Page 43 - Semiconductor Manufacturing Handbook

Geng(SMH)_CH04.qxd 04/04/2005 19:36 Page 4.6

COPPER, LOW-k DIELECTRICS, AND THEIR RELIABILITY

4.6 SEMICONDUCTOR FUNDAMENTALS AND BASIC MATERIALS

ALCVD contributes several advantages. The thickness of the films can be controlled in a straight-
forward manner by controlling the number of reaction cycles, therefore enabling the controlled
growth of ultrathin layers. New seedless techniques have also been developed. Metals like ruthenium
and rhodium can be used both as a diffusion barrier and seed layer for copper metallization. 13
Chemical mechanical polishing is done to remove the copper overburden post-electrochemical
deposition and to achieve planarization. In this method the wafer is mechanically abraded in a slurry
containing abrasives and etching agents. The process combines chemical reactions and mechanical
forces to selectively remove unwanted material. In addition to copper, liners are also removed using
this method. Tools have been developed to remove both copper and liners in the same process mod-
ule, thus increasing the throughput.
Across wafer uniformity, pattern factor sensitivity and chemical (corrosion, copper and liner
residuals, particulates, moisture absorption) and mechanical (delaminating, peeling, and cracking)
damage to the surface are some of the key issues associated with this methodology. They can impact
both yield and reliability of the product. Advanced solutions like e-CMP (similar to electropolish-
ing) with low down force have been introduced. Such solutions are required especially for technolo-
gies with finer line geometries and mechanically weak low-k dielectrics. 14
Often integration routes use a permanent hardmask to protect the dielectric surface during CMP.
It has been shown that the CMP slurry can damage the surface of the dielectric leading to high line
to line leakage. This also adds to the effective dielectric constant and reduces the performance.
Silicon nitride and oxide-based materials are typically used for this purpose.

4.3 LOW-k DIELECTRICS TECHNOLOGY

The capacitive component of the RC delay can be reduced by incorporating lower dielectric constant
(low-k) insulators in the back end. The ratio of the increased capacitance to the vacuum capacitance
due to polarization of the dielectric is known as the dielectric constant. Polarization at the molecu-
lar level is made up of three components—electronic, atomic, and orientational. Electronic polariza-
tion is due to the displacement of the electron cloud with respect to the positive nucleus under an
external electric field. It can react to very high frequencies and is responsible for the refraction of
light. Atomic polarization is due to the distortion of the atomic nuclei arrangement in a molecule or
a lattice. This movement of heavy nuclei is more sluggish than electrons so that atomic polarization
cannot occur at as high frequencies as electronic polarization. Hence typically it is not observed
above infrared frequencies. If the molecules of the material already possess a permanent dipole
moment, there is a tendency for these to be aligned by the applied field to give a net polarization in
that direction. This gives rise to orientational polarization. This component can make a large contri-
bution to the total polarization in an applied field but may be slow to develop.
Although copper containing chips were introduced in 1998 with silicon dioxide insulators, the
lowering of insulator dielectric constant predicted by International Technology Roadmap for
Semiconductors (ITRS) has been problematic. Several chemistries were evaluated to find the low-k
solution for interconnect technology. Oxide-based dielectrics like fluorosilicate glass (FSG) and
fluorine-doped tetra-ethyl-ortho silicate (FTEOS) were among the first set of materials studied for
15
low-k solutions. The introduction of FSG (k = 3.7) at the 180-nm technology node in conjunction
with the copper dual-damascene process represented the first major break in the historical extension
of silicon dioxide (k = 4.1). Fluorine being the highest electronegative element reduces the electronic
polarization and thus achieves lower dielectric constant. Issues like fluorine mobility, reactivity with
refractory metal-based liners, and adhesion and moisture sensitivity had to be resolved before it
could become a manufacturing process. Noguchi et al. have shown the degradation of TDDB life
16
with increasing delay between CMP and cap deposition. FTEOS is another form of fluorine-doped
oxide that is being used as the back-end dielectric. Materials like organosilicate glass (OSG) and car-
bon doped oxide (SiCOH) offer k values in the range of 2.9 to 3.2. Several polymers with k values
17
between 2.5 and 2.9 can also be used as low-k dielectrics. SiLK, methyl silsesquioxane (MSQ) and
hydrogen silsesquioxane (HSQ) are some polymer-based low-k dielectrics. Ultra-low-k dielectrics

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.