It is shown that a new type of singularity is formed on Si wafer surface by the Standard Cleaning 1 (SC1) of the RCA cleaning process″. Such singularities are perceived by laser particle counters as small particles on wafers. It is revealed that the singularities correspond to small shallow pits caused by the etching effect of the SCI cleaning solution. The origin of the pits seems to be some kind of defects in the melt–grown Si crystals. The authors named such “particles” as crystal originated “particles” (COPs).

The size–distribution of COPs after single SC1 cleaning cycle is estimated on the basis of variation in the number of COPs with the repeated cleaning cycles. It is revealed that the crystal pulling rate affects the size distribution of COPs. As the pulling rate becomes faster, COP becomes larger in number. From the total number of COPs after the first cleaning cycle and the etching depth, the volume density of origin of COPs can be estimated. The results show that, as the pulling rate becomes faster, the volume density of origin of COPs increases.

An experimental protocol using secondary ion mass spectrometry to measure Al in vapor phase decomposition (VPD) solutions has been developed to achieve a detection limit of 108 atoms/cm2. The analytical technique is called VPD/SIMS. The procedure utilizes sample preparation techniques developed for microvolume SIMS. Preliminary SIMS analysis of simulated VPD solutions consisting of dilute HF indicate that a theoretical detection limit of 5.7×105 Al atoms/cm2 is possible from a 6″ wafer.

Metal contamination on the surface of silicon substrates before gate oxidation is known to affect gate oxide reliability. For the first time this study presents a non-destructive, analytical measurement of transition metals in an 8nm gate oxide grown by a 920 °C-10min-dry oxidation of an intentionally contaminated silicon surface. The TECHNOS TREX 610 TXRF anglescan of the gate oxide provides qualitative information on the location of the metals. The data indicate the Fe is on or in the oxide, the Cu is below the oxide, the Zn is on the oxide, and the Ni may be both in the oxide and below the oxide layer. In addition, quantitative estimates from the TXRF data indicate that all the original Fe and Cu are present, while only portions of Zn and Ni are detected after the oxidation.

A contamination control study of a Silicon Valley Group Thermco Systems Vertical Thermal Reactor(VTR) is presented. Trace elements of contaminants such as water vapor and oxygen have been shown to significantly affect the integrity of the silicon nitride film deposited by the low pressure chemical vapor deposition (LPCVD) process. This study documented the effects of process parameters on gaseous contamination levels, i.e., O2 and H2O vapor. Starting with a baseline process, the effects of an excursion of pre-deposition temperature ramp-up and stabilization condition, wafer load/unload and various post deposition conditions were explored. An axial profile of moisture and oxygen levels along the wafer load was obtained using Linde's Low Pressure Reactor Analysis(LPRAS) methodology. In addition, other process parameters such as gas flow rates during load and unload of wafers, pre-deposition N2 purge and process tube exposure time to ambient environment were- investigated. The wafers were analyzed for contaminants on the wafer surface or in the deposited silicon nitride film using FTIR and Auger spectroscopy techniques. They showed low levels of Si-O and no measurable Si-H or N-H bonds.

KOH, an anisotropic etchant of monocrystalline Si, may cause roughness and defects whose shapes are related to crystallographic orientations. This paper studies the effect of processing steps on the formation of geometric etch defects. Implantation, thermal treatment, epitaxial growth or photoresist were not the source of such defects. In the scope of this study, only unwanted damage caused geometric etch defects. This makes the observation of the wafer after KOH etch a good indicator of the quality of previous steps.

Chlorine radicals generated with UV irradiation are effectively used to remove residual metal contaminants after vatious LSI processes. Aluminum, iron, and copper atoms are effectively removed as volatile chloride species from silicon surfaces. The mechanism has been studied using silicon wafers intentionally contaminated with those metal ions. Contaminants are involved in native oxides as metal oxides or hydroxides. Chlorine radicals penetrate the native oxides and attack those oxides or hydroxides vaporizing metal chlorides. The cleaning process is accompanied with slight etching of silicon surfaces to a depth as thin as a few nm. However, this is not essential because the substrate temperature is more important than the etching depth. Alkaline metals are also removed from the surfaces to the level as small as the detection limit of atomic absorption spectroscopy. We believe that those are removed through a lift-off process. The, wixture of fluorine and hydrogen gases removes a native oxide of silicon under UV irradiation. Hydrogen fluoride radicals react with native oxide resulting in hydrogen termination on silicon dangling bonds. These cleaning processes are advantageous for low temperature silicon epitaxy, reliable contact formation, and thin gate oxide growth.

