Failure Analysis of Elevated Contact-Chain Resistance and Corrective Action
Failure Analysis of Elevated Contact-Chain Resistance and Corrective Action
Wednesday, October 7, 2026: 8:40 AM
Summary:
The resistances of N+ and P+ contact chains of two technology nodes were elevated which had caused production yield loss. In one technology node A, the cobalt (Co) silicide thickness decreased under the contact and near the trench-isolation edge, and the contact etch terminated directly on the doped silicon (Si). The P+ contact chains of a different technology node B were measured to be electrically open. The silicide became thinned and corrugated in its interface with the underlying Si. The contact also landed on Si. These structural features in both nodes elevated the contact chain resistance. A second set of observations showed the presence of Si and oxygen (O) entrapped within two layers of silicide in node A. Root cause analysis points to an inadequate pre-Co clean of the active surface, evidenced by increased reflective power of the pre-Co clean tools. Taken together, the data suggest a unified mechanism of the issue of node A: an incomplete surface preparation locally impeded silicidation, causing the reduced silicide mass/thickness, contact landing on Si and oxide entrapment in silicides. The proposed mechanism is compatible with the issue of technology node B, and aligns with established kinetic models where even ultrathin interfacial oxide can strongly alter Co silicide formation. The issues were successfully resolved by restoring the efficacy of the pre-Co sputter clean, highlighting the sensitivity of advanced silicide integration to surface boundary condition
The resistances of N+ and P+ contact chains of two technology nodes were elevated which had caused production yield loss. In one technology node A, the cobalt (Co) silicide thickness decreased under the contact and near the trench-isolation edge, and the contact etch terminated directly on the doped silicon (Si). The P+ contact chains of a different technology node B were measured to be electrically open. The silicide became thinned and corrugated in its interface with the underlying Si. The contact also landed on Si. These structural features in both nodes elevated the contact chain resistance. A second set of observations showed the presence of Si and oxygen (O) entrapped within two layers of silicide in node A. Root cause analysis points to an inadequate pre-Co clean of the active surface, evidenced by increased reflective power of the pre-Co clean tools. Taken together, the data suggest a unified mechanism of the issue of node A: an incomplete surface preparation locally impeded silicidation, causing the reduced silicide mass/thickness, contact landing on Si and oxide entrapment in silicides. The proposed mechanism is compatible with the issue of technology node B, and aligns with established kinetic models where even ultrathin interfacial oxide can strongly alter Co silicide formation. The issues were successfully resolved by restoring the efficacy of the pre-Co sputter clean, highlighting the sensitivity of advanced silicide integration to surface boundary condition