Cu was deposited onto Si(100) wafers from 1% HF solutions containing CuCl2. Cu K edge glancing-angle x-ray fluorescence (GAXF) measurements were used to determine surface concentrations and to elucidate the chemical state and morphology of the deposits. Total reflection x-ray fluorescence (TXRF) measurements were used to calibrate edge jumps to absolute surface concentrations. X-ray absorption near-edge structure (XANES) spectra indicated that the Cu deposits were metallic at concentrations of 1014 - 1016 atoms cm−2, and extended x-ray absorption fine structure (EXAFS) spectra evidenced that the Cu was present as three-dimensional fcc crystallites. Cu0 deposition from HF solutions is inferred to proceed by an electroless process, that is made more favorable by complexation of Si4+ by F. Contaminated wafers were stored in air and the effects of Cu on regrowth of Si native oxide were investigated using Auger electron spectroscopy (AES). The induction period for native oxide regrowth on unrinsed wafers markedly decreased when Cu microcontamination was present, consistent with a catalytic effect. Rinsed wafers had higher initial oxide coverages but were oxidized more slowly in air than unrinsed wafers.

X-ray photoelectron spectroscopy (XPS) measurements on silicon surfaces cleaned by HF/Ethanol treatments are described using silicon wafers with (100) vicinal surfaces. The composition of the surfaces (Silicon, Oxygen, Fluorine and Carbon) has been measured leading to the conclusion of good passivated surfaces. Moreover, for each vicinal direction, the F1s, C1s,O1s and Si2p lines have been decomposed into several components corresponding to different chemical bonds.

The more significant results show that the percentages of the two components for Fluorine depend on the total amount of Oxygen and are related to the Si-F and Si-O-F bonds. On the other hand, the two complementary doublets necessary to fit the Si2p band can be related to the Si-H and Si-F bonds and change with orientation and/or roughness of the considered surface.

The native oxide formation also points out significant differences in the behaviour of each vicinal direction. The thickness and structure of the oxide and the remaining amount of Fluorine reflect the influence of the initial RMS-microroughness determined by Atomic Force Microscopy which changes from 0.05nm for (100) to 0. lnm for (911).

Corrosion of the Si(100) surface by concentrated NH4F solutions was studied using high spatial resolution secondary electron imaging in an ultra-high vacuum scanning transmission electron microscope. Various corrosion mechanisms were investigated by varying the ex-situ chemical treatment of the samples. Entirely different surface morphologies were obtained for surfaces that were kept in solution for short times (-minutes) vs. long times (∼24 hours) and surfaces which were rinsed in DI H2O vs. those which weren't. These measurements confirm that corrosion continues after the samples are removed from the solution and seems to be correlated with the formation of the Si hexafluorometallate salt. This salt is extremely electron beam sensitive and desorbs in UHV at temperatures below 250°C.

The use of hot water immersion to aid in the drying of wafers subsequent to liquid chemical processing has recently been increasing. Both the TREBOR HydrodryTM and the Isopropyl Alcohol (IPA) dry used by CFM Technologies Inc. use water rinses at elevated temperature. These treatments are of particular concern when they follow HF immersions where surface oxides are removed and bare silicon surfaces are exposed to the hot water. Studies by Watanabe et al.1 show that Si(l 11) facets can be produced by hot water immersion. This, in turn, implies that hot water etches silicon with an anisotropic etch rate. In this work, the etching of oxide patterned Si(100) wafers by immersion in hot water is studied. Etching is clearly observed in images produced by Transmission Electron Microscopy (TEM) with a concomitant increase in surface roughness. The rate of etching varies dramatically with the dissolved oxygen concentration of the water used. Possible mechanisms to explain the etching will be discussed.

The scanning tunneling microscopy (STM) images for samples treated by the NH4F (pH=8) solutions exhibited flat terraces and monolayer steps directed toward <112> directions. In addition, there exist two types of surface structures on the same surface: nearly flat parts consisting of terraces with 20-60 nm in width; step-bunched parts consisting of terraces with 1.0-1.5 nm in width which are close to (221) or (332) facets, respectively. The equilibrium width for the large terrace is determined by the competition of two etching processes with weakly-alkaline solutions (pH = 8 ): etching from the step edges and etching on the terrace. On the other hand, the narrow terrace with a few atomic distance is mainly etched from the step edges. In this case, the equilibrium shape is determined by the steric hindrance effect in addition to the anisotropic etching effect which makes the steps uniform.

Atomic step structure and chemical termination of Si(lll) surfaces after wet chemical treatments which precede the various oxidation, deposition or growth processes for the patterning of microelectronic devices—i.e. various cleaning, HF or NH4F etching, and H2O rinsing steps—are investigated by scanning tunneling microscopy (STM), Auger electron spectroscopy (AES) and thermodesorption spectroscopy (TDS) after sample transfer into ultrahigh vacuum: Concentrated HF selectively removes surface oxide and, thus, chemically prepares the initially burried, isotropically rough Si/SiO2 interface. A surface termination by mono–, di–, and trihydrides results from the differently coordinated interface defect sites and can be distinguished by distinct characteristic H2 desorption peaks in TDS. NH4F, on the contrary, attacks the bulk silicon anisotropically to form large, ideal monohydrideterminated (111) terraces with triangular etch pits on precisely (111)–oriented samples or to result in a rapid silicon removal by a flow of the preferentially etched atomic steps on samples with an initial miscut. At weak miscut the etching anisotropy dictates a flow of (kinked) steps made up of straight monohydride–type step portions perpendicular to <211> while at strong miscut the miscut–promoted step type survives due to strong stepstep interaction and even a flow of (unstable) dihydride–type steps may be observed if the sample surface is tilted towards <211>. During the experiments the sample cleanliness (carbon–hydrides) is monitored by AES, comparing various wet chemical and gaseous phase ex situ cleaning and etching techniques.

The Si(111) surface treated in NH4F has been investigated using photoemission and surface infrared spectroscopy (SIS). We confirm both with photoemission and SIS measurements that the NH4F-treated Si(111) surface is dominantly terminated with the monohydride Si (Si-H) oriented perpendicular to the surface and is free from silicon oxide. Photoemission and SIS data also demonstrate that ammonium fluorides react with the Si substrate to generate the hexafluorosilicate salt, (NH4)2SiF6. We propose that the formation of (NH4)2SiF6 or SiF62− ions plays an important role in the NH4F etching of Si crystals.