Clean Substrate Surface Research Articles

I n s i t u cleaning of silicon substrate surfaces by remote plasma-excited hydrogen

We have demonstrated a low-temperature cleaning technique for removing both carbon and oxygen from a Si surface. It uses a combination of ex situ wet chemical clean and an in situ remote rf plasma-excited hydrogen clean in an ultrahigh vacuum chamber. Since a remote rf plasma is used, there is insignificant plasma damage or other deleterious effects on surface morphology. A combination of in situ Auger and RHEED analysis has been used to confirm the removal of surface contaminants and the reconstruction of the Si surface. From mass spectroscopy studies, we believe that the hydrogen cleaning is due to a chemical etching of the Si by atomic hydrogen produced by the plasma. This clean is compatible with UHV processing and yields Si substrates that can be used for successful very low temperature (220–400 °C) Si homoepitaxy by remote plasma-enhanced chemical vapor deposition.

Room-temperature copper metallization for ultralarge-scale integrated circuits by a low kinetic-energy particle process

Copper films were epitaxially grown on (100)Si substrates at room temperatures utilizing low kinetic-energy particle bombardment of growing copper film surfaces. The crystallographic structure of the film, such as (100) or (111) orientation, was selected by controlling the energy of incident particles. Low-temperature, damage-free substrate surface cleaning has also been realized by the low kinetic-energy particle process, which has made it possible to form ideal metal-semiconductor contacts without employing any alloying heat cycles.

Bond-angle relaxation and electronic structure of Si and Ge overlayers on (110) surfaces of III–V semiconductors

We present the results of a detailed study of relaxational and electronic properties of Si and Ge overlayers on nonpolar (110) surfaces of GaP, GaAs and GaSb. Systems ranging from one up to three homopolar monolayers on semi-infinite heteropolar substrates have been investigated. Structure optimizations, restricted to relaxational degrees of freedom of the overlayer system give rise to a “pseudoionic” outermost adlayer. This pseudoionicity is induced by the adlayer relaxation and by the heteropolarity of the substrates and it leads to pronounced splittings of the Ge and Si surface dangling bond bands similar to the anion- and cation-derived dangling bond bands of the clean relaxed substrate surfaces. Considering the surface layer relaxation asymmetry as a function of overlayer thickness, we find that three monolayers of adatoms are sufficient to decouple the surface from the interface in the overlayer systems. The total energy minimization and electronic structure calculations are carried out in the framework of a second-nearest neighbour empirical tight-binding model. The semi-infinite configurations are treated by scattering theory using one-particle Green functions.

Effect of ion bombardment at low energy on (100)InP surfaces, studied by auger electron spectroscopy

Surface cleaning of (100)InP substrates with an Ar + ion beam of 250–400 eV is analysed by AES and shown as a function of time. The results obtained show the possibility of removing the contamination layer without any significant chemical damage to the InP surface.

Effect of ion bombardment at low energy on (100)InP surfaces, studied by Auger electron spectroscopy

Surface cleaning of (100)InP substrates with an Ar + ion beam of 250–400 eV is analysed by AES and shown as a function of time. The results obtained show the possibility of removing the contamination layer without any significant chemical damage to the InP surface.

Laser damage studies of polymer oxide coated silica optics at 350 nm

A single layer SiO2 film with graded porosity applied from a polymer solution derived from organometallic silicon compounds on fused silica lenses provides antireflectivity over the transmission range covering the first, second, and third harmonics of neodymium glass lasers. The damage thresholds of initial samples at 350 nm (3ω) varied from less than 1 to 9 J/cm2. The main causes of this large variation in the damage threshold were determined to be substrate polishing, substrate surface cleaning, and the coating solution chemistry and processing, which, under certain conditions, leads to carbon formation during the heat treatment of the coating.

Growth of Sr1−xBaxF2 films on InAs by molecular beam epitaxy

Heteroepitaxial films of Sr1−x BaxF2 have been deposited onto (001), (110), and (111)B oriented InAs substrates by molecular beam epitaxy. The epitaxial growth was studied by reflection high energy electron diffraction (RHEED), electron microscopy, x-ray diffraction, and secondary ion mass spectrometry (SIMS). The films grown epitaxially on the (001) and (110) substrates exhibit {111} faceting, and smooth and featureless growth occurs only on the (111)B oriented substrates. Single crystalline Sr1−x BaxF2 films lattice-matched to the InAs substrates are grown at temperatures higher than 200, 450, and 400 °C on the (001), (110), and (111)B faces, respectively. SIMS profiles indicate that the interdiffusion between epitaxial film and substrate is rather small even at a growth temperature of 520 °C. The change in RHEED patterns during the thermal cleaning of substrate surfaces is also described.

