Stress and Adhesion of CVD Copper and TiN
- PDF / 264,432 Bytes
- 4 Pages / 414.72 x 648 pts Page_size
- 108 Downloads / 234 Views
ABSTRACT The use of copper for interconnect metallization on advanced submicron integrated circuits is attractive because of its inherent low resistivity. To date, chemical vapor deposition (CVD) processes which produce copper films with nearly bulk resistivity and good uniformity have been produced. However, adhesion of the copper deposit to underlying barrier materials, in particular to titanium nitride (TiN), has been problematic. In this work, adhesion has been measured using both a qualitative and a semi-quantitative method and has been correlated with process conditions and precursor formulation.
INTRODUCTION In recent years, there has been great interest in the development of copper metallization as a component of very large scale integration (VLSI) since it is an improvement over the standard aluminum or aluminum with tungsten plug metallization schemes currently in use. The reason for this is obvious, with the exception of silver, copper has the lowest resistivity, 1.673 ýtQ cm at 20°C, of any candidate material for the fabrication of interconnects. It also has high electromigration resistance in comparison to aluminum materials.", 2 However, other material properties such as adhesion are not ideal and require significant developmental effort. Copper deposition in this study was carried out in a load locked, single wafer deposition system consisting of an entrance/exit load lock, a vacuum transport chamber, and a process chamber. The precursors used in this study were various formulations of CupraSelectTM available from Schumacher, Inc. Titanium nitride barrier materials were reactively sputter deposited from a pure titanium target in a nitrogen/argon ambient. A more detailed description 3 of the copper deposition process has been presented elsewhere.
EXPERIMENTAL All experiments were carried out using 150 mm [100] silicon substrates prepared with an initial coating of 450 nm of tetraethylorthosilicate (TEOS) based plasma enhanced chemical vapor deposited silicon dioxide. This was followed by sputter deposition of TiN. Various thicknesses of TiN were used in this study, however, the nominal thickness used was 50 nm. Thickness of TiN was found to have no effect on the adhesion or the deposition characteristics of copper films. For stress measurements, initial wafer bow was measured prior to copper deposition. Following the TiN deposition, copper films with a nominal thickness of 300 nm were deposited using the CVD process. Film stress and adhesion were measured on these substrates. Film stress was measured using a Flexus stress gauge. 859 Mat. Res. Soc. Symp. Proc. Vol. 356 01995 Materials Research Society
Two methods were used to evaluate film adhesion. The first is a qualitative tape pull test which consists of scribing the film in a parallel line pattern, attaching adhesive tape to the copper deposit, and pulling the tape off of the deposited film and substrate. Bad adhesion is indicated if the copper film is removed with the tape, conversely, good adhesion is indicated if the copper film remains att
Data Loading...
