Physical forces control particle deposition on wafers in virtually all manufacturing environments: open air, liquid baths, and subatmospheric chambers. Physical forces also often dominate particle adhesion and removal. In aerosol systems, gravity and diffusion are important universal mechanisms contributing to particle deposition velocity (= particle flux to a surface/particle concentration adjacent to that surface). Under appropriate conditions, electrical and thermal forces are also important and sometimes dominating. Both these conditional mechanisms can be introduced so as to minimize particle deposition velocity and thus protect wafers from particulate contamination.

In liquid systems where particle and liquid densities are similar, gravity is a less important particle deposition mechanism. Convective diffusion remains an important mechanism but typically contributes a much lower deposition velocity than in aerosol systems (particle diffusion coefficients are smaller in liquids than in gases). In many liquid systems, electrical forces, arising from particle and surface charging by ion adsorption from the liquid or ionization of surface groups, dominate particle deposition by introducing electrostatic double layer repulsion.

The primary force binding a particle to a surface is van der Waals attraction which is universal and dominating when separation distances between a particle and a surface decrease below I to 2 nm. Van der Waals attractive forces are lower in liquids than in gases so particles are easier to remove in liquids, partly because of this reduced adhesion.

Physical forces—centrifugal, hydrodynamic, vibrational—typically dominate particle removal technology as well. These forces depend directly on fluid density and hence are larger in liquid systems than in gases, another reason why liquid systems remain more successful at particle removal than dry methods.

The smoothness of silicon wafer surfaces is defined by polishing. Subsequent cleaning steps take account of surface conditioning of the wafer and meet the demands for low metal surface contamination as well as low particle level. Wet chemical treatments, however, can also influence surface morphology. The coexistence of oxidation and etching of the Si surface has a pronounced feedback on the resulting surface roughness. This holds in particular for SC1 cleaning solutions and modified HF treatments. Buffered HF enhances surface roughness of the technological important Si (100) surface, in contrast to a smoothening effect on Si (111). The same holds for other modifications of HF treatments with an enhanced oxidation behaviour involved, like HF-H2O2 solutions. In parallel with the change of surface morphology and with the increase of surface roughness an increase in Light Point Defect (LPD) densities is also found. Small LPDs are predominantly not particle related, depend on the respective silicon substrate material and were not found to influence gate oxide integrity.

The effect of metal contamination and silicon surface defects on the gate oxide yield is investigated. The characteristics of various cleaning procedures are studied and correlated with the integrity of thin gate oxides. The standard wet cleaning recipe is optimized and a new cleaning strategy is proposed. Selective contamination experiments in chemicals and on Siwafers are used to investigate the effect of small amounts of metal contaminants on the gate oxide integrity. It is found that the characteristics of the silicon substrate play a dominant role in this. HF-last processes are investigated and a new wet cleaning strategy is proposed.

A critical step for successful growth of ultra-high vacuum chemical vapor deposition (UHV-CVD) epitaxial silicon films at low temperatures is the preparation and maintenance of a hydrogen passivated surface. A final 10:1 HF dip has been shown to provide a robust and extremely clean substrate for this purpose; interfacial oxygen and carbon levels of less than 1 × 1013 atoms/cm2 and low levels of structural defects are routinely observed after epitaxial growth. In this work, our understanding of the Si (100) hydrogen passivated surface is reviewed with emphasis on its stability to air and moisture exposure. Additionally, in an effort to reduce interfacial contamination levels even further and explore clusterable processes, other precleaning chemistries such as high pH NH4F and vapor HF have also been evaluated. While surface analysis measurements suggest that the NH4F preparation should be as good or better than the HF dip, the epitaxial film defect density is higher by several orders of magnitude than for the dilute HF preparation. The vapor HF process, while potentially clusterable and therefore desireable, has generally not provided hydrogen passivated surfaces that are clean enough for this application.

Hydrogen terminated on Si surface actively contributes to the succeeding selective chemical vapor deposition (CVD) of Al using dimethylaluminum hydride \DMAH, (CH3)2AIH] and H2. In this article, we describe our recent progress of area-selective Al CVD on H-terminated Si surface. By thermal decomposition, 0.3µm-diam./1µm-deep SiO2-via-holes on Si surface are filled with the selective single-crystal Al. As a selective growth mechanism of Al on Si, we have proposed a “surface electrochemical reaction” model. The most significant point is (1) catalytic contribution of surface free–electrons, (2) surface terminated–H, and (3) selective reaction between terminated–H and CH3–radical of DMAH. Recently, we have performed area–selective deposition of Al on Si by patterning the monolayer terminated–H, i.e. atomic hydrogen resist process. The terminated–H on Si is intentionally desorbed by the electron beam exposure. The activated Si surface is oxidized in the air environment. Then, single–crystal Al has been successfully deposited on the remaining H–terminated area and not deposited on nanometer–thick oxidized area formed by the electron beam lithography.

