Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Analysis of electromigration-induced backflow stresses in Cu(Mn) interconnects using high statistical sampling

: Kraatz, Matthias; Sander, Christoph; Clausner, André; Hauschildt, M.; Standke, Yvonne; Gall, Martin; Zschech, Ehrenfried


Institute of Electrical and Electronics Engineers -IEEE-:
IEEE International Reliability Physics Symposium, IRPS 2018 : Burlingame, California, USA 11 – 15 March 2018
Piscataway, NJ: IEEE, 2018
ISBN: 978-1-5386-5480-4
ISBN: 978-1-5386-5479-8
International Reliability Physics Symposium (IRPS) <56, 2018, Burlingame/Calif.>
Fraunhofer IKTS ()
Copper alloys; Electromigration; failure analysis; integrated circuit interconnections; integrated circuit reliability; manganese alloys

Electromigration-induced stresses play a major role in the analysis and the explanation of this important reliability degradation mechanism. It is generally assumed that the directed atomic transport generates tensile stress at the cathode end of an interconnect structure while compressive stress is generated at the anode end. At short on-chip metal interconnect lengths, the electromigration-induced stress gradient between the cathode and the anode counteracts the primary electromigration effect (i.e. the directed atomic transport), leading to the well-known Blech effect. In this study, Wheatstone bridge devices with 4 times 100 Cu interconnects are used in a configuration which is different from the usual application. All 400 Cu interconnects per bridge are wired in a parallel/series arrangement without the regular resistance tapping, disabling the original approach of a high detection probability of early failures. Instead, a very high statistical sampling for measurements of resistance increases occurring in each individual interconnect can be achieved, leading to an almost ldquoperfectrdquo resistance increase behavior as a function of stress time at elevated temperatures. Furthermore, at conditions close to, but above the critical current density . length product (jL)c, the linear drift portion as well as saturation effects at later times can be used to evaluate the critical product with statistical significance. Selecting the robust contact/M1 interface in a first proof-of-concept analysis, a value of (jL)c = (510 +or- 110) mA/mum is determined for 40 mum long Cu(Mn) M1 interconnects with 45 nm width and 100 nm height.