Oxide-free Silicon Research Articles

Characterization of silicate/Si(001) interfaces

Many of the proposed high permittivity gate dielectrics for silicon-based microelectronics rely on a stack configuration, with an SiO2 buffer layer to provide an interface. We describe a means for creating gate dielectrics with a direct yttrium silicate–silicon interface through the solid-state reaction of yttria and silicon oxynitride, avoiding the preparation of an oxide-free silicon surface. Characterization by medium-energy ion scattering indicates complete consumption of the underlying oxide through silicate formation during high-temperature annealing. Furthermore, the silicate dielectric exhibits small flat-band voltage shifts, indicating low quantities of charge, without passivation steps. Creation of a silicate–silicon interfaces by a simple route may enable the study of an alternate class of dielectrics.

Thin films and nanowires of molecule-based conductors and magnets

Nanowires of tetrathiafulvalene (TTF)[Ni(dmit) 2] 2 (dmit 2−: 2-thioxo-1,3-dithiole-4,5-dithiolato) are obtained by successively dipping SSCC in acetonitrile solutions of (Bu 4N)[Ni(dmit) 2] and (TTF) 3(BF 4) 2, observed by S.E.M. and AFM, and characterized by Raman micro-probe. Thin films of (TTF)[Ni(dmit) 2] 2 are obtained by the same process but using a new original silicon conversion coating (SiCC). The deposit is observed by S.E.M., and characterized by Raman micro-probe and transport measurements. TTF·tetracyanoquino-dimethane (TCNQ) individual nanowires and (TTF)[Ni(dmit) 2] 2 nano-needles scattered on the surface of the substrate or aggregated in wads, are obtained by the dipping process on oxide-free micro-rough silicon (O-f.Mr.Si). Adherent thin films of the V(TCNE) 2 (TCNE: tetracyanoethylene) magnet are deposited on (O-f.Mr.Si) substrates by chemical vapor deposition from bis(benzene)vanadium and TCNE as precursors.

Front-end, single-wafer diffusion processing for advanced 300-mm fabrication lines

Single-wafer, diffusion processing offers the possibility to significantly reduce cycle times, provide ‘greener’ processes and fulfil key technology requirements of ever-reducing thermal budgets and improving critical thin-film interface control. This paper will discuss the cycle time and other potential gains of a transition to 300-mm single-wafer diffusion processing and go on to describe work carried out on such processing for advanced nitride CMOS gate and SiGe applications. Multi-product, multi-process ASIC/SOC 300-mm processing will require aggressive cycle times to ensure early time-to-market, rapid time-to-volume and accelerated development of new technologies and functionality. One of the key cycle time detractors in current multi-process facilities is the batching requirement in diffusion. The advantages of a transition to single-wafer diffusion processing will be presented. A nitrided CMOS gate application has been studied on a 300-mm single-wafer tool featuring a module for RTA/RTO/RTN, a module for mono-Si/poly-Si and SiGe CVD deposition and a cool-down chamber equipped with optical metrology to allow direct control of different steps. Different pre- and post-nitridation steps on oxide or oxide-free silicon surface will be reported that demonstrates the advantages of such an integrated single-wafer tool.

Nanotribology of Clean and Oxidized Silicon Surfaces.

Atomic force microscopy (AFM) has been used for nanotribological studies of silicon surfaces covered by oxide layers of various kinds: chemical oxides prepared by SC 1 (NH4OH/H2O2/H2O) and SC 2 (HCl/H2O2/H2O) treatments, a thermally treated silicon surface and a H-terminated oxide-free silicon surface. Only in the case of SC 1 chemical oxide, scratching of the oxide and ploughing of the silicon by a Si3N4 AFM tip were observed. On the other hand, no wear of the sample was noted on other surfaces. By annealing the SC 1-treated surface in N2 gas above 200oC for 30 min, the oxide layer could not be scratched. By soaking the thermal oxide in KOH, the oxide layer was then scratched. It is suggested that the presence of OH bases is a requirement for the nano-scratching of the oxide layers.

Stability in photoluminescence of porous silicon

The effect of aging on photoluminescence (PL) of porous silicon has been studied for different storage media with a view to find suitable conditions for stabilizing the PL spectra. IR studies have been performed on both initially oxidized and oxide-free porous silicon samples to get an insight of the possible chemical changes in the porous layer after treating it in a number of environments. The changes in the PL spectra, both blue shift and red shift in appropriate environments, are ascribed to growth in Si=O bonds leading to trapped electron states at the Si/SiO 2 interface in conformity with the recently proposed model [M.V. Wolkin et al., Phys. Rev. Lett. 82 (1999) 197]. In addition, a mechanism and a theoretical model for oxidation of silicon nanocrystallites in porous silicon have been proposed to explain the observed experimental results. An excellent agreement between the observed rate of shift of PL peak wavelength with storage time and that calculated from the proposed model has been found. Possibility of stabilizing the PL intensity and peak wavelength by keeping it in a non-oxidizing environment and to tune the PL peak wavelength with different storage media have also been discussed.