The effects of various ex-situ and in-situ silicon surface cleaning techniques on low temperature epitaxial growth down to 650°C have been characterized. For ex-situ cleaning, wet HF cleans with and without a chemically grown oxide were compared along with vapor HF cleaning. In-situ cleaning was achieved by rapid thermal cleaning (RTC), a 1min. hydrogen bake at various temperatures. When the residual surface oxide was >2e15/cm2 or fluorine >4e13/cm2, a continuous interface was observed and epitaxial growth could not be achieved (conical/columnar poly deposition). By optimizing the in-situ RTC process for either wet or vapor HF ex-situ clean, high quality low temperature epitaxial growth was realized.

The use of an aqueous HF dip as the final ex situ cleaning step following a modified RCA clean before growth of epitaxial Si layers by Remote Plasma-enhanced Chemical Vapor Deposition (RPCVD) has been found to cause haze. Two approaches have been employed to eliminate the haze: addition of alcohol to lower the surface energy of the HF solution, and the application of the HF solution by a spin etch technique. We have studied final ex situ treatments with water/HF, ethanol/HF and methanol/HF solutions applied by both dip and by spin methods, and their effects on haze in low temperature Si homoepitaxy by RPCVD. The ratios of solvent to 49% HF (aq) were 40:1, 10:1 and 5:1. Surface carbon and oxygen levels, as measured by Auger Electron Spectroscopy (AES) after the HF solution treatments were found to be comparable for the alcohol/HF and water/HF dips. For the spin application, methanol/HF resulted in higher surface carbon coverage than for ethanol or water HF solutions. Epitaxial silicon films, 2000–4000Å thick, were grown on the wafers at 450°C using RPCVD. Secondary Ion Mass Spectroscopy (SIMS) measurements showed elevated levels of oxygen and carbon contamination at the epi-substrate interface to result when the wafers were treated with water/HF solutions rather than alcohol/HF, but no advantage was observed for the spin versus the dip technique. Films which were cleaned by water/HF exhibited patterns of haze over the wafer surface showing flow lines where the liquid drained from the surface for both dip and spin applications. The alcohol/HF solutions resulted in substantial reduction in haze for the dip application and elimination of haze for the spin application.

The deposition of perfect growing layers requires an atomically clean surface. In this paper we will describe a new in situ cleaning technique which is able to reduce the temperature/time load of silicon substrates significantly.

First we will investigate the efficiency of the in situ UV/ozone cleaning immediately before deposition. Such an UV/ozone treatment lowers the level of impurities below the detection limit of in situ XPS. The known recontamination if exposing to air after ex situ cleaning is prevented by this in situ technique.

Nevertheless a significant contamination within the UHV chamber is detected by SIMS. A comparison of various UHV pumping systems (base pressure in the range of 10-8 Pa) shows the specific influence on recontamination at an atomically clean surface.

A blocking mechanism by a thin oxide layer below (0.1 ML) to avoid the growth of stacking faults within an epitaxial silicon layer was successfully tested.

An argon/hydrogen discharge plasma was created with a UHV compatible plasma source. This low energy plasma was utilized to remove – in a single step – the native oxide and the hydrocarbons from the wafer surface at substrate temperatures between 100°C and 400°C. During the plasma cleaning procedure, residual gas ions were monitored to elucidate the process chemistry. To optimize the procedure, the interfaces between the plasma cleaned wafer and the MBE grown epilayer were investigated by scanning electron microscopy (SEM) and cross-sectional transmission electron microscopy (XTEM). The cleaning process was also applied to patterned silicon substrates. Subsequent local epitaxial growth by MBE at 550°C without the typical high temperature annealing step was achieved, suggesting that the in–situ dry cleaning procedure caused no surface damage.

