The use of sulfuric acid to strip photoresist from silicon wafers is a widely employed technique in the semiconductor manufacturing community. In most cases, the acid is combined with hydrogen peroxide to oxidize stripped photoresist material, though the use of sparged ozone in sulfuric solution is also used to remove resist residues on wafers following the ashing process. Although they have been used for many years, sulfuric acid processes also have proven to be costly. This is due to the need for frequent bath change-outs and the use of high temperature which impose safety and environmental concerns. As a result of these and other considerations, the use of mixtures of ozone and de-ionized water for photoresist stripping has been investigated. Results show that the technique effectively removes hard-baked resist (ashed and un-ashed) from bare silicon and patterned wafers and produces cleaner surfaces (i.e. particles and metals), compared to outcomes from SPM processes.

Over the recent years, many new approaches in wafer cleaning have utilized hot water rinsing. Benefits of hot water rinsing are (1) it aids the removal of residual chemicals and (2) it improves the performance of the dry, e.g. TREBOR hydrodry. However, there have been a number of significant disadvantages documented with regard to device performance. Earlier work done at this laboratory showed that the use of DI rinse water at elevated temperatures can cause etching of the Si surface producing unwanted surface topography and microroughness. The surface erosion created under these conditions had a negative device impact which led to product yield loss. In this paper, we provide a fundamental understanding of the mechanism by which Si is etched during hot water rinsing. To this end, we have explored the interaction of dissolved oxygen concentration, water temperature, and rinse time as it relates to the attack/dissolution of H-terminated silicon. Contact angle measurements, surface roughness measurements, and etch rate data from TEM measurements as a function of water temperature, rinse time, and dissolved oxygen concentration have been made. The etch rate data strongly correlates with the rinse water temperature, however little dependence was observed with dissolved oxygen concentration. A mechanism to explain the etching will be discussed.

To form atomically flat H-passivated Si(100) surfaces, wet chemical etching of sacrificial SiO2 layer has been examined. Roughness and chemical overlayer thickness, as monitored by ellipsometry shows a minima at an optimal solution of 1:0.5:30 HF(49wt\%):H2SO4 (98wt\%):H2O. A mechanistic study offers no evidence for a chemical smoothing from preferential non-Si(100) facet etching. Silicon planarization can be induced by rapid thermal annealing RTA of chemical oxides. The H-terminated Si(100) surfaces are found to be moderately reactive to ambient conditions as monitored by in-situ ellipsometry and Auger analysis. Atomic force microscopy (AFM) measurements show Si(100) surfaces to have a rms ∼1.0Å and Rmax values of 1.6–0.9Å. With measured roughness incorporate into ellipsometric model, a 5Å native oxide overlayer is rapidly incorporated into the Si(100) surface.

The hydrogen bonding features of Si(100) surfaces treated with BHF (NH4F/HF/H2O) have been studied by Fourier transform infrared attenuated total reflection spectroscopy (FT-IR-ATR). The amount of residual silicon-fluorine bonds on Si(100) surfaces has been evaluated by X-ray photoelectron spectroscopy (XPS). It is found that Si-H3 bonds appears to be preferentially removed by OH ions so as to increase the surface SiH2 and SiH bonds. On the other hand, it is likely that fluorine-containing ionic species such as HF2 might attack the backbonds of surface hydrides to produce Si-H3 and Si-F bonds. ATR spectra have shown that a BHF treated Si(100) surface in 5∼10% NH4F with molar ratios of HF/NH4F=0.37∼0.56 (pH=3.7∼4.0) at a treatment time of 5 minutes is atomically flatter than that treated in BHF containing 15∼20% NH4F. This is because the amount of residual Si-F bonds on Si(100) increases with HF2 concentration in BHF and these Si-F bonds enhance attacking of silicon backbonds of Si-F bond by OH− and HF2− ions.

By using a combination of smooth epi substrates (0.7 Å rms roughness), in-situ UV/Cl2 processing, and rapid thermal oxidation, highly reliable ultrathin gate oxides were produced with ≤ 1.0 Å rms interface roughness.

