Influence of Small Weight Percentages of Bi and Systematic Coefficient of Thermal Expansion Variations on Sn Whiskering

Abstract

In Part A of this paper, we have investigated how small, controlled amounts of Bi influence Sn whiskering. Three custom Sn–Bi sputter targets of 0.5%, 1.0%, and 2.0% Bi (by weight) were used to generate $\sim 2000$ A Sn films on Si substrates. The samples were incubated at room temperature (RT) followed by thermal cycling ( $- 40 ) to accelerate whisker growth. The control group was not thermal cycled. Whisker densities dramatically dropped to zero as the concentration of Bi increased to 2.0% for the RT specimens (100-day incubation). However, subsequent thermal cycling turned ON whisker growth ( $\sim 25$ 000 whiskers/cm $^{\mathrm {\mathbf {2}}}$ ) in all the three Bi% cases. Thus, Bi additions suppress whiskering during RT incubation, but the same films readily grow whiskers during thermal cycling. In Part B, a systematic range of coefficient of thermal expansion (CTE) variations between the substrate and Sn (pure, i.e., no Bi or Pb) were investigated to determine the effect on whisker growth. CTEs close in value to Sn ( $0 ) were Al, Ag, and brass; intermediate in value to Sn ( $25 ) were Zn, Ni, and Ta; and far in value to Sn ( $75 ) were semiconductors Si, GaAs, and InP. A thickness of 0.5 $\mu \text{m}$ of sputtered Sn was deposited on each coupon. The thermal cycling range was −40 °C to 125 °C, with 2-h ramps and 4-h dwells, for a total of 12 h per cycle. A second, comparative set of specimens underwent isothermal annealing at 100 °C. All the samples were incubated for 37 days (74 cycles) before observation. The results show that samples with a % $\Delta \text {CTE} >75$ % had drastically higher whisker densities (when cycled) compared to those with % $\Delta \text {CTE} %. There appears to be a critical activation (or nucleation) threshold CTE mismatch that “turns ON” whisker growth when using semiconductors as substrates.