Glass Transition Temperature of Poly(tert-butyl methacrylate) Langmuir−Blodgett Film and Spin-Coated Film by X-ray Reflectivity and Ellipsometry

The glass transition temperature (Tg) of poly(tert-butyl methacrylate) (PtBMA) Langmuir−Blodgett (LB) films relative to that of spin-coated films on a silicon substrate is investigated by ellipsometry and X-ray reflectivity. The Tg of spin-coated films on an oxide-free silicon wafer decreases as the film thickness decreases, which is consistent with the majority of the previous reports in the literature. However, the transition temperature of the LB films is nearly independent of the total thickness, that is, the number of accumulated layers, which is much different from the case for spin-coated films. After sufficient thermal annealing of the LB film, it recovers the thermal transition behavior of the spin-coated film with an equivalent thickness. The layer structure of the LB film is deemed to be responsible for the departure from the behavior of the spin-coated film.

Influence of Microscopic Chemical Reactions on the Preparation of an Oxide-Free Silicon Surface in a Fluorine-Based Solution

It is shown that microscopic chemical reactions influence the hydrogen termination of the Si surface in fluorine-based solution. The hydrogen termination reaction was quite sensitive to the surface structure and the composition of the solution. The remaining hydroxyl group on some specific surface sites was suppressed using a neutral pH solution. It was confirmed that the neutral pH solution was suitable for silicidation on a heavily boron-implanted surface and improved the sheet resistance of titanium silicide on the surface.

Fermi level pinning on HF etched silicon surfaces investigated by photoelectron spectroscopy

A widely used approach to obtain smooth oxide-free and (partially) H-terminated silicon (Si) surfaces is to immerse Si wafers into CP4A (a mixture of H2O, HNO3, CH3COOH and HF in a volume ratio of 22:5:3:3) and/or HF solutions of varying concentrations. It is usually assumed that such treatments result in a dramatic reduction of the surface density of states and that, therefore, no surface band bending can occur. In our experiments we investigated the electronic surface structure of a number of CP4A/HF treated n- and p-Si wafers with varying doping densities by x-ray photoelectron spectroscopy (XPS). XPS allows a straightforward detection of surface stoichiometry as well as one of band bending and surface photovoltages (SPV) on semiconductor materials because the positions of the core level peaks directly depend on the position of the Fermi level within the band gap at the surface. Our experiments show that on all surfaces investigated Fermi level pinning still exists after the samples were immersed in the CP4A/HF solutions and that the pinning states are located close to the conduction band. Most of the samples also showed SPV when measured under illumination. The measurements also show that up to 36.6% of the surfaces are covered by F atoms depending on the treatment and the doping density. From the amount of band bending we estimated the density of surface states present on the various samples.

Depth distribution of zinc adsorbed on silicon surfaces out of alkaline aqueous solutions

In the mass production of advanced microelectronic devices (chips), contamination by transition metals is known as a major risk for device yield and reliability. Although zinc belongs to the most frequent contaminants, not many efforts have been made to understand the chemistry of the interactions between this element and silicon surfaces during wet chemical treatments. This paper deals with the interactions of oxide-free and chemically oxidized silicon(100) surfaces with aqueous zinc solutions of pH=9. TXRF (total-reflection X-ray fluorescence spectrometry), XPS (X-ray photoemission spectroscopy) and ISS (ion scattering spectroscopy) were applied in combination with chemical desorption experiments to characterize the silicon surfaces and the concentration and depth distribution of zinc. These studies show that zinc is partially incorporated in the native oxide which is growing under these conditions on an oxide-free silicon surface. In the case of a chemically oxidized surface, only a small part of the adsorbed zinc is incorporated in the oxide in a region close to the oxide surface due to ion exchange reactions during the contamination process.

Nanoscale Study of the Hydrogenated Amorphous Silicon Surface

ABSTRACTA scanning tunneling microscope has been used to study the topology of the surface of device-quality, hydrogenated Amorphous silicon deposited by rf discharge from silane or “hot wire” CVD. The substrates were oxide-free single-crystal silicon or GaAs. Films studied were either grown in our laboratory and observed with no air exposure, or grown at other laboratories producing device-quality photovoltaic cells and viewed after air exposures of less than 30 Minutes. Thin films (10 nm) representing early growth stages appear significantly smoother than the thicker films. The topology of thick films (> 50 nm) has large variations over individual samples. While many regions can be characterized as “rolling hills”, atomically flat areas are sometimes observed nearby in our films. In most regions the observed slopes were 10% or less from the horizontal, but some steep-sided valleys, indicating incipient voids, are seen. Overall surface roughness measured on sub-Micron areas of our films is very inhomogeneous. Uniformity of the films grown off site was much better, although no atomically flat regions were observed, surface roughness can be estimated.

