Intermetallic growth kinetics in gold ball bonds on Al-1%Si-0.5%Cu bond pads at 175^∘C

The purpose of this study is to demonstrate that isothermal intermetallic growth data for gold ball bonds can be non-parabolic with explanations of why deviation from parabolic kinetics may occur.

Intermetallic thickness measurements were made at the centre of cross-sectioned ball bonds that were isothermally annealed at 175°C. Intermetallic growth kinetics were modelled with a power law expression(x(t) − x₀)² = α₁t^α₂. The parameters of the power law model were obtained by transformation of the response and explanatory variables followed by data fitting using simple linear regression (SLR).

Ball bonds made with 4 N (99.99%Au) and 3 N (99.9%Au) gold wires exhibited two consecutive time regimes of intermetallic growth denoted Regime I and Regime II. Regime I was characterised by reactive diffusion between the gold wire and the aluminium alloy bond pad, during which Al was completely consumed in the formation of Au–Al intermetallics with non-parabolic kinetics. In Regime II, the absence of a free supply of Al to sustain intermetallic growth led to the conclusion that thickening of intermetallics was caused by phase transformation of Au₈Al₃ to Au₄Al. Ball bonds made with 2 N (99%Au) wire also exhibited non-parabolic kinetics in Regime I and negligible intermetallic thickening in Regime II.

The analysis of intermetallic growth is limited to total intermetallic growth at a single temperature (175°C).

The value of this study lies in showing that the assumption that only parabolic intermetallic growth is observed in isothermally aged gold ball bonds is incorrect. Furthermore there is no need to assume parabolic growth kinetics because with an appropriate data transformation, followed by fitting the data to a power law model using SLR and with the use of statistical diagnostics, both the suitability of the kinetic model and the nature of the growth kinetics (parabolic or non-parabolic) can be determined.