The microstructural changes in oxygen implanted silicon-on-insulator material (SIMOX) at intermediate annealing steps and the changes by rapid thermal annealing (RTA) were studied with transmission electron microscopy. Defects found in as-implanted SIMOX, including multiply faulted defects, short stacking faults, and {113} defects, were all removed in anneals from 900°C to 1100°C. The threading dislocations in annealed material form at these temperatures during thermal ramping. RTA shows that the microstructure is significantly influenced by the ramp rate. The very high ramp rate in RTA results in very flat interfaces and a buried oxide layer with no Si islands, but significantly increases the defect density. Overall, the results show that intermediate thermal processing steps strongly affect the final microstructure of SIMOX material.

One of the most promising silicon on insulator (SOI) fabrication techniques under development is Separation by IMplanted OXygen (SIMOX) substrates. The objective of this paper is to evaluate the feasability of employing spectroscopic ellipsometry (SE) for the determination of SIMOX substrate quality. Defect density measurements obtained from planar view transmission electron microscopy (TEM) studies are presented and concur with the Raman experimental results. The need for the development of a more sophisticated ellipsometric optical response model to mirror the complexity of the annealed Si overlayer microstructure is demonstrated.

Agglomeration has occurred during zone-melting to obtain single-crystal germanium layers because of insufficient wettability between germanium and silicon-dioxide. The agglomeration could be suppressed by controlling the capping layer thickness, germanium layer thickness, and the island shape. A rotation of crystal orientation was found in the stripe shape island, which was presumably caused by the stress relieving within an island.

We have investigated the effect of the presence of oxide particles on the surface of silicon wafers during high energy, high dose implantation of oxygen into silicon. It was found that for single implants with doses of 1.5 × 1018/cm2 or 1.8 × 1018/cm2, such particles produce a non-continuous buried oxide layer in the as-implanted condition as well as after annealing. Etching results showed that no defects, which formed etchable paths through the buried oxide, were produced for particles with diameters 0.43 um or below for the lower dose and 0.53 um for the higher dose.

The equilibrium configuration produced when the SIMOX process is carried to the extreme of complete conversion of the silicon film to oxide is described. The thickness of the “equilibrium oxide” produced is modelled using a joined Gaussian approximation to the oxygen implant profile. The “equilibrium oxide” for a 190KeV implant is investigated. This material is found to have a thickness of approximately 5400 angstroms, and to be a very homogeneous film with essentially stoichiometric SiO2 composition. The implications of this work for use of oxides in patterned SIMOX are discussed.

A two dimensional numerical simulation of incandescent lamp Zone-Melting Recrystallization (ZMR) was performed. A parametric study examined the thermal effects of lamp intensity, susceptor temperature, and ambient reflectivity, on the melt zone. The melt zone was found to vary linearly with lamp intensity and parabolically with susceptor temperature and ambient reflectivity.

Implants of B, P, and As in Si lead to dislocation formation after 900°c annealing if a critical amount of implant damage is exceeded. However, it is possible to implant higher doses without forming dislocations if the dose is implanted in several sub-critical steps. Annealing between each step removes the (sub-critical) implant damage and dislocations do not form. Such avoidance of dislocation formation is demonstrated for 80 keV implants of B and MeV implants of B, P, and As.

This paper will show that annealing of Si implanted with moderate doses of 725 keV B results in the formation of secondary defects, the so-called category I dislocations. Surprisingly, 12C, with roughly the same mass as 11B, behaves in a very different way. Carbon implant damage does not result in dislocation formation even for damage levels > 100 times higher than that required for B implants. C is also able to avoid dislocation formation of co-implanted B ions. The C-dose needed to avoid dislocation formation for the B implant increases nonlinearly with B-dose. Special C-related secondary defects remain after annealing if the C-dose is higher than 4×1015 /cm2.

High-Resolution X-Ray Diffraction is investigated as a technique for analysis of low-dose, high-energy implanted (001) silicon. The choice of the proper Bragg reflection for the rocking curve measurements is showed to be of crucial importance. The (026)1 reflection appears to be the most suitable for strain caused by implantation damage. By a qualitative analysis of rocking curves taken on B and P implanted Si samples, it could be established that the minimum dose, for which HR-XRD is sensitive enough, is about 1.5.1014cm−2 and 5.1013cm−2 for P and B, respectively. The necessary minimum peak temperature that was determined for a complete damage anneal with the T-RTA of P implanted Si with energies ranging from 0.5 to 1.5 MeV, was 1400 K for all doses of P considered. For B the required optimum peak temperature is about 1375 K.

