Abstract

In this study, the line-type Cu/Sn/Cu and Cu/Sn/Ni interconnects were used to determine the growth kinetics of interfacial intermetallic compounds (IMCs) under electromigration (EM), and the current crowding effect and thermomigration are expected to be avoided in this line-type interconnects because of their symmetric structure. The Cu/Sn/Cu interconnect was under the current density of 5.0×10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> A/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> at 100°C and 150°C, and the Cu/Sn/Ni interconnect was under the same current density at 150°C. For the purpose of comparison, the Cu/Sn/Cu and Cu/Sn/Ni interconnects were aged at the same temperatures for the same durations. In the case of Cu/Sn/Cu interconnect, the same types of IMCs, Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> and Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> Sn, formed at the Sn/Cu interface, which was independent of electric current. EM caused a polarity effect, i.e., the interfacial IMCs on the anode side were significantly thicker than those on the cathode side. The growth kinetics of the interfacial IMCs on the anode side during EM were significantly enhanced compared with that of the aging (the no-current case), and still followed a t <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/2</sup> law with time. The temperature was one of the critical factors that influenced the EM. The effect of EM became more significant at higher temperature under the same current density. The growth behavior of the interfacial IMCs on the cathode sides was complicated. When the initial interfacial IMCs were very thin, the inward atomic fluxes were larger than the outward fluxes and thus the interfacial IMCs grew. After the IMCs reached a critical thickness, the inward atomic fluxes were less than the outward fluxes and thus the thickness of the interfacial IMCs decreased. In the case of Cu/Sn/Ni interconnects, Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> and Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> IMCs formed at the as-soldered Sn/Ni and Sn/Cu interfaces, respectively. The Cu content in the IMCs at the Sn/Ni interface increased with the increasing aging time, and the original Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> IMC at the Sn/Ni interface transformed into (Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.56</sub> Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.44</sub> ) <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> after aging at 150°C for 200 h; while the IMC at the Sn/Cu interface remained Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> , which contained less than 0.5 at% Ni even after aging at 150°C for 200 h. When electrons flowed from Cu side to Ni side in the Cu/Sn/Ni interconnects during EM at 150°C, the original interfacial Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> IMC at the Sn/Ni interface (anode side) had already transformed into (CuNi) <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> type after EM for 100 h. After EM for 200 h, (Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.60</sub> Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.40</sub> ) <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> formed at the Sn/Ni interface and Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> (containing less than 0.1 at% Ni) formed at the Sn/Cu interface. When the direction of electron flow was reversed, after EM at 150°C for 200 h, the types of IMCs remained unchanged, i.e., Ni <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> (containing 2 at% Cu) and Cu <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">6</sub> Sn <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">5</sub> (containing less than 2 at% Ni) formed at the Sn/Ni and Sn/Cu interfaces, respectively. The diffusivity of Cu atoms in Sn is two orders of magnitude higher than that of Ni in Sn. Thus, Cu atoms would easily diffuse across the bulk solder than Ni atoms and influence the interfacial reactions on both the anode and cathode sides when electrons flowed from Cu side to Ni side. However, the diffusion of Cu atoms was blocked while against the electron wind, i.e., when electrons flowed from Ni side to Cu side.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call