The structure and configuration of threading dislocations (TDs) in AlN films grown on (0001) sapphire by metal–organic vapor phase epitaxy (MOVPE) were characterized by high-resolution transmission electron microscopy (HRTEM). It was found that the TDs formed in the films were mainly the perfect edge dislocations with the Burgers vector of b = ⅓〈11¯20〉. The majority of the edge TDs were not randomly formed but densely arranged in lines. The arrays of the edge TDs were mainly observed on the {11¯20} and {10¯10} planes. These two planes showed different configurations of TDs. TD arrays on both of these planes constituted low-angle boundaries. We suggest that these TDs are introduced to compensate for slight misorientations between the subgrains during the film growth.

Initial nucleation and succeeding coalescence in supercritical fluid deposition (SCFD) of Cu was studied in situ by using our devised monitoring technique, i.e. surface reflectivity measurement of visible white light. Fabrication of 10nm-thick smooth and continuous Cu film required by ULSI metallization, was succeeded. Complete filling without any seems and voids into various via patterns of 50 - 200 nm in diameter and 1 μm in depth was also achieved, which was revealed through angled polishing of patterned substrate.

High-temperature silicon dioxide chemical vapor deposition (HTOCVD), using SiH2Cl2 and N2O, can realize dense and conformal oxide film, not only on large size silicon wafers, but even inside of microscopic silicon trenches, at high-temperature around 800°C.

In this work, we investigated the kinetics of HTOCVD using a commercial scale low pressure (LP) CVD reactor, focusing on the correlation between deposition rate and surface-to-volume ratio (S/V ratio), which is a specific surface area of substrate wafer divided by the space volume between two adjacent wafers. We also investigated the deposition rate profile on wafers, and along the axial direction of the reactor near the region where one, two or three substrate wafers are extracted from the quartz holder. The deposition rate profiles on wafer characteristically change from skillet-like to pancake-like, according to the increase of wafer spacing. The influence on the deposition rate of wafer spacing spreads to ranges not only downstream, but also upstream in the gas flow. These experimental results strongly suggest that in the HTOCVD gas-phase reactions through intermediate states of active species contribute to deposition reaction as well as direct deposition reaction of source gases on Si surface.

Thin and continuous CVD Cu seed layer was successfully deposited on Ru under-layer by Cu oxide deposition and reduction method at 100°C with novel chemistry. Cu oxide was formed with Cu(hface)TMVS and H2O2 at 100°C, and this film was reduced with formic acid at 100°C. Deposited Cu oxide films were Cu2O that was confirmed by XRD and XPS. The morphology of oxide films showed smooth and continuous on Ru and Ta substrate. The reduced Cu film on Ru maintained good surface morphology, and no impurity was detected not only in the Cu film but also the interface between Cu and Ru. However, that on Ta had poor morphology by agglomeration of Cu film during reduction due to poor Cu wettability on oxidized Ta that was oxidized during oxide deposition. The readiness of reduction is very important merit of using Ru under-layer for this oxide deposition and reduction process. The oxide deposition and reduction method on Ru under-layer can be a promising candidate for thin and continuous seed layer deposition method.

We study new deposition chemistry in Cu-SCFD (supercritical fluid deposition), especially effect of acetone is investigated as reducing agent and additives into H2 reducing Cu-SCFD. Acetone reduction yields Cu film deposition onto Ru coated Si substrate, because Ru has catalytic effect to generate reducing agent from acetone. When acetone is added into H2 reducing chemistry, the solubility of precursor is enhanced approximately 50 %, which is so-called “entrainer effect”. Ethanol can enhance the deposition of Cu in H2 reduction chemistry, which is a solvent effect. Two effects of entrainer effect by acetone and solvent effect by ethanol works independently, without interfering each other.

Thickness uniformities of poly-crystalline silicon thin films, deposited by a commercial LPCVD reactor, were investigated through a phenomenological and elementary reaction analysis. To understand the deposition rate and its profile in radius direction of ø 6”h silicon wafer, concentration distributions of film precursors were examined by solving basic diffusion equations of film precursors in the CVD system. The experimental thickness distribution can be simulated very well with the solution by optimizing η, the sticking probabilities of the precursors. While most of the silicon deposition was made by source precursor (mono-silane, SiH4), two kinds of intermediate species with sticking probabilities of 5X10-2 and 7X10-4 were found to contribute the deposition. Subsequently, elementary chemical reaction analysis of poly-crystalline silicon CVD process was performed using ChemKinTM and two chemical species, SiH2 and Si2H6, were identified as the possible candidates for the intermediate species.

ZrO2 films were investigated as high-k gate insulators for possible gate dielectric applications. These thin films were deposited by chemical vapor deposition (CVD). We had investigated kinetics of ZrO2-CVD reaction system and thermal stability of ZrO2 film deposited from zirconium-tetra-tertiary-butoxide (ZTB) and tetrakis-diethylamio-zirconium (TDEAZ) as Zr source. In the case of ZTB, ZTB decomposed into two active intermediate precursors by gas phase reaction. These intermediate precursors also contributed to the film growth. The step coverage, the contents of carbon and oxygen, and film structures depended on the growth precursor contributed to the film growth. Film structures also depended on the deposition temperature. As the films subjected to the rapid thermal oxidation annealing, the film structure changed amorphous into cubic and then into monoclinic. However, in the case of the films deposited at 300°C, the film structure kept amorphous state after 900°C annealing. On the other hand, in the case of TDEAZ, TDEAZ directly deposited on the growing surface and the decomposition of TDEAZ by gas phase reaction did not exist. The films deposited by TDEAZ contained more carbon than by ZTB. The film structure kept amorphous state independent of the deposition temperature. As the films subjected to the rapid thermal oxidation annealing, the film structure changed amorphous into monoclinic and then into amorphous again. This structural change resulted from the decrease of the impurity contents of the residual nitrogen and the residual carbon and the increase of the content of silicon diffused from the substrate in the ZrO2 films.

TiN films deposited by CVD utilizing TiCl4 and NH3 are widely used in ULSI processes. The residual chlorine in the film is one of the major issues of TiCl4 based chemistry [1–5]. We have examined the kinetics of TiN-CVD using TiCl4 and NH3 as precursors to determine method for controlling the residual chlorine concentration.

We analyzed the deposition rate profiles obtained from a simple tubular isothermal reactor. Deposition was performed over the temperature range from 300°C to 500°C and with varying initial concentrations of TiCl4 and NH3. We found that the deposition rate was independent of the TiCl4 concentration at higher concentrations. The deposition rate was proportional to the TiCl4 concentration in the low concentration range. This phenomenon can be well explained by a Langmuir-Hinshelwood type mechanism. Examination of the deposition rate dependence on the concentration of NH3 showed that both the maximum deposition rate at high TiCl4 concentrations and the first order kinetic rate constant in the lower TiCl4 concentration range were proportional to the NH3 concentration. These results indicate that the deposition was controlled by the Eley-Rideal mechanism. The residual chlorine concentration was found to be proportional to the surface coverage factor of adsorbed TiCl4. We conclude that the best way to obtain good quality TiN films can be obtained theoretically through these kinetic studies.