Reactive ion etching of GaAs in a chlorine plasma

Chlorine was used to carry out the reactive ion etching (RIE) of GaAs. At 5 mTorr pressure and 270 V bias, it was found to etch at a rate >1 μ/min, and produced features having vertical sidewalls and a clean substrate surface. The mechanism of chlorine RIE of GaAs was studied by examining etch rates and profiles obtained using gases of different mixtures of chlorine and argon. The results are consonant with proposed mechanisms of GaAs etching by chlorine in both the plasma and reactive ion beam etching regimes.

Chlorine adsorption on copper: I. Photoemission from clean Cu(001) and Cu(111) substrates

As a first step to an understanding of the atomic chemisorption of C1 on Cu(001) and Cu(111) we have studied high-resolution angle and photon-polarization dependent photoelectron spectra of the clean substrate surfaces. In a subsequent paper we will (besides presenting the energy dispersion curves of the adsorbate-derived bands) in particular discuss and interpret the adsorbate-induced changes within the substrate 3d-band emission spectra. Therefore a detailed understanding of the clean substrate emission was a necessary prerequisite. We demonstrate that a clear identification of bulk direct transitions, surface state emission and contributions from density of states transitions can be given. In particular we investigated the photoemission spectra from Cu(001) in the ΓXWK bulk mirror plane excited by NeI, Hel and HeII radiation, from Cu(001) in the ΓXUL plane excited by HeI radiation, and from Cu(111) along ΓLUX and ΓLKL (HeI). The experimental bulk direct transitions are in very good agreement with simple calculations based on Burdick's initial state band structure and a free electron like final state.

Cleaning chemistry of GaAs(100) and InSb(100) substrates for molecular beam epitaxy

The widely accepted sample-cleaning procedure for MBE substrates, involving the thermal desorption of native oxide-passivation layers in UHV, is examined using high resolution x-ray photoelectron spectroscopy. Oxide composition, carbon contamination, and substrate surface stoichiometry are followed for a variety of temperatures, for GaAs, Inx Ga1−xAs, and InSb substrates. The oxide desorption mechanism observed involves thermal reduction of As+5 to As+3 resulting in the further oxidation of the GaAs substrate surface and the additional formation of Ga2O3 and As0. At higher temperatures, first the As2O3 desorbs followed by the loss of Ga2O3. Carbon-contamination levels were minimized (<0.5% of a monolayer) by the etch removal of the chemical passivating layer by a HC1/EtOH strip immediately prior to vacuum introduction of the sample. The etch is done in inert-atmosphere conditions and gives total removal of both arsenic and gallium oxides, yet leaves the substrate stoichiometry unmodified. This method was quite effective for the Inx Ga1−xAs (x=0–1) system; however, halide-passivation films gave better results on InSb.

Elastic low-energy electron diffraction from GaAs(110)-p(1×1)-Sb(1 ML)

Auger electron spectroscopy and low-energy electron diffraction are used to study the interface formed by the evaporation of Sb on room temperature GaAs(110). In contrast to Al, the Sb overlayer is ordered and produces a (1×1) diffraction pattern with intensity profiles very different from those measured from the clean substrate surface. The interface is sharp and stable under heat treatment, the Sb film is continuous and thermal desorption experiments reveal the strong bonding that exists between the first Sb monolayer and the substrate.The atomic structure of the GaAs(110)-p(1×1)-Sb(1 ML) system is analyzed with multiple scattering computations. Three types of structures have been examined: chains of Sb atoms parallel and antiparallel to the top layer Ga–As chains, Sb2 dimers attached to the surface Ga species, and a ’’jellium’’ type structure in which one Sb is bonded to the surface Ga atom and the other Sb is randomly distributed above the surface. Only single scattering computation, however, has been used for the later model. Qualitative description of the measured intensities are achieved by structures of the first (antiparallel chains) and third models.