Removal of adsorbed organic contaminants from Si surfaces by reaction with molecular O3 and photo-generated atomic oxygen species in a UHV-compatible photochemical reactor was investigated. Treatment of contaminated wafers with externally generated O3 at 25°C was effective in removing adsorbed organics, but surface cleaning rates were enhanced by simultaneous 254-nm UV irradiation of the reactor contents. In situ photo-generation of O3 and atomic oxygen species by 185- and 254-nm irradiation of O2 gave comparable results. A simplified gas-phase kinetics model describing O3 generation by a low-pressure Hg lamp was developed and used to gain insight into the effects of relative humidity and O2 partial pressure on steady-state O3 concentrations.

An investigation of the parameters in H-plasma cleaning influencing H-diffusion and surface etching is described. The Si surface and subsurface regions were characterized with Raman spectroscopy and high resolution transmission electron microscopy (HRTEM), and the plasma parameters were monitored with a double Langmuir probe and optical emission spectroscopy. The parameters varied in the rf plasma cleaning were the substrate temperature, the rf power, and the plasma exposure time. It was found that low pressure and low power H-plasma exposure was effective in terms of cleaning the surface. In the 300°C, 20 Watts, 2 min. H-plasma exposure case, the Raman spectra indicated that there was no detectable H incorporation into the Si bulk and HRTEM showed no obvious defect microstructure in the near surface region and that the surface was smooth. In contrast, in the 150°C, 50 Watts, 60 min. H-plasma exposure case, HRTEM indicated that H-induced platelet defects formed in the Si(100) subsurface region and that the surface morphology was rough. Raman scattering revealed Si-H features at ∼2100 cm−1.

The effects of SiO2 reactive ion etching (RIE) in CHF3 / C2F6 on the surface properties of the underlying Si substrate have been studied by X-ray photoelectron spectroscopy (XPS) and secondary ion mass spectrometry (SIMS) techniques. The observed two major modifications are (i) a ∼50nm thick silicon layer which contains carbon and fluorine and (ii) 2∼3nm thick residue layer composed entirely of carbon, fluorine, oxygen and hydrogen on the silicon surface. The thermal behaviors of attributed peaks for C 1s, Si 2p, O 1s and F 1s of residue film have been analyzed after in-situ resistive anneal under ultra high vacuum (UHV) condition. C-F1, C-F2 and C-F3 bonds decompose and form C-CFx (x≤3) bonds above 200°C. Above 400°C, C-CFx bonds also decompose to C-C/H bonds. For recovery of the modified silicon surface, reactive ion etched specimens have been exposed to an oxygen plasma. By XPS analysis, the effect of an O2 plasma treatment has been revealed to be completed within 20min. With an O2 plasma pre-treated, a rapid thermal anneal (RTA) treatment as low as 500°2 is found to be effective for removal of impurities in the silicon.

BPSG films show good flow characteristics with the increase of boron and phosphorus concentration. As–deposited BPSG films with high boron and phosphorus concentration, however, have a dopant instability which results in film bursts, cracks, and precipitates, due to water absorption. Elimination of dopant instabilities of BPSG films by N2O plasma treatment is investigated. Film structure, composition and flow properties are evaluated by FT-IR, SIMS, and SEM, respectively. Dopant stability is examined after H2SO4 boiling. After H2SO4 boiling without N2O plasma treatment, BPSG films with high boron and phosphorus concentration show poor flow characteristics. Boron and phosphorus are leached out of the BPSG films by reaction with absorbed water during the H2SO4 boil. N2O plasma treatment results in a thin surface layer composed of Si-O-B bonds. These bonds do not dissolve in water. It is proposed that the thin layer consisting of Si-O-B bonds prevents the penetration of water into BPSG film during H2SO4 boiling.

