Hisao Kojima

Passivation effect of silicon nitride against copper diffusion

The use of Cu in ultralarge scale integrated (ULSI) conductors has resulted in the need to prevent Cu diffusion. We evaluated the passivation effect of plasma-enhanced chemical-vapor-deposited silicon nitride (PECVD-SiN) using secondary ion mass spectrometry and atomic absorption spectrometry. From these measurements, it was found that a large amount of Cu diffused through PECVD-SiN films during the heat treatments of the metallization process, probably due to the rapid diffusion paths along the microdefects of PECVD-SiN films. However, Cu contamination was barely detected in the current–voltage measurements and bias-temperature stressing tests of Cu/PECVD-SiN/SiO2/Si capacitors because the leakage current through SiN films slightly increased as a result of Cu diffusion. This result is attributed to the electric-field relaxation caused by a large number of electrons trapped in the PECVD-SiN films, of which the negative charge compensates the positive charge of Cu ions. Although the degradation of electric...

X‐ray photoelectron spectroscopy study of submonolayer native oxides on HF‐treated Si surfaces

This article investigates, by a novel spectral analysis technique, the possibility of applying conventional x‐ray photoelectron spectroscopy to characterize the initial oxidation process. Quantitative analysis of submonolayer native oxides is made possible by spectral decomposition of Si 2p into Si4+, Six+, and Si0+ based on the O 1s binding energy, and by calculation of the SiO2 and SiOx thicknesses using the decomposition results. Here, the sensitivity for oxide‐thickness change is about 1/10 monolayer. Using this technique, the initial oxidation processes of HF‐treated Si(111) and Si(100) are precisely characterized. The influence of the dissolved‐oxygen concentration in the HF solution is also investigated.

Study on solvent-loading effect on inductively coupled plasma and microwave-induced plasma sources with a microliter nebulizer

Abstract To investigate the solvent-loading effect, a microliter nebulizer, called a sonic spray nebulizer (SSN), was used to introduce a sample solution at the microliter per minute level into an inductively coupled plasma (ICP) and into a nitrogen microwave-induced plasma (MIP). We compared the ICP to the MIP, and the SSN to a conventional concentric nebulizer (CCN). Our results confirm that a certain amount of water solvent is useful for analyte excitation in an Ar-ICP, since hydrogen released from the water enhances the energy transfer inside the plasma. However, we found the dominant water–solvent effect in the N2-MIP source was to consume MIP energy, rather than to promote the analyte excitation. Therefore, a means of solvent removal should be used with a N2-MIP. We also discuss other differences between the ICP and the MIP. With the SSN, several organic solvents were successfully introduced into the ICP without degrading the plasma stability.

Phosphosilicate Glass Passivation for ULSI Cu Metallization

Phosphosilicate glass (PSG) passivation for Cu metallization is proposed, and the Cu diffusion rate into Si substrates through various oxide structures is measured quantitatively. Cu diffusion is suppressed effectively by the PSG passivation resulting in reduction of the diffused Cu atoms to about an order of magnitude less than when passivated by nondoped SiO 2 films. The passivation is effective when the annealing temperature is less than 600 o C. The apparent activation energy of Cu diffusion for the nondoped SiO 2 /Si substrate and PSG/nondoped SiO 2 /Si is 1.2 and 1.6 eV, respectively

Influence of ion-implantation on native oxidation of Si in a clean-room atmosphere

Abstract We investigated the native oxidation of ion-implanted Si(100) surfaces in anticipation of a future necessity for controlling native oxidation during semiconductor device fabrication. Quantitative analysis of XPS spectra was used to estimate SiO2 and SiOx thicknesses. Native oxidation of Si(100) wafers in which were implanted with As, P, B or Si was examined. The results show that oxidation of As or P-implanted Si is much faster than that of Si without implantation, however, no conclusive difference was found between the oxidation rates of B or Si-implanted Si and that of Si without implantation. These results indicate that native oxidation is influenced mainly by the species of implanted ions. Neither ion-implantation induced defects nor surface roughness was found to have major effect on the native oxidation rate.