The passivation of silica surfaces and the subsequent adsorption/desorption of (hfac)Cu(2-butyne) and (hfac)Cu(VTMS) precursors were studied. The goal was to enhance the selective deposition of copper from these compounds during chemical vapor deposition (CVD). The passivation reagents were hexamethyldisilazane (HMDS) and triisobutylaluminum (TIBA). Surface passivation with HMDS shielded the hydrogenbonded surface hydroxyl groups and the strained siloxane surface sites making them inaccessible to both (hfac)Cu(VTMS) and (hfac)Cu(2-butyne) where hfac = 1,1,1,5,5,5-hexafluoroacetylacetonate and VTMS = vinyltrimethylsilane. The adsorbed copper(I) compounds molecularly desorbed from the HMDS-passivated surface below 125 °C. The surface groups resulting from the reaction of TIBA with the silica surface reacted with the copper compounds subsequently adsorbed at 275 °C. The reaction products were present on the surface above 450 °C. These results demonstrate the utility of SiO2 surface modification using HMDS for control of selective CVD of copper.

We have studied selective chemical vapor deposition (CVD) of copper using (hfac)Cu(VTMS). Copper films were deposited selectively onto tungsten (W) and titanium nitride (TiN) in the presence of silicon dioxide (SiO2) surfaces in a warm-wall CVD reactor. Selective deposition was achieved by in-situ passivation of SiO2 surface hydroxyls by exposure to a silylating agent. These agents, such as dichloro-dimethylsilane and hexamethyldisilazane (HMDS), suppressed the nucleation rate of copper on SiO2 compared to metal surfaces. Control over loss of selectivity during CVD was achieved by maintaining the flow of the silylating agent during deposition. The passivation chemistry was studied by FTIR on a model Cab-O-Sil surface which showed that dichlorodimethylsilane and HMDS pre-dosing shielded the hydrogen-bonded surface hydroxyl groups making them inaccessible to the precursor molecule. An understanding of SiO2 surface modification and control over loss of selectivity during CVD is a key for successful selective copper deposition.

A new surface cleaning method for Si MBE is described in which the Si surface is exposed to Ge beams while the substrate is kept at certain temperature. It has been proved that the thin passivation layer of SiO2 on the Si substrate will react with Ge at a relatively low temperature (620°C), and the products are volatile. The residual Ge on Si substrate can be reduced to less than 0.1 monolayer (ML). Ge beam treatment turns out to be an effective low temperature technique for preparing Si substrate, especially for the heteroepitaxial growth of GexSi1-x/Si.

Chemically vapor deposited borophosphosilicate glass (BPSG) has been widely used in microelectronic device fabrication as interlayer dielectric film due to its excellent planarization, gettering and flow properties. A BPSG reflow step, performed either in a diffusion furnace or a RTP reactor, normally follows the deposition step. With device geometry reducing to sub micron levels, there is an increasingly greater emphasis on detection and elimination of surface contaminants and sub micron defects particularly on deposited film. In this paper, we report on the implications of surface contaminants in a BPSG deposition/reflow process. The deposition and reflow processes have been carried out in an atmospheric-pressure CVD system and a vertical diffusion furnace respectively. Ellipsometric measurements, SIMS, TXRF, and vapor phase dissolution (VPD) methods have been used to detect both metallic and non-metallic contaminants as a function of both BPSG deposition and furnace reflow process parameters. It was found that surface contaminants emanating under specific deposition conditions can cause an uncontrolled, enhanced oxidation during the reflow process. Process modifications and in line defect monitoring schemes are discussed.

This paper presents the results of surface cleaning and passivation of Si surfaces and sulfurized SiO2 surfaces for the growth of Si/SiO2/Si structures. Silicon surfaces were cleaned by the conventional cleaning process prior to the growth of silicon oxide, in which Si wafers were vibrated in ACE and TCE solutions, boiled in H2SO4 solution, rinsed in DI water, and dipped in HF solution. A polycrystalline silicon film was deposited using SiH4/H2 gas mixtures after the subsequent surface cleaning and sulfur treatment on the surface of SiO2/Si at 250° C. Sulfurization of SiO2 surfaces using (NH4)2Sx (amonia sulfide with excess sulfur) and (NH4)2S was compared to investigate the influence of surface treatment. The passivation of SiO2 surfaces via hydrogen plasma was also investigated. Both the SiO2 and the silicon layers were grown in a plasma-enhanced chemical vapor deposition (PECVD) system. The sulfurized SiO2 surface contributed to the low temperature (≈250°C) poly-Si growth on SiO2/Si. The results of this technique are essential to thin film transistor (TFT) and silicon-on-insulator (SOI) device fabrication. The interface quality between layers of deposited Si/SiO2/Si and grain size of poly-Si were examined by cross section transmission electron microscopy (XTEM).