Severe oxidation inhibited epitaxy when buried Sb profiles in single crystal silicon were formed from evaporated layers irradiated in atmosphere with a pulsed Q-switched ruby laser. Oxygen concentrations as high as 5×1017atoms/cm2 (equivalent to 105nm SiO2) were measured. However, structures prepared without the Sb layer and irradiated under identical conditions, showed no oxidation. Oxidation of Sb as a source of the measured oxygen was ruled out, while the total heating time during laser irradiation is so short (nano- to milliseconds) that normal oxidation kinetics cannot account for the amount of SiO2 measured. Irradiations in vacuum and in a helium ambient showed that the oxygen responsible for these effects is supplied from the ambient in which irradiations are carried out. Also no oxidation was observed when structures, prepared on a substrate heated to 350°C, were irradiated in atmosphere. A model to account for these oxidation effects is proposed.

We have performed a comprehensive experimental study of profiles of arsenic implanted into (100) silicon for a wide range of energies, doses, tilt angles, and rotation angles. Critical angles for channeling of arsenic ions in single-crystal silicon have been calculated and are found to agree well with experimental results. The <100> axial channels and the {110} planar channels are found to be primary sources of channeling. The optimal tilt and rotation angles which minimize channeling and maximize uniformity across a wafer are deduced.

Over 250 boron and over 250 fluorine profiles have been obtained from BF2+ implants over a wide range of implant energies, doses, tilt angles, and rotation angles. A detailed study has been conducted on the boron and fluorine profile variations with the tilt and rotation angles over the available range of energies and doses. Channeling through a few low index axial and planar channels in (100) silicon has been found to account for the observed profile variations with implant angle. Tilt and rotation angle combinations which minimize channeling and ensure process uniformity have been deduced.

The effect of channeling on the diffusion of ion implanted phosphorus in silicon has been investigated. Silicon samples, implanted with 25–100 keV P along the [100] channeling and the random equivalent directions, were subjected to thermal annealing over a temperature range of 600–1050 °C. Secondary Ion Mass Spectrometry (SIMS) and Spreading Resistance Probe (SRP) have been used to determine the atomic and carrier concentration depth profiles, respectively. The findings show that after annealing, the P profiles by implantation along the random equivalent direction can be kept shallower than the profiles obtained by implantation along the [100] channeling direction. Through proper annealing and electrical activation, only minimal diffusion in the tail region of the profiles occurred. For 50 keV P at 1×1015 at./cm2, changing the implantation from the [100] to the random equivalent direction leads to a reduction in the profile depth of about 50% (at 1×1017at./cm3). After 10 seconds of rapid thermal annealing (RTA) at 1050 °C, the profile depth remains more than 30% shallower than the channeled profile.

We have developed a new computationally-efficient two-dimensional model for boron implantation into single-crystal silicon. This new model is based on the dual Pearson semi-empirical implant depth profile model [1] and the UT-MARLOWE Monte Carlo boron ion implantation model [2]. This new model can predict with very high computatational efficiency two-dimensional as-implanted boron profiles as a function of energy, dose, tilt angle, rotation angle, masking edge orientation, and masking edge thickness.

Band tailing produced by electron irradiation of n-type GaAs at helium temperatures is studied. It has been observed that conductance in the band tailing occurs via localized states as evidenced by the observation of a conductance having a frequency dependence of the form σ⊥ωs where ≈ 1.

Sufficient degree of damage induces variable range hopping, σ ∝ exp (-b/Tl/4) in a defect band. The frequency dependence of the conductivity in this band is very weak.

Bombardment damage produced by Si+ ions in AlxGa1−xAs/GaAs layer structures has been studied using transmission electron microscopy and ion channeling and backscattering spectrometry. The damage resistance of A1xGa1−xAs alloy layers increases with Al concentration. In particular, by comparison of complementary Si+ ion doses yielding similar nuclear displacement densities at 150keV and 2MeV bombardment energies, it is demonstrated for the first time that the local concentration of implanted Si impurity is likely to be a significant factor in controlling lattice damage build-up, especially for the highest Si+ ion implantation doses. It is also shown that, in a manner analogous to A1As, the alloy layers can confer a significant protection from ion damage upon adjacent, epitaxially-bonded narrow zones of crystalline GaAs.