X‐ray photoelectron spectroscopy study on native oxidation of As‐implanted Si (100)

The native oxidation of As‐implanted Si surfaces used in actual ultralarge scale integrated processes are investigated. Quantitative analysis of oxidation is made possible by x‐ray photoelectron spectroscopy spectral decomposition of Si 2p into Si4+, Six+, and Si0+, and by calculation of SiO2 and SiOx thicknesses using the decomposition results. Here, the sensitivity is such that less than 1 A change is detectable. The native oxidation of As‐implanted (1×1015–1×1016/cm2, at 25 kV) Si(100) is compared to that of Si(100) without implantation. The results show that the oxidation rate of As‐implanted Si is faster than that of Si without implantation, that the native oxide on As‐implanted Si includes more suboxide than that on Si without implantation, and that As is oxidized in deeper regions than Si. These results indicate that As implantation changes the Si native oxidation mechanism itself. We propose an oxidation model of As‐implanted Si to explain our observations, and discuss the validity of this model.

Improved phosphosilicate glass passivation against Cu contamination using the rapid thermal annealing process

Passivation against Cu contamination is a key technology for realizing multilevel Cu interconnection for ultralarge scale integrated circuits (ULSI). Phosphosilicate glass (PSG) with rapid thermal annealing (RTA) is found to provide no less of a passivation effect than furnace-annealed (FA) PSG films, while significantly reducing the thermal budget of the PSG annealing from 900 o C×20 min (FA) to 900 o C×60 s (RTA). This suggests that the PSG-RTA process is applicable to 0.25 μm level metal oxide semiconductor (MOS) ULSIs, since both the amount of Cu diffusion and the thermal budget are within the limits of 0.25 μm MOS device technology

Ultra-Shallow Depth Profiling of Arsenic Implants in Silicon by Hydride Generation-Inductively Coupled Plasma Atomic Emission Spectrometry

High resolution depth profiling of arsenic (As) implanted into silicon wafers by a chemical technique is described. Silicon wafers are precisely etched through repeated oxidation by hydrogen peroxide solution and dissolution of the oxide by hydrofluoric acid solution. The etched silicon thickness is determined by inductively-coupled plasma atomic emission spectrometry (ICP-AES). Arsenic concentration is determined by hydride generation ICP-AES (HG-ICP-AES) with prereduction using potassium iodide. The detection limit of As in a 4-inch silicon wafer is 2.4×1018 atoms/cm3. The etched silicon thickness is controlled to less than 4±2 atomic layers. Depth profiling of an ultra-shallow As diffusion layer with the proposed method shows good agreement with profiling using the four-probe method or secondary ion mass spectrometry.

The Increase of the Native Oxide Thickness on H-Terminated Si Surfaces by Gaseous Contamination in a Clean Room Atmosphere

Gaseous impurities in a clean room (CR) atmosphere cause an increase in the native oxide thickness on H-terminated Si surfaces. The increase effect of ammonia is greater than other impurities. The increase mechanisms can be explained by a three-stage model. Ammonia generates an OH- group by reacting with H 2 O. The OH-group etches off H-terminated Si atoms and the effectiveness of the H-termination on reducing the native oxide growth is lost. As a result, the native oxide thickness increases. The native oxide growth can be suppressed by reducing the level of alkaline contamination in the CR atmosphere.

A Ceramic Plasma Torch for Determining Silicon Content in HF Solutions by Inductively Coupled Plasma Atomic Emission Spectrometry

A ceramic plasma torch has been developed for plasma spectrometers–such as an inductively coupled plasma atomic emission spectrometer (ICP-AES) or an inductively coupled plasma mass spectrometer (ICP-MS)–used to measure trace amounts of Si in HF solutions. Pyrolytic-boron nitride (PBN) and alumina are studied as ceramic torch materials. The PBN torch can be used when the HF concentration is low; however, it cannot be used when the HF concentration is high since contamination of B due to thermal decomposition is significant. When the alumina torch is used, the background spectrum of Si can be decreased to 1/65 of that obtained when using a conventional torch–this increment corresponds to less than 0.2 µg/m l of Si and a 1-µg/m l Si solution can be determined with a relative standard deviation of about 8% even at 10% (v/v) HF concentration.