The problem is occurring that gate oxide integrity (GOI) might be influenced on the residual subsurface damages induced by mirror polishing. The correlations between mirror polishing conditions, micro roughness, and photoconductivity amplitude (PCA) signals are measured and discussed in a small roughness region of micro roughness Rrms such as 0.095 nm to 0.247 nm. The theoretical analysis shows that PCA signal is possible to reflect a subsurface lifetime.The discussions lead to result that the GOL is mainly dependent on subsurface damages in such small roughness region, PCA signals measured with an UV/mm-wave noncontact technique correlate closely with the GOI yield. It is concluded that the PCA technique makes it possible to characterize subsurface damages and to estimate GOI yield.

Comparative studies on the effect of Ultra-dilute RCA cleans, chemical ratios in excess of 300:1, and Dilute RCA cleans, chemical ratios around 50:1, on the integrity of thin gate oxides have been performed. Ultra-dilute RCA chemistries have shown particle removal efficiency, metallic contamination removal, surface roughness, Qbd, BVox and defect density equivalent to those obtained using dilute RCA chemistries. Furthermore ultra-dilute chemistries use less chemical leading to shorter rinse times and thus increased throughput as compared to the dilute RCA chemistries.

Dilute HF/RCA and IEMC/SC2 cleans have been evaluated on two process lines with different metallic contamination levels. VPD-DSE-TXRF and SPV techniques were used to monitor the metallic contamination. Gate oxide integrity(GOI) tests were performed on several structures. Both HF/RCA and IMEC/SC2 cleans have shown good Qbd and Ebd results for the clean process line. Lower Qbd and Ebd values were obtained for both cleans in the relatively contaminated process line. These results suggest that poor GOI is related to the metallic contamination in the oxide or at the SiO2/Si interface.

The drive to reduce chemical consumption for cost savings is high in the semiconductor industry. Recently, studies have shown that the traditional 1:1:5 ratios of the SC1, SC2 cleans can be reduced and still maintain good cleaning. The optimization of process parameters must be done in accordance with the type of cleaning tool. Processes in wet decks may be optimized differently from single chamber tools, especially if extended bath lifetimes are expected. In this study, temperature reduction and dilution of the SC1 and SC2 cleans in a wet deck are examined for their effects on metals removal efficiency from silicon surfaces as measured by TRXRF. For an HF, SC1, SC2 sequence, good metal removal can be maintained by dropping the SC1 temperature down to 35°C and chemical ratio to 1:1:40. At too low an SC1 dilution and temperature, the Cu removal efficiency drops. In SC2, good metals removal remains without peroxide and a lower limit for HC1 is determined. It is also found that peroxide must be reduced if HCl is reduced in SC2. otherwise metals plating out from SC1 cannot be removed effectively. To understand the impact of these reduced cleans on gate oxide integrity, the electrical properties of 30 A gate oxides grown using these experimental pre-gate cleans is discussed.

The effect of different silicon wafer surface preparation in modulating gate oxide quality performance has been studied through an experimental design which examines key phases of wafer cleaning and polishing processes. An interpretation of the root causes of GOI degradation has been proposed and discussed.

A comparative study of pregate cleaning solution chemistries has been conducted, for 30Å furnace oxides, to investigate the feasibility of continuing to use current cleaning technologies in this thickness range. The experiment focused on correlating changes of device properties with changes in chemistry. Several options were investigated for growing or etching the surface passivation oxide. Chemistries studied included standard SC1 (with Megasonic)/SC2, final HF, as well as room temperature single bath HF based chemistries (HF only, HF+H2O2, HF+HCl, HF+H2O2+HCl, SPM+HF). These chemistries were evaluated in terms of oxidation rate, metallic contamination from solutions, capacitance-voltage (CV) characteristics, and gate oxide integrity. Data suggest that furnace oxides can be reliably grown using conventional cleaning technology. HF-based mixtures might be justified for even more aggressive gate oxide thicknesses.