Ion beam mixing and damage production at GaAs-AlAs interfaces was studied by Rutherford backscattering and channeling methods. It was observed that the general features of the intermixing of GaAs with AlAs at 100K are typical of that in other semiconductor and metallic systems but that the damage production is not. The GaAs layers amorphize at a very low ion dose whereas the AlAs layers are very resistant to amorphization. Damage in the AlAs begins at one interface of the GaAs and grows through the AlAs layer, but damage at the other interface never nucleates. The ratio of nuclear to electronic stopping influences the growth of the damage zone.

Photoluminescence (PL) measurements have been carried out as a function of temperature and of excitation power on heavily magnesium (Mg) implanted ultra-pure indium phosphide (InP) with a dose concentration, [Mg], ranging from l×lO15cm-3 to 3×1020cm−3. Several undefined emissions have been observed; a strong emission, Z, emerges near the band-acceptor emission for [Mg]= 3×1020cm−3 and two novel emissions, Y1 and Y3 appear below the energy of donor-acceptor pair emission for [Mg] smaller than 3×1018cm−3. In order to obtain more explanation on such emission, temperature and excitation power dependence of the emissions were investigated. Temperature dependent PL experiments revealed that Z is composed of three emissions named ZA, ZBand ZC. Measured activation energies suggest Z represents a pair type transition associated with Mg acceptors. While, Y1 and Y3 emissions are attributed to donor-acceptor type transition because of a significant blue-shift with increasing excitation power.

Solid phase epitaxial regrowth of Sb implanted strained Si1−x Gex alloy layers is reported. Two sets of Si1–xGex alloys with compositions of x=0.08 and x=0.18, MBE grown on (100)Si substrates, were implanted at room temperature with Sb− ions at energies of 200 and 100 keV, respectively, and a dose of 1015cm−2. These alloys were heat-treated in a rapid thermal annealing system at temperatures of 525, 550 and 575°C for durations between 5 and 600 sec. The study of the solid phase epitaxial regrowth was performed by Rutherford backscattering in the channeling mode. The measurements show a significant difference in the regrowth mechanism between the two alloys. For the Si0.92Ge0.00 alloy a fast regrowth process (faster than for Sb implanted Si or Si implanted SiGe layers) occured with an activation energy of 2.92±0.2eV. For the Si0.02Ge0.10 alloy the regrowth took place in two steps: a) a very fast initial process over a short distance, b) a regrowth process of the majority of the amorphous layer.

Sputter etching normally removes material with an overall composition equal to the target composition. At room temperature, this steady-state removal process is usually quickly established by the formation of an altered surface layer which compensates for preferential sputtering, together with an altered subsurface layer which reflects the existence of bombardment-induced Gibbsian segregation (GS). At elevated temperature, ordinary GS, diffusion and the presence of compound phases may strongly affect the development of the steady-state condition. In this paper, we describe two cases in which diffusion and GS become significant processes during sputter etching of a compound material AxB1−x on an underlying substrate of B atoms. In the first case, diffusion is dominated by the substrate species. This case is demonstrated by the sputtering of TiSi2 on (100) Si by 300 eV Ar+. At room temperature, normal sputtering of the TiSi2 layer occurs. At elevated temperature (500–700°C), however, Si diffuses from the substrate to the surface, where it undergoes GS. This allows the development of a steady-state highly selective sputtering of Si atoms, accompanied by movement of the TiSi2 layer into the substrate. In the second case, diffusion is dominated by the non-substrate species. This case is demonstrated by CoSi2, for which the development of steady-state etching occurs in two distinct stages which are strongly temperature-dependent. In this material, Co is the dominant diffusing species, and it is unclear whether or not Si GS occurs. What is clear, however, is that Ar+ bombardment at 500–600°C leads to partial decomposition of the CoSi2 layer and diffusion of Co towards the substrate Si, where new CoSi2 is formed. The combination of sputter etching and elevated temperature is shown to be a controllable environment for compound phase formation at a buried interface.