits characteristic of dislocation loops in the substrate, and inclined dislocation dipoles in the layer. For layers grown on dislocation-free substrates we find that (i) the surface densities of both defects are equal (∼5 × 10 5cm -2), and (ii) the average diameter of the dislocation loops in the substrate is roughly the same as the average distance between the two dislocations of the dislocation dipoles (0.5−1μm). Hence the perfection of these layers is determined by interfacial dislocation loops. Because the density of dislocation loops is only about a factor of two lower in highly (∼1 × 10 5 cm -2) dislocated substrates, growth on these substrates results in layers which have even slightly lower dislocation density than layers grown on dislocation-free substrates. In the former case also single dislocations in the substrate propagate into the layer. Minority-carrier lifetime data indicate that minority-carrier recombination at dislocations is a restrictive factor for the luminescence quality of layers grown both on dislocation-free and highly-dislocated substrates.