1033
Post Salicidation Clean: Removal of Unreacted Pt from High Pt Content NiPt Silicide
Silicides have been used in self-aligned processes for several generations of CMOS devices, with the aim of reducing the sheet resistance and providing stable Ohmic contacts with low contact resistance on gate, source and drain areas.1 To avoid bridging at the spacers and the isolation areas, after salicide formation, unreacted NiPt needs to be removed; simplified process flow is shown in Figure 1. Theoretically, strongly oxidizing wet clean chemistries should be able to remove Ni and/or Pt in the Ni Pt salicide. Xu et al.2 used a HF and HNO3 acid mix to remove unreacted NiPt (for 5% and 10% Pt content) at high etch rates with high selectivity to other materials exposed at the wafer surface.
However, for higher Pt contents in Ni Pt salicide, the HF and HNO3 acid mix was unable to remove all the unreacted Pt (Figure 2). Other strongly oxidizing wet cleans like SPM (sulfuric peroxide mix) and aqua regia were used and found to be inconsistent; Ni Pt residues were found on the wafer surface intermittently. Therefore, we needed to find a stable high volume manufacturing (HVM) process for removing Ni high Pt residues.
Approach
The unreacted Ni high Pt residue was analyzed. Unlike previous generations of Ni-Pt, this residue was almost pure Pt. That meant that Ni was being preferentially dissolved from the Ni high Pt surface eventually resulting in a pure Pt layer. So to prevent the preferential dissolution of Ni in Ni high Pt binary alloys, wafer spin changes were used to add a physical component to the chemical dissolution. The intent was to use centrifugal force to remove the unreacted Ni Pt stringers from the wafer surface. Unfortunately, this approach did not work and pure Pt residues were still found on the wafer surface.
Extending the duration of the current wet clean steps was found to increase the attack of the NiPt salicide while the unreacted Ni Pt residue was not eliminated.
So a new wet clean chemistry was developed to completely removing all the unreacted Ni Pt consistently making the HVM ready process.
Results
Figure 3 shows the effect of using the new wet clean at the Ni Pt Strip 1 step. Complete removal of Ni high Pt residues was found. The experiment was found to be repeatable and verified on lots staggered in the production line by 2 weeks.
Additional details and results will be provided in the complete presentation.
References
[1] T. Schram et al., Microelectronic Engineering, 120 (25), pp. 157–162, (2014).
[2] K. Xu et al., ECS Transactions, 11 (2), pp. 327-334, (2007)