The Role of Neutral Atoms in Ion Plating

Abstract Ion plating is essentially vapour deposition on to a clean substrate in a glow discharge. Usually, the substrate is the cathode of the glow discharge and is cleaned by the sputtering action of the discharge. The adhesion between coating and substrate is excellent, even in those cases where the materials do not alloy.1,2 The requirements for good adhesion have been discussed by Mattox3 and in brief are a clean substrate surface and the formation of a graded interface, causing not only a good metallurgical bond, but dispersing stresses due to differences in properties, e.g. thermal expansion coefficients, between coating and substrate. Deep graded interfaces have been reported in ion plated specimens where alloying is possible,4 but not for incompatible materials. One of the main reasons given in the past for the excellent adhesion of ion plated films has been penetration of the substrate lattice by the ionised depositing vapour atoms, under the action of the substrate bias voltage, i.e. ion implantation. However, this view can be criticised on three counts. Even if the ions had the full energy of the discharge, i.e. 5 keV for a 5 kV discharge the depth of penetration would not be more than about 50 Å5 rather than the depths of several microns reported,4 also the ions must lose energy in collisions with neutrals and finally, only a small percentage of the atoms are ionised.6,7 Davis and Vanderslice8 have presented a theory of the energy distributions of ions in a glow discharge and Teer7 has extended this to obtain an approximate expression for the number of energetic neutrals produced together with the distribution of energy. The conclusion reached was that ion plating was the deposition of a small number of ions and a large number of neutrals, the mean energies of both ions and neutrals being of the order of 100 eV. There is little direct evidence with which to test this conclusion, but Komiya and Tsuruoka9 found that the major contribution to heat input to the substrate in ion plating experiments using a hollow cathode source and low substrate bias voltages came from energetic neutrals. In order to test the importance of energetic neutrals, under more conventional bias voltages and gas pressures, the experiment described below has been performed.

The Initial Epitaxy of Copper on (111) Gold Single Crystals

Two of the primary requirements needed for a successful study of epitaxy are a contamination free environment (UHV) and an extremely smooth, clean substrate surface. The deposition should be studied in situ by means of some surface sensitive technique (e.g. RHEED or LEED) and then in TEM and TED. Such a study has been conducted with Cu on (111) Au.Extremely smooth (111) single crystals of Au were evaporated on NaCl/mica in a UHV-RHEED system whose base pressure was 7 x 10-10 torr. After a 15 minute anneal at 625°C the films were cooled to room temperature. The initial growth of Cu on (111) Au at room temperature was studied is the following manner. Fractions of a monolayer of Cu were deposited (~1A per minute) onto the Au substrate and immediately examined in RHEED.

Gas adsorption studies by ellipsometry in combination with low energy electron diffraction and mass spectrometry

Isotherms for physical adsorption on a silver single crystal surface from ultrahigh vacuum have been determined by ellipsometry. Molecular volumes, derived from the optically determined monolayer film thickness, indicate that the classical macroscopic theory of ellipsometry for film covered surfaces is surprisingly well suited for physically adsorbed coverages of molecular dimensions. For a fractional monolayer coverage, the electron diffraction pattern has been found to remain characteristic of the clean substrate surface. The results also show the difficulties in the preparation and definition of clean solid surfaces, and indicate a sensitivity of the ellipsometer to surface roughness on an atomic level.

Mean Adsorption Lifetimes and Activation Energies of Silver and Gold on Clean, Oxygenated, and Carburized Tungsten Surfaces

Direct measurements of the mean adsorption lifetimes and the activation energies of silver and gold have been made on a polycrystalline tungsten substrate. The lifetime was determined by the time constant of the exponential increase of the surface concentration when an atomic beam impinged on the substrate. The activation energy was determined by the slope of the logarithm of the lifetime as a function of the inverse substrate temperature. The adsorption lifetime of atoms on a metal surface is τ=(1/ν0) exp (E/kT), where ν0 is the vibrational frequency and E is the activation energy. The experiments were conducted in an ultrahigh-vacuum system free of hydrocarbon contamination. A clean substrate surface was obtained by heating in a partial pressure of oxygen for a long period of time and flashing to 2500°K before each data point was taken. The well-defined surface condition and the precisely known temperature allow reliable interpretation of the experimental results. The experimental results for silver and gold on clean, oxygenated, and carburized tungsten substrates when τ is in seconds and E is in electron volts are the following: Ag on W, τ=(1/3.6×1012) exp (2.9/kT); Ag on O–W, τ=(1/1.43×1012) exp (2.06/kT); Ag on C–W, τ=(1/8.2×1012) exp (3.18/kT); Au on W, τ=(1/8.2×1013) exp (4.54/kT); Au on O–W, τ=(1/4.7×1011) exp (1.99/kT); Au on C–W, τ=(1/1.73×1014) exp (4.725/kT).

