Lifetime of high- k gate dielectrics and analogy with strength of quasibrittle structures

Jia Liang Le, Zdeněk P. Bažant, Martin Z. Bazant

Research output: Contribution to journalArticlepeer-review

10 Scopus citations

Abstract

The two-parameter Weibull distribution has been widely adopted to model the lifetime statistics of dielectric breakdown under constant voltage, but recent lifetime testing for high- k gate dielectrics has revealed a systematic departure from Weibull statistics, evocative of lifetime statistics for small quasibrittle structures under constant stress. Here we identify a mathematical analogy between the dielectric breakdown in semiconductor electronic devices and the finite-size weakest-link model for mechanical strength of quasibrittle structures and adapt a recently developed probabilistic theory of structural failure to gate dielectrics. Although the theory is general and does not rely on any particular model of local breakdown events, we show how its key assumptions can be derived from the classical dielectric breakdown model, which predicts certain scaling exponents. The theory accurately fits the observed kinked shape of the histograms of lifetime plotted in Weibull scale, as well as the measured dependence of the median lifetime on the gate area (or size), including its deviation from a power law. The theory also predicts that the Weibull modulus for breakdown lifetime increases in proportion to the thickness of the oxide layer and suggests new ideas for more effective reliability testing.

Original languageEnglish (US)
Article number104119
JournalJournal of Applied Physics
Volume106
Issue number10
DOIs
StatePublished - 2009
Externally publishedYes

Bibliographical note

Funding Information:
Partial financial support for the underlying statistical theory was received under NSF Grant No. CMS-0556323 to Northwestern University.

Fingerprint Dive into the research topics of 'Lifetime of high- k gate dielectrics and analogy with strength of quasibrittle structures'. Together they form a unique fingerprint.

Cite this