H. S. Luftman

PHYSICAL MECHANISMS OF TRANSIENT ENHANCED DOPANT DIFFUSION IN ION-IMPLANTED SILICON

Implanted B and P dopants in Si exhibit transient enhanced diffusion (TED) during annealing which arises from the excess interstitials generated by the implant. In order to study the mechanisms of TED, transmission electron microscopy measurements of implantation damage were combined with B diffusion experiments using doping marker structures grown by molecular-beam epitaxy (MBE). Damage from nonamorphizing Si implants at doses ranging from 5×1012 to 1×1014/cm2 evolves into a distribution of {311} interstitial agglomerates during the initial annealing stages at 670–815u2009°C. The excess interstitial concentration contained in these defects roughly equals the implanted ion dose, an observation that is corroborated by atomistic Monte Carlo simulations of implantation and annealing processes. The injection of interstitials from the damage region involves the dissolution of {311} defects during Ostwald ripening with an activation energy of 3.8±0.2 eV. The excess interstitials drive substitutional B into electric...

The interstitial fraction of diffusivity of common dopants in Si

The relative contributions of interstitials and vacancies to diffusion of a dopant A in silicon are specified by the interstitial fraction of diffusivity, fA. Accurate knowledge of fA is required for predictive simulations of Si processing during which the point defect population is perturbed, such as transient enhanced diffusion. While experimental determination of fA is traditionally based on an underdetermined system of equations, we show here that it is actually possible to derive expressions that give meaningful bounds on fA without any further assumptions but that of local equilibrium. By employing a pair of dopants under the same point-defect perturbance, and by utilizing perturbances very far from equilibrium, we obtain experimentally fSb⩽0.012 and fB⩾0.98 at temperatures of ∼800u2009°C, which are the strictest bounds reported to date. Our results are in agreement with a theoretical expectation that a substitutional dopant in Si should either be a pure vacancy, or a pure interstitial(cy) diffuser.

GaAs surface reconstruction obtained using a dry process

We report attaining Ga‐terminated (4×2) surface reconstruction on virgin GaAs substrates using a completely dry process at temperatures below the oxide sublimation temperature and without group V overpressure. The native oxides are removed with an electron cyclotron resonance hydrogen plasma treatment, followed by annealing at 500u2009°C in ultrahigh vacuum, which yields a reconstructed surface suitable for epitaxial overgrowth. Characterization by secondary ion mass spectroscopy and transmission electron microscopy reveals the complete removal of O, reduced C, and high structural order at the epilayer/substrate interface when this preparation method is used before molecular beam epitaxy. Annealing the substrate at a lower temperature yields a nonreconstructed surface possessing significant impurity concentrations, and leads to dislocation defects at the epilayer/substrate interface.

Surface roughness‐induced artifacts in secondary ion mass spectrometry depth profiling and a simple technique to smooth the surface

We report on secondary ion mass spectrometry (SIMS) depth profile artifacts induced by surface roughness. The formation of a TiSi2 film at 800u2009°C on a boron doping superlattice (DSL) of Si results in a rough (22.0 nm root mean square) interface between the film and Si DSL. After chemically etching off the TiSi2 film, SIMS information is collected while sputtering through the surface of the Si DSL. The resulting depth profiles are irreproducible due to the initial surface roughness. By chemo‐mechanically polishing the Si prior to SIMS analysis, we smooth the surface and the resulting depth profiles are then consistent and easily explained by current diffusion theory.

Removal of GaAs surface contaminants using H2 electron cyclotron resonance plasma treatment followed by Cl2 chemical etching

We report a novel dry process to remove the surface contaminants C, Si, and O from GaAs substrates. This method utilizes an electron cyclotron resonance hydrogen plasma to remove the native oxides, followed by a very brief Cl2 chemical etching of GaAs to further reduce C and Si residues, and a final vacuum anneal. Characterization by secondary ion‐mass spectrometry (SIMS) typically reveals the removal of C, Si, and O at the overgrown/processed interface to the levels below the SIMS detection limit. The as‐processed GaAs surface, a Ga‐stabilized reconstructed (4×6), is atomically smooth, and is as clean as a surface of freshly grown GaAs epilayers.