Quantification of Si 3N 4 LPCVD inhibition on oxide surfaces

Si 3N 4 films were produced in an integrated vacuum processing system with separate modules for HF vapour etching, silicon thermal oxidation and LPCVD of Si 3N 4. It appears that the initial nucleation rate of the nitride growth depends strongly on the substrate surface. Si 3N 4 nucleates immediately on HF-vapour etched and thus oxide-free silicon, whereas on silicon oxide an appreciable retardation of growth takes place, which depends on the conditioning of the oxide prior to deposition. Conversion of the (O)N film into (O)NO, the dielectrics of choice in advanced memory capacitors, reveals that the degree of resistance of the nitride film against wet oxidation (“punch-through”) is directly related to the growth inhibition time and hence to the film micro-roughness. Details are presented on the relation between wafer pre-treatment, nitride process conditions, growth inhibition timw and the ultimate wet-oxidation resistance of the nitride film.

Kinetic interpretation ofα- andβ-Si3N4 formation from oxide-free high-purity silicon powder

A kinetic analysis of the isothermal nitridation of high-purity oxide-free silicon powder is described. The kinetic analysis suggests that theα andβ polymorphs of Si3N4 are formed by separate and parallel reaction paths. This analysis provides for the decoupling and quantitative kinetic interpretation ofα- andβ-Si3N4 formation reactions. Consistent with existing microstructural and thermodynamic evidence, theα-forming reaction is shown to obey a first-order rate law, whereas a phase-boundary controlled rate law describes theβ-forming reaction. A kinetic model employing these rate laws is developed and is used to predict theα/β phase ratio as a function of isothermal reaction temperature and extent of reaction. Theα/β phase ratios so obtained are shown to be in good agreement with experimental observations made under a variety of reaction conditions.

Electromigration behavior of aluminum films deposited on silicon by ionized cluster beam and other techniques

Electromigration experiments have been carried out on aluminum thin films which have been deposited on oxide-free (111) silicon by employing the ‘‘ionized cluster beam’’ technique as well as other related deposition methods. It has been found that the same high electromigration resistance and film structure is obtained with or without utilizing a nozzled crucible and with or without applying ionization and acceleration voltages during deposition. The conclusions are supported by electromigration lifetime measurements, scanning electron micrographs, optical micrographs, reflective high-energy electron diffraction, and x-ray diffraction.

Simple method to determine thickness of thin silicon dioxide film

The depth of the rocking mode infrared absorption peak at 450 cm−1 from IR response at 500 cm−1 has been observed to be a linear function of silicon dioxide thickness grown on silicon wafers. At the critical oxide thickness of 100 A°, the depth becomes zero, i.e., the IR response is flat. With further decrease of oxide thickness, the absorption peak changes sign and for oxide-free silicon, the absorption minima is shifted by about 2% from the IR response at 500 cm−1. Such IR response has been successfully used for determining oxide thicknesses in the range of 100–1200 A° with ±50 A° accuracy.

Characterization of low-pressure chemical-vapor-deposited and thermally-grown silicon nitride films

Low-pressure chemical vapor-deposited (LPCVD) silicon nitride films on silicon have been characterized by means of Rutherford backscattering (RBS), Auger electron spectroscopy (AES) combined with ion sputtering, and spectroscopic ellipsometry. It appeared that all LPCVD samples in the examined thickness range of 50 –500 Å had an oxygen-containing layer equivalent to 15–20 Å of SiO2 at the nitride-silicon interface. This interfacial layer originates from the native silicon oxide present at the silicon substrate when the deposition of nitride is started. For comparison, oxide-free silicon substrates were nitrided in ammonia at temperatures between 800–1160 °C. The thermal nitride films were found to be very thin, at the most 30 Å, even after 5 h of nitridation. Both the LPCVD and thermal nitride films oxidize slightly when transferred into the ambient; a surface layer equivalent to 8 Å of SiO2 was detected. Auger and RBS results agree very well for all nitride films investigated. It is shown that RBS can be made a very sensitive probe for such thin multilayer structures by performing glancing-angle scattering experiments under channeling conditions. Spectroscopic ellipsometry proves to be more sensitive to the interface structure than to the surface composition. Although quantitatively deviating results were obtained, in the extremely thin thermal nitrides this optical technique traces chemical intermediates, not perceived with RBS or AES.

Influence of Oxide Layers on the Determination of the Optical Properties of Silicon

Experimental reflectance data for silicon are corrected for the presence of a surface oxide layer and analyzed using the Kramers-Kronig relation to yield a set of optical constants for oxide-free silicon. For etched silicon samples exposed to the atmosphere for ∼ 1 h, this layer is 12–15 Å thick and has the approximate composition SiO. Absorption data are also given for epitaxial silicon on spinel and used as an aid in the analysis.