Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Introduction of Ru metal in the MOL liner stack -- Replacement of IMP Ti

IP.com Disclosure Number: IPCOM000188723D
Original Publication Date: 2009-Oct-20
Included in the Prior Art Database: 2009-Oct-20
Document File: 2 page(s) / 68K

Publishing Venue

IBM

Abstract

32 nm node requires CDs well below 40 nm. At these dimensions, IMP Ti overhang can severely compromise the W fill and degrade contact resistance. Also, since the TiN is thinner, more fluorine may find its way to the underlying Ti and create opens. Replacing the IMP Ti layer with CVD/ALD Ru metal may eliviate these issues, the Ti overhang and the Ti attack

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Introduction of Ru metal in the MOL liner stack -- Replacement of IMP Ti

Why do we use Ti ?

CA pre-clean (i.e. Ar sputter clean) leaves behind traces of oxygen on the silicide surface at the bottom of the CA. Since Ti is conductive and an excellent oxygen getter, it is used to "trap" the oxygen and provide better electrical contact with the silicide.

What is the problem with Ti ?

Ti is deposited via PVD. PVD is a "line of site" deposition technique that leads to significant overhang for small CAs (> 50 nm). Extensive overhang causes poor fill (key holes) and degrades the final contact resistance. CVD and ALD Ti techniques are not mature enough to replace PVD Ti primarily due to the absence of available precursors and the high deposition temperatures required (~ 500 oC). Ti also violently reacts with fluorine. Any fluorine that slips through the liner (TiN) may react with Ti and create opens. The effect is known as the "volcano effect". As the liner is scaled down, fluorine attack becomes a major reliability concern.

Ruthenium can be used to replace Ti for the following reasons

It can be deposited both by CVD and ALD methods which are not "line of site" like PVD (free of overhang issues)

Ru nucleation on oxides is slower that on metallic surfaces. It is expected that Ru will form a continuous film faster at the bottom of the CA (on the silicide surface) than on the sidewalls (SiOx and SiN) 1,2. This is beneficial for two reasons: a) it creates the necessary electrical co...