We report on experiments in which the picosecond ultrasonics technique is used to detect the existence of thin layers of CFx at the interface between a metal film (Al) and a silicon substrate. Acoustic vibrations are excited in the metal film when a picosecond light pulse is absorbed. The thickness of the CFx can be estimated from the rate at which these vibrations are damped out via transmission of sound into the silicon substrate through the CFx layer. We show that thin CFx layers can also be detected via their effect on the rate at which heat flows from the metal to the silicon.

The surfaces of Si(100) and Ge(100), cleaned at low temperatures by H-plasma, were studied by Angle Resolved UV-Photoemission Spectroscopy(ARUPS). In the case of Si(100), H-plasma exposure produced ordered H-terminated crystallographic structures with either a 2×1 or 1×1 LEED pattern while for Ge(100) a weak H-terminated 2x1 pattern was obtained. The hydride phases, found on the surfaces of the cleaned Si(100), were shown to depend on the temperature of the surface during H-plasma cleaning. The electronic states for the monohydride and dihydride phases were identified by ARUPS. When the plasma cleaned surface was annealed, the phase transition from the dihydride to monohydride was observed. For the Ge(100) surface, an ordered 2x1 monohydride phase was obtained from the surface cleaned at 180°C. After plasma exposure at ≤170°C a 1×1 surface was observed, but the ARUPS indicated that the surface was predominantly composed of disordered monohydride structures. After annealing above the H-dissociation temperatures, the dangling bond surface states were identified for both Si and Ge. The Si(100) and Ge(100) ARUPS spectra shifted after annealing, indicating that the H-terminated surfaces were unpinned.

Microwave plasma surface cleaning opens an interesting possibility for thin layer preparation. We have investigated the effect of hydrogen and NF3 plasma induced by a microwave source. The sample is exposed in the downstream region achieved by a plenum. The cleaning process has been applied to a silicon (100) substrate. The cleaned surface has been analyzed by different methods of investigation: HRTEM, XPS, Therma-Wave and MOS capacitor (C-V) characteristics. The HRTEM observations show a defect-free interface, between the hydrogen cleaned silicon surface and the thermal oxide grown immediately after cleaning. In addition, the silicon crystalline network is not perturbed by induced defects. This was confirmed by Thermal Wave measurements which gave an excellent surface qualification.

Silicon wafers as obtained from the manufacturer are immersed in a large area argon/hydrogen plasma for surface cleaning. The plasma discharge is maintained between a heated cathode and the grounded process chamber at discharge voltages of about 30 V for which discharge currents up to 100 A can be chosen. In this regime, neither the chamber walls nor the substrates are sputtered. Chemical reactions at the wafer surface are assumed to be mainly stimulated by low energy electron bombardment. The etch rates for diamond-like carbon (DLC) on silicon wafers were determined for selected discharge parameters and compared with the previously obtained results for SiO2. It was found that 5 minutes in-situ cleaning prepares the silicon wafers for homoepitaxy at 500 °C and higher substrate temperatures, whereas a short anneal at 500 °C was necessary to obtain low temperature homoepitaxy at 300 °C on (100) and 400 °C on (111) silicon. This seems to be due to hydrogen passivation of silicon during the cleaning procedure.

Air and UV/ozone exposure have been used for growing sacrificial thin oxide layers on HCl etched GaAs (100) surfaces. Passive films have been then desorbed by vacuum thermal cleaning in order to prepare suitable surfaces for molecular beam epitaxy (MBE) growth. Structure, thickness and chemical composition of passive films and desorbed surfaces have been studied by ellipsometry and angle dependent X-ray photoelectron spectroscopy (ADXPS). The results have indicated an increased chemical reactivity of both arsenic and gallium during UV/ozone exposure, compared to that of air exposure, that produces Ga (III) oxide enriched films. Furthermore, XPS results have also shown that the thermal desorption behavior are different. In particular, GaAs (100) after short term UV/ozone exposure and oxide film desorption, has an As/Ga surface atomic ratio close to unity and a level of carbon contamination below the XPS detectability. On the contrary, air exposed surfaces never have a stoichiometric composition and carbon is not completely removed by a vacuum heating.