Si surface cleaning has been studied intensively in these days. One of the major concerns is about the removal of metallic impurities on Si surface. In this study we choose contaminant element of Cu which shows high electronegativity value compared to Si's. The Si substrate was cleaned with using piranha (H2SO4:H2O2=4:1) and HF (HF:H2O=1:100) solutions to eliminate the organic impurities and native oxide. The initial Si substrate was then contaminated intentionally by dipping into the 1 ppm standard solution of Cu and cleaned by the cleaning splits of the chemical solution HF combined with UV/O3 treatment and the chemical mixture of H2O2 and HF. The initial substrates which were contaminated with the standard Cu solution exhibited the contamination levels of 1013–1015 atoms/cm2 and these substrates were cleaned and showed Cu concentration down to the levels as lower as 1010 atom/cm2 which were monitored by TXRF (total reflection X-ray fluorescence). And repeated treatments of these cleaning splits improved the surface microroughness of these initial substrate which were measured by AFM (atomic force microscope). The surface and interface morphologies contaminated with Cu were examined by SEM (scanning electron microscope) and TEM (transmission electron microscope). It was observed that Cu impurities were adsorbed on Si surface not in a thin film but in a particle form as a hemispherical shape. The chemical composition of Cu impurities and interface between Cu and Si substrate was investigated by AES (Auger electron spectroscope). The result exhibited that Cu element adsorbed on Si surface by chemical adsorption and this resulted in oxidation of Si substrate.

We investigated a dependence of cleaning efficiency for Cu on the initial contamination method with different initial concentration and contamination solution. Through this study, we found that the cleaning efficiency of typical dissoluble chemical solutions such as O3-UPW, SPM and HPM for Cu contaminated in diluted-hydrofluoric acid (DHF) was better than for that in ultrapure water (UPW). And, it is shown that there is a dependence of cleaning efficiency for Cu contaminated in UPW. The different desorption behaviors of Cu contaminated in UPW and DHF can be interpreted by the fact that the Cu is included in metal-induced-oxide (MIO) grown during contamination in UPW and can not be removed by only dissolution process of O3-UPW, SPM and HPM solutions which have no etching effect on SiO2 and Si. In contrast, a relatively low concentration Cu contaminated in DHF which has no MIO effect can be removed below the detection limit by only dissolution process of these chemical solutions. However, Cu contaminated in DHF is dissolved and inserted in the chemical oxide grown during cleaning process as the initial Cu concentration is a relatively high. Consequently, we conclude that Cu on Si surface can be fundamentally removed by just dissolution process of typical dissoluble chemicals except the amount of Cu inserted in MIO grown during contamination in UPW and chemical oxide grown in chemical solutions.

The use of H2SO4:H2O2:HF (SPFM) at low HF concentrations (10 to 1000 ppm) has been investigated as the preparation procedure prior to formation of Si/Si interfaces by wafer bonding. The SPFM cleaning process makes it possible to form a hydrophilic (OH terminated) silicon surface, thereby achieving a spontaneous and strong room temperature bond. Electrical characterization using current vs voltage and spreading resistance measurements shows that this cleaning procedure can be used to form Si/Si junctions with excellent electrical properties. Some of the problems related to hydrophobic wafer bonding can thus be circumvented by the proposed technique.

Cu particle growth behavior on two silicon substrates, amorphous and single crystal silicon, has been investigated using two contamination solutions. This study reveals that the growth behavior of Cu particle depends on substrate conditions and copper contamination solutions. Contamination level is independent of split conditions. From the SEM images of an amorphous silicon shows a big difference in the number of particles depending on copper contamination solution. The amorphous silicon has similar native oxide thickness in ultrapure water spiked with CuF2 and CuCl2, whereas the single crystal silicon is different from the native oxide thickness depending on copper contamination solution. When 1 ppm of Cu in ultrapure water was spiked as a function of time, the amount of Cu impurity on amorphous silicon in the early dipping stage was measured 10 times higher than that on single crystal silicon for both of copper contamination solutions.

Three different HF:alcohol solutions are investigated to etch native SiO2 and passivate Si(100) surfaces with H which can the be desorbed at low temperature (T < 600°C). The resulting passivated Si(100) surfaces are compared using as a reference Si(100) passivated by a standard aqueous HF: solution (1:98 parts of HF: H2O). After a modified RCA cleaning, Si(100) etched by HF:Methanol, HF:IPA, or HF:Ethanol, is characterized by Ion Beam Analysis (IBA), Tapping Mode Atomic Force Microscope (TMAFM), and Fourier Transform Infrared Spectroscopy (FTIR). The absolute coverage of O and C is measured by nuclear reaction analysis (NRA) combined with ion channeling at 3.05 MeV for O and 4.265 MeV for C. Hydrogen is measured via the elastic recoil detection (ERD) of 4He2+ at 2.8 MeV.

