2001Journal of Vacuum Science & Technology B Microelectronics and Nanometer Structures Processing Measurement and PhenomenaRequires access

Effect of thinning a WSiN/WSix barrier layer on its barrier capability

Akihiko Hirata, Tomohiko Maeda, Katsuyuki Machida, Masahiko Maeda, Kanji Yasui, Hakaru Kyuragi

Open publisher page 4 citations

Abstract

We investigated the effects of tungsten silicon nitride/tungsten silicide (WSiN/WSix) barrier layer thickness on its barrier capability. WSiN was obtained by nitridizing the WSix surface with electron cyclotron resonance nitrogen plasma. The total thickness of the WSiN/WSix barrier layer was reduced by thinning the initial WSix layer. When 5-nm-thick WSix was nitridized, the N and Si contents in the WSiN/WSix layer became smaller than when WSix initial thickness was 20 nm. This barrier layer diffused into the copper (Cu) layer when annealed, and did not act as a barrier layer. On the contrary, a WSiN/WSix barrier layer formed by nitridizing 10-nm-thick WSix showed good barrier capability against Cu diffusion. We evaluated the leakage current between Cu damascene interconnections with this barrier layer and found that this barrier layer formed on the trench side wall prevents Cu diffusion when the thickness on the side wall is over 10 nm.

About this research paper

What this paper is about

We investigated the effects of tungsten silicon nitride/tungsten silicide (WSiN/WSix) barrier layer thickness on its barrier capability. WSiN was obtained by nitridizing the WSix surface with electron cyclotron resonance nitrogen plasma. The total thickness of the WSiN/WSix barrier layer was reduced by thinning the initial WSix layer. When 5-nm-thick WSix was nitridized, the N and Si contents in the WSiN/WSix layer became smaller than when WSix initial thickness was 20 nm. This barrier layer diffused into the copper (Cu) layer when annealed, and did not act as a barrier layer. On the contrary, a WSiN/WSix barrier layer formed by nitridizing 10-nm-thick WSix showed good barrier capability against Cu diffusion. We evaluated the leakage current between Cu damascene interconnections with this barrier layer and found that this barrier layer formed on the trench side wall prevents Cu diffusion when the thickness on the side wall is over 10 nm.

Why it matters

OpenAlex reports 4 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

We investigated the effects of tungsten silicon nitride/tungsten silicide (WSiN/WSix) barrier layer thickness on its barrier capability. WSiN was obtained by nitridizing the WSix surface with electron cyclotron resonance nitrogen plasma. The total thickness of the WSiN/WSix barrier layer was reduced by thinning the initial WSix layer. When 5-nm-thick WSix was nitridized, the N and Si contents in the WSiN/WSix layer became smaller than when WSix initial thickness was 20 nm. This barrier layer diffused into the copper (Cu) layer when annealed, and did not act as a barrier layer. On the contrary, a WSiN/WSix barrier layer formed by nitridizing 10-nm-thick WSix showed good barrier capability against Cu diffusion. We evaluated the leakage current between Cu damascene interconnections with this barrier layer and found that this barrier layer formed on the trench side wall prevents Cu diffusion when the thickness on the side wall is over 10 nm.

Key concepts: Diffusion barrier, Barrier layer, Materials science, Layer (electronics), Tungsten, Silicide, Composite material, Metallurgy

Related papers

Back to paper searchBrowse research topicsOriginal source
Effect of thinning a WSiN/WSix barrier layer on its barrier capability — Research Paper | ScholarLens