A laser/microwave method using two lasers of different wavelengths for carrier injection is proposed to evaluate near surface regions. These lasers are a He-Ne (wavelength=633nm, penetration depth=∼3μm) and a YAG lasers (wavelength=1060nm, penetration depth=∼500μm). Using a microwave probe, the amount of injected excess carriers can be detected. The carrier concentration is mainly dependent on the condition of the surface when carriers are excited by the He-Ne laser. It is mainly dependent on the condition of the bulk region when carriers are excited by YAG laser. We refer to microwave intensities detected by the He-Ne and the YAG lasers as the surface-recombination-velocity-related microwave intensity (SRMI) and bulkrelated microwave intensity (BRMI), respectively. We refer to the difference between SRMI and BRMI as relative SRMI (R-SRNI), which is closely related to the condition of surface and surface active region. We evaluate the near surface regions of the samples after plasma and wet etching for removing the photoresist layer. And we evaluate the near surface regions of the samples after a heat treatment which is done to recover the damage introduced by plasma etching. It is found that the R-SRMI method is better suited to near surface region evaluation than conventional lifetime measurements.

We report for the first time on the application of Field Enhanced Plasma Oxidation (FEPO) to the passivation of Hg1-xCdxTe. Oxide layers of thickness 250-500Å were grown on Hg0.4Cd0.6Te at temperatures in the range from 30°C to 100°C. The index of refraction of the oxide layer, as determined by ellipsometry, was between 1.99 and 2.10. Auger depth profiles of the surfaces showed a compositional variation through the oxide with predominantly a mix of Cd and Te oxides in a film depleted in Hg. The oxides had dielectric constants of 3.6 to 4.5, and dielectric strength greater than 1 × 107 V/cm. The oxide/semiconductor interfaces had high densities of “slow” states but were capable of being depleted and accumulated.

The etching of CdZnTe substrates by bromine-methanol solutions was studied by chemi-mechanical polishing. Over a bromine concentration of 0.5 to 3.5 volume percent, the etch rates varied linearly with bromine concentration, were independent of substrate orientation, and, within experimental uncertainties, were identical to those for the etching of CdTe. The etched samples were used as substrates for the organometallic vapor phase epitaxy (OMVPE) growth of CdTe in an impinging jet reactor. The bromine concentration used in etching did not affect the growth rate or the morphology of the deposited CdTe films, but a higher defect density was found for the film grown on a substrate with the orange-peel structure.

Chemical treatment of GaAs with sulfur-containing compounds has been shown to improve GaAs surface electronic properties. There is still considerable controversy, however, regarding the chemical nature of the surface film which results from the sulfidation, and of the basis of the electronic improvement and of the decay in the improved electronic properties with time. We have investigated the surface chemistry of the chemical sulfidation treatment of GaAs with Na2S.9H2O and (NH4)2S. Using surface infrared reflection spectroscopy (SIRS) and x-ray photoelectron spectroscopy (XPS), we have studied the GaAs surface and its behavior with time after such treatments. Results show that both of these sulfidation treatments removed the chemical oxide of GaAs, leaving behind a thin film on the surface. XPS results show that the Ga-O and As-O peaks were removed after treatment and that As-S and no Ga-S peaks were formed. Infrared results show that the film deposited after Na2S.9H2O treatment slowly reacted in air to form sodium carbonate and rhombic sulfur. In addition, this film contains compounds with sulfur-oxygen bonds, which most likely were arsenic sulfate, sulfite, and thiosulfate salts. The film deposited on the (NH4)2Streated GaAs surface was identified as ammonium thiosulfate and slowly decomposed with time. Rinsing with water removed the thin film formed after either sulfidation treatment.

Two techniques were used to apply a thin layer of selenium to GaAs (100) surfaces: (a) the thermal decomposition of hydrogen selenide at 120°C, and (b) the deposition of gaseous selenium formed in the reaction of hydrogen atoms with H2Se. Angle dependent XPS measurements show that AsxSey, but not GaxSey, forms on the surface in both cases. This work therefore establishes a surface behavior for selenium on GaAs(100) which contrasts with that described earlier for sulfur on this surface.