Electrochemical cleaning processes generally use preheated (50-60°C) concentrated alkaline type soaps together with a low voltage current that can either be direct, reverse, or periodic. The effect of this cleaning process on metal surface and its influence on long term strength on metal-adhesive bonding is superior to normal industrial cleaning processes such as immersion or soaking, rotary and spray wash. Therefore, in this electrochemical cleaning and electro-rinsing method, the metal parts are treated in hot alkaline, hot acid solutions with anodic current prior to bonding, The metal surface electrochemical/electrorinse anodizing process was used in order to improve the metal surface and activate it for long term strength bonding.

Various cleaning procedures used to prepare InP surfaces for epitaxial growth and regrowth have been investigated. They include standard surface chemical cleaning treatments, UV/ozone oxidation, (NH4) 2Sx sulfidation and in situ ECR hydrogen and oxygen plasma treatments. The quality of the various treatments is evaluated on the basis of the surface chemical composition prior to growth and from the structural and electrical properties of the corresponding epilayers.

X-ray photoelectron spectroscopy was used to estimate the surface Fermi level position at S-passivated InP samples. Low temperature band edge photoluminescence intensity and Schottky diode current-voltage characteristic measurements were used to investigate the surface electronic properties of S-passivated InP. The results show that surface electronic properties were greatly improved by the S-passivation. The passivated InP samples have good resistance to heat treatment, and to the formation of non-radiative recombination centers. The barrier height and ideality factor of Schottky diodes were improved by the passivation. These superior surface electronic properties are attributed to the formation of strong and stable S-In bridge bonds on the passivated InP surfaces.

The chemical species present at the surface of InP substrates at the different stages of a classical cleaning (degreasing, etching, de-oxidation and UHV annealing) were studied by infrared absorption (IR). Traces of oxides were detected from nuclear reaction analysis (NRA). With these analysis techniques, after the complete treatment, no defects were revealed at the surface of the substrates. RHEED patterns showed that they were well-reconstructed and SrF2 thin films were succesfully epitaxially grown on the semiconductor. However, capacitancevoltage (C-V) measurements performed on MIS (SrF2/InP) structures showed that to obtain a large modulation of the surface potential of the semiconductor, a sulfurization of the substrates ((NH4)2Sx) was required after the etching phase. These results show that many complementary analysis techniques must be used to completely describe the surface state of a cleaned semiconductor.

Simple chemical sulfur treatments of GaAs have been shown to passivate the GaAs surface. These treatments result in lower surface state density, lower surface recombination velocity, and shifting or unpinning of the Fermi level, in addition to improvement in the performance of GaAs devices. Electrochemical sulfur treatment, however, has only recently been explored and pursued as a method of growing anodic surface layers which have good passivating characteristics on semiconductors. In this study, using surface infrared reflection spectroscopy (SIRS), x-ray photoelectron spectroscopy (XPS), and Raman spectroscopy, we have investigated the electrochemical sulfidation of GaAs as a method to produce a GaAs surface that has good electronic properties and is stable chemically and electronically. We have found that anodic treatments with Na2S and (NH4)2S solutions resulted in the removal of the pre-existing oxide of GaAs and in the formation of films comprising sulfur, sodium carbonate, ammonium thiosulfate, and various sulfide and sulfur-oxygen compounds of arsenic. The surface state density of this anodically treated surface was significantly better than that of untreated GaAs. Rinsing the GaAs with water removed the bulk of the film, leaving behind a surface on which only arsenic sulfide was detected. The surface state density after rinsing has degraded slightly, however, but it was still better than that of an untreated GaAs.

We apply the reflectance microwave probe (RMP) method to characterize the surface condition of the sulfur-treated GaAs. Undoped semi-insulating GaAs wafers are dipped in (NH4)2S or (NH4)2Sx [1<x≦3] after an etching. The intensity of the reflected microwave under He-Ne laser (λ=633nm) irradiation is proportional to the excess carrier density, which strongly depends on the surface condition. The (NH4)2Sx treatment increases the RMP signal intensity, while no significant change in the signal intensity is observed after the (NH4)2S treatment. The signal intensity of the (NH4)2Sx-treated samples is decreased by the annealingyhe results observed in this study are consistent with those previously found with other measurement methods. This method will be able to in-situ evaluate the surface condition of GaAs during the device fabrication process.