In-Situ Studies of Epitaxial Thin-Film Growth

Early stages of oriented overgrowth of Ag, Au, and Pd on thin, single-crystal substrates of mica, molybdenite, Au and Pd were studied by high-resolution electron microscopy and diffraction. Cleaning of substrate surfaces and deposition of evaporated materials were conducted inside an electron microscope. High-magnification, continuous observation during growth permitted investigation of the kinetics of growth. A number of probably elementary epitaxial processes were studied in detail. Nucleation and growth behavior was examined for different supersaturations and free surface energies of substrate and overgrowth materials. The influence of alloying on growth and the spacing of parallel moiré structures was investigated.

Effect of Substrate Cleanness on Permalloy Thin Films

The use of mechanical and chemical cleaning techniques in the production of a clean, random, and smooth microslide glass substrate surface for the vacuum deposition of uniaxial thin Permalloy film elements has been studied. The principal cleaning techniques intensively investigated were: (1) a mild detergent wash followed by a vapor degreasing cycle in isopropyl alcohol; (2) a chalk paste scrub followed by an ultrasonically agitated distilled water rinsing cycle and a forced hot air drying process; and (3) hydrofluoric acid etching followed by a distilled water rinsing cycle and a forced air-drying process. The surface condition of the cleaned microslide glass was assessed with electron micrographs of preshadowed carbon replicas (magnification 88 000×). The micrographs show that the chalk-cleaned glass substrate has the smoothest surface. The variously cleaned substrates were then used for the vacuum deposition of 1700-A 4-mm-diam Permalloy film elements (melt composition 83% nickel, 17% iron). This work was carried out in a 10−6 mm Hg vacuum system. Twenty-five evaporations were made with a total of 54 elements per evaporation. The Permalloy film elements deposited on the differently cleaned glass substrates were then examined by means of a 1000-cycle hysteresis loop apparatus and the following measurements were made: (1) coercive force HC; (2) saturation flux φs; (3) orientation of the anisotropy axis θ; (4) magnetoelastic strain coefficient η; and (5) anisotropy field HK. Dispersion measurements of the easy axis were made on a 1000-cycle crossed field hysteresis loop apparatus. Measured in this manner, the chalk-cleaned glass substrates yielded consistently the lower average value of HC; i.e., 1.4 oe. The average value of HK for the three different cleaning techniques was 2.5 oe. Within the total range of values of coercive force and the anisotropy field, the chalk-cleaned and the acid-etched glass substrates yielded the same values; i.e., HC±0.2 oe, HK±0.3 oe. Measurements of angular dispersion varied from 8° to less than 2° with an average value of less than 3°. The chalk-cleaned substrates yielded the lowest values of angular dispersion.

Monolayers of the globulin, arachin.

VARIOUS methods for spreading proteins have been devised. The method used by Gorter and Grendel1 is to drop a fairly concentrated aqueous protein solution very carefully from the least possible height on to the clean water surface in the trough. Hughes et al.2 found that solid particles of protein spread very rapidly on a clean substrate surface. These methods proved to be unsuccessful for the spreading of arachin. Arachin was also not soluble in the aqueous propyl alcohol–sodium acetate mixture recommended by Ställberg and Teorell3. Very satisfactory monolayers of egg albumin and β-lacto-globulin were obtained by Bull4 by using 35 per cent ammonium sulphate as the underlying solution. Accordingly, arachin was next spread on strong ammonium sulphate solutions, and films stable up to high film pressures were obtained. The ammonium sulphate was of analytical quality and the solutions were treated with activated carbon black to remove all surface-active impurities.