Compared to aqueous HF, HF:alcohol passivates Si(100) leaving a lower O residue by an average factor of 0.62 and a similar C residue. H coverage is higher by an average factor of 1.43. Surface coverages are found to be reproducible in average by 1.4 × 1014 atoms/cm2 for C, and by 1.25 × 1014 atoms/cm2 for O when measured by IBA on samples identically processed. H coverage is reproducible within 5.5% when measured by ERD.

Selective area analysis by TMAFM shows that an increasing number of particulates is responsible for the apparent increase in root-mean -square (rms) surface roughness when the rms is measured over a whole image. Taking this effect into account, all passivated surfaces exhibit similar roughness when compared to the original Si(100) surface with little difference between alcohols and with the reference aqueous HF solution.

FTIR in the attenuation total reflection (ATR) mode detected SiHx species mostly as a dihydride. Both IBA and FTIR detected significant levels of oxygen on surfaces passivated HF in alcohol and aqueous HF. This indicates that while Si(100) exhibits more H when passivated with HF in alcohol and can be desorbed at lower temperature than when treated with aqueous HF, H is not bonded to Si only but likely bonds into a more complex surface termination, such as SiOH.

We have investigated the oxidation of clean reconstructed silicon surfaces in-situ using a fourpoint probe technique. The measured conductance variations, on Si(111) and Si(100) surfaces as a function of oxygen exposure are markedly different. On Si(100) surfaces, the conductance displays a rapid fall during the first 100 L exposure to O2, followed by a slower steady decrease at higher oxygen exposures. This behavior is similar for both n-type arid p-type silicon. The conductance of Si(111) surfaces increases significantly at the onset of the oxidation. This conductance increase is found on both n-type and p-type Si(111) samples. We interpret this as being due to the molecular precursor which is known to form in the peroxy bridge position.

We show by ex-situ STM that pit-free H-Si(111) surfaces can be prepared by immersion of pre-oxidized Si(111) in deoxygenated 40% NH4F for 15 minutes. In contrast, H-Si(111) prepared in air-saturated 40% NH4F for 15 minutes results in small etch pits randomly covering ca. 1% of the surface. Added dioxygen, shorter etching times or stirring of the air-saturated etch solution increase the etch pit density. Dissolved oxygen is responsible for the initiation of etch pits on the (111) monohydride terraces. The reaction between dissolved oxygen and the H-Si bond of the (111) terrace is limited by mass transport of dissolved oxygen to the silicon surface. A proposed reaction mechanism of dissolved oxygen with H-Si is briefly mentioned, which explains not only etch pit initiation in ammonium fluoride but also the oxidation of hydrogenterminated silicon in wet, oxygenated environments. The results presented here for H-Si(111) lead to a greater understanding of the wet oxidation and etching processes of this important model surface.

A comparison of SurfaceSIMS measurements of surface metal contamination on silicon wafers with TXRF analyses of the same wafers is presented. These two important surface analysis techniques are characterized by differing analytical areas and sampling depths. For Fe, agreement between SurfaceSIMS and TXRF results is good (generally within a factor of two) for silicon wafers from a variety of sources with contamination levels ranging from 8×109 to 2×1015 atoms/cm2. Differences between SurfaceSIMS and TXRF results occur for other elements (such as W and Cu) when the contamination is not uniform across the surface of the wafer (particles are present), or extends a significant depth below the surface of the wafer (as is frequently the case for ion-implanted wafers).

The continuously increasing integration of today's advanced semiconductors requires increasingly tight process control in the IC manufacturing steps. This paper demonstrates the use of conductivity sensors to monitor and control the chemical concentrations of RCA cleaning and HF etching solutions. Electrodeless conductivity sensors were used to monitor and regulate the concentration of these process chemicals. A linear relationship between the conductivity of the solution and the chemical concentration was obtained within the range studied. A chemical monitoring and concentration scheme (ICE-1™) was developed. Different concentrations of RCA and HF solutions were investigated. Results show that these techniques are suitable for monitoring and controlling the concentration of chemicals in the process tanks for better process control. These techniques provide a lower cost of ownership of the process due to longer bath lives and the use of dilute chemicals.