Hydrogen plasma removal of AlGaAs oxides before molecular beam epitaxy

We report a method for the removal of AlxGa1−xAs native oxides for 0≤x≤1, prior to molecular beam epitaxial overgrowth. The oxides formed on epilayers of AlGaAs after atmospheric exposure are removed in an electron cyclotron resonance hydrogen plasma with a substrate temperature less than 400u2009°C. Reflection high energy electron diffraction indicates the plasma‐prepared AlGaAs surface are oxide‐free and crystalline; after a vacuum anneal to 250–500u2009°C, GaAs or AlGaAs are epitaxially overgrown on these surfaces. Secondary ion mass spectroscopy detects C, O, and Si impurities at the interfaces, where their concentrations increase with increasing Al content of the exposed surface. The quality of the interface and the overgrown film, as observed by transmission electron microscopy, are found to be better for lower interface impurity densities.

Hydrogen plasma processing of GaAs and AlGaAs

Hydrogen plasma processing of GaAs and AlGaAs using an electron cyclotron resonance plasma reactor, which is vacuum‐linked to a molecular‐beam epitaxial (MBE) growth chamber is reported. Native oxide removal and surface cleaning of GaAs is characterized using hydrogen plasma processing, subsequent thermal Cl2 etching, and vacuum annealing. It is shown that surface reconstruction and excellent GaAs/GaAs interfaces can be achieved using these dry vacuum procedures. It is also shown that AlxGa1−xAs native oxides can be removed for 0≤x≤1 using hydrogen plasma processing before MBE overgrowth. The best AlGaAs/AlGaAs interfaces are obtained using low microwave power during hydrogen plasma processing. O and C impurities detected at these interfaces increase with higher Al composition; Si interface impurities tend to increase with higher microwave power. In general, hydrogen plasma processing is judged effective for surface preparation before MBE growth for the complete range of AlGaAs alloys.

Thermal nitridation enhanced diffusion of Sb and Si(100) doping superlattices

Depth dependent nitridation enhanced diffusion (NED) of Sb was investigated by annealing Si(100) doping superlattice (DSL) structures in NH3 at 810–910u2009°C for 15–180 min. These multilayered DSLs consisted of six 10 nm wide Sb doping spikes spaced 100 nm apart. Antimony NED, attributed to vacancy injection, indicated vacancy supersaturation values of 3–5. From the spatial decay of Sb NED, lower bounds for vacancy diffusivities of (7.9±1.4)±10−14, (1.2±0.2)×10−12, and 2.1×10−11 cm2/s were obtained at 810, 860, and 910u2009°C, respectively. Evidence of trap limited vacancy diffusivity was observed.

INTERFACIAL CHARACTERISTICS OF ALGAAS AFTER IN SITU ELECTRON CYCLOTRON RESONANCE PLASMA ETCHING AND MOLECULAR BEAM EPITAXIAL REGROWTH

Regrown/processed AlGaAs interfaces using secondary ion mass spectrometry, cross section transmission electron microscopy (TEM), and reflection high energy electron diffraction have been characterized. Two sets of samples, GaAs/Al0.4Ga0.6As (with GaAs on top) and Al0.4Ga0.6As/GaAs (with Al0.4Ga0.6As on top), are used as starting materials. For the GaAs/Al0.4Ga0.6As samples that are first exposed to atmosphere, the experiment is performed in an integrated processing system where etching and regrowth chambers are linked together by ultrahigh vacuum transfer modules. The etching process includes electron cyclotron resonance (ECR) hydrogen plasma cleaning of GaAs native oxides, ECR SiCl4 plasma anisotropic deep etching into Al0.4Ga0.6As, and an optional, brief Cl2 chemical etching. Regrowth is carried out using solid‐source molecular beam epitaxy (MBE). Despite the in situ processing, significant amounts of C, Si, and O impurities at the 10, 5, and 50×1012 cm−2 levels exist at the interfaces. However, the imp...

Electron‐beam induced current determination of shallow junction depth

Electron‐beam induced current (EBIC) was investigated as a possible method of determining shallow junction depth. Molecular‐beam epitaxy Si n+/p junctions 200–1800 A deep were explored with both cross‐sectional and plan‐view EBIC geometry. Cross‐sectional EBIC analysis proves to be accurate and reproducible in determining the center of the depletion region for the deep junctions (1800 A) when using a conventional scanning electron microscopy (SEM) with a LaB6 filament. Confidence in the location of shallow junctions decreases due to sample drift and the resolution limits of the SEM and EBIC techniques. In the plan‐view geometry, in which the EBIC current is recorded as a function of electron beam energy, we can distinguish between shallow junctions with greater precision than deeper junctions. Collection efficiency versus electron‐beam energy curves reveal junction depth through shifts in both peak position and height. The collection efficiency versus electron‐beam energy curves were modeled assuming the ...