Understanding the impact of surface engineering, structure, and design on electromigration through Monte Carlo simulation and in-situ SEM studies

Z. H. Gan*, W. Shao, M. Y. Yan, A. V. Vairagar, T. Zaporozhets, M. A. Meyer, A. Krishnamoorthy, K. N. Tu, A. Gusak, E. Zschech, S. G. Mhaisalkar

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

The Cu dual-damascene interconnects are investigated by studying the effects of surface engineering including surface cleaning and alloying, design including M1/M2 architectures, buried Ta layers, and interconnect tree structures through Monte Carlo simulations, TEM and in-situ SEM. The Cu surface was treated with reducing, inert, and reactive gases to induce compositional changes at the interface capped by different layers including silicides, nitrides and their variants. In untreated samples, the asymmetry of the test structures leads to higher MTFs for the upper layer (M2) test structures, however upon surface treatments, the MTFs of lower layer (M1) test structures showed a 2× improvement and could achieve lifetimes comparable to M2 test structures. Cu-Sn chemical bonding was introduced to the Cu/SiNx interface by immersion Sn surface treatment. This chemical force effectively blocked the dominant surface diffusion path in Cu interconnect, thus leading to 10× lifetime improvement. Failure mode changed from via corner voiding to voiding in the middle of the line. Direct evidence of electromigration (EM) mechanism in interconnect tree structures consisting of void nucleation and movement in opposite direction to electron flow along the Cu/SiNx interface was unraveled by in-situ SEM. It was observed that the EM mechanism of a segment in a Cu interconnect tree depends on the current configuration in neighboring interconnect segments. Reservoir effect on electromigration lifetimes in copper interconnects was also studied. It was found that there is a critical extension length (∼60 nm) beyond which increasing extension lengths ceases to prolong electromigration lifetimes. The underlying mechanism was investigated by means of current density distributions calculated from finite element analysis. A dual damascene structure with an additional 25 nm Ta diffusion barrier embedded into the upper Cu layer was also fabricated. With this structure, a lifetime improvement of at least 40 × was achieved. Finally, to understand the process of void nucleation, migration, and coalescence, a kinetic Monte Carlo simulation of 3D nano-void behavior at the metal/dielectric interface under strong electric current was undertaken. Major stages of failure modes of Cu-lines were simulated, including trapping at and detachment from grain-boundaries and triple junctions, migration and shape evolution of voids along interface and along the grain-boundaries at this interface. A good correlation was obtained between the simulation and the in-situ SEM observations.

Original languageEnglish
Title of host publicationSTRESS-INDUCED PHENOMENA IN METALLIZATION
Subtitle of host publicationEighth International Workshop on Stress-Induced Phenomena in Metallization
Pages34-42
Number of pages9
DOIs
Publication statusPublished - Feb 7 2006
Externally publishedYes
Event8th International Workshop on Stress-Induced Phenomena in Metallization - Dresden, Germany
Duration: Sept 12 2005Sept 14 2005

Publication series

NameAIP Conference Proceedings
Volume817
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616

Conference

Conference8th International Workshop on Stress-Induced Phenomena in Metallization
Country/TerritoryGermany
CityDresden
Period9/12/059/14/05

ASJC Scopus Subject Areas

  • General Physics and Astronomy

Keywords

  • Electromigration
  • In-situ SEM
  • Interconnect tree
  • Monte Carlo simulation
  • Reservoir
  • Surface strengthening

Fingerprint

Dive into the research topics of 'Understanding the impact of surface engineering, structure, and design on electromigration through Monte Carlo simulation and in-situ SEM studies'. Together they form a unique fingerprint.

Cite this