This research relates electrochemical failure of encapsulated microelectronics to surface moisture and to surface impurities. It draws heavily on “osmotic blistering”, a phenomenon known to produce corrosive failure of coated/painted metals [1].

Leakage currents were measured on aluminum comb specimens as a function of relative humidity (RH) and temperature. We studied both bare combs and combs encapsulated with polysiloxane. 9.05 Vdc bias was used. We performed “contamination-by-design” experiments by deliberately introducing known amounts of NaCl, CaCl2 and sucrose onto the comb surface. Results were compared with corresponding data taken on well-cleaned specimens.

Our principal findings are

(1) Under dry conditions (RH<1%), small leakage currents are observed, ranging between 1-10 pA, which are insensitive to surface contamination levels. This implies that solid surface impurities Rer se do not promote electrochemical IC failure.

(2) In extremely moist environments (RH>99%), surface-contaminated samples exhibit large leakage currents, ranging between 1 and 10 pA, that are roughly proportional to surface loading.

(3) Different surface chemical compounds produce leakage-current steps at specific RH values, corresponding to solid-to-saturated solution transitions. For example, CaCI2 exhibits a transition at 21% RH; NaCl at 75% RH. At RH values above the transition, aqueous droplets, or vacuoles, were observed at surface sites occupied by solid deposits. The RH location of the transitions is largely unaffected by the presence or absence of polymer encapsulant. Leakage current steps were typically four to six orders of magnitude. The size of the step change varied between bare and encapsulated samples, and with surface loadings.

(4) Variable temperature studies, performed at constant external water vapor, exhibited step decreases in leakage current at temperatures corresponding to saturated solution to solid transitions.

(5) Sucrose, a nonelectrolyte, exhibited a leakage current step similar to those observed with CaCl 2 and NaCl. The size of the sucrose step change was significantly less than that observed with the electrolytes.

(6) Electrochemical attack patterns varied among the different chemical compounds. For example, as shown in Fig. i, CaCl 2 exhibited anodic attack on alternate metallization lines. NaCl produced attack on all metallization lines.

A full report of this work will appear elsewhere [2].