Characterization of a spray acid tool for selectively etching under bump metallurgy films used in flip chip applications

A methodology for optimizing and characterizing processes for selectively etching Cu underbump metallurgy in the presence of Pb-Sn solder in spray acid tools is discussed. Wafers with the following UBM and bump structure were used in this study: sputtered TiW/sputtered Cu/electroplated Cu stud/electroplated Pb-Sn solder. The development of a manufacturable etch process is illustrated by discussing the method used to establish the process for selectively etching the sputtered Cu film using a commercial, two-component, ammoniacal Cu etch chemistry. The average etch rate on monitors with blanket sputtered Cu films, placed in all slots of the tool clamshell, was used to establish a non-uniformity factor (η), which is defined as the ratio of the etch rate in the slot displaying the fastest etch rate to the etch rate in the slot displaying the slowest etch rate. Under ideal conditions the non-uniformity factor should be unity. Over the parameter space investigated it was found that the non-uniformity factor was minimized with the following conditions: temperature of 27°C, RPM of 20, pump pressure of 52 psi, etchant flow rate of 6 liters/min, and a 25% solution of the two-component etch chemistry, with the two-components in a 1:3 proportion. Flip chip packages containing electroplated Cu stud and solder bump structures require selective etching of the underlying underbump metallurgy layers that serve as a diffusion barrier and as an electrical bus layer for the electroplating process. In this paper the characterization of a spray acid tool for selectively etching the sputtered Cu bus layer from multiple wafers simultaneously is discussed. The use of the non-uniformity parameter, η, defined earlier, to establish the preliminary process is presented. η was also used to understand the parameters that controlled the uniformity of the spray pattern. In particular the effect of pump pressure, flow rate, and etchant composition on η was used to establish the levels of these variables for the Cu etch process. The preliminary Cu etch process was confirmed with a full lot containing 25 blanket Cu wafers. A maximum overetch of 39% was obtained for the full lot.