Issues with Gold Electroplating for Microelectromechanical System Applications

  • PDF / 179,487 Bytes
  • 6 Pages / 612 x 792 pts (letter) Page_size
  • 98 Downloads / 196 Views

DOWNLOAD

REPORT


ISSUES WITH GOLD ELECTROPLATING FOR MICROELECTROMECHANICAL SYSTEM APPLICATIONS Caroline A. Kondoleon and Thomas F. Marinis Electronics Packaging and Prototyping Division, The Charles Stark Draper Laboratory, Cambridge, MA 02139 ABSTRACT Electro-plated gold films are used extensively in packaging of MEM sensors to make connections to signal conditioning electronics. Over the past two years, production lots of gold plated substrates, procured from various vendors, failed an accelerated aging qualification test. In this test, 0.0025 [mm] diameter aluminum wires were ultrasonically welded to the film, aged at 120 °C for 48 hours, and then pulled to destruction. The criterion for passing this test was that the wires should break both before and after aging. In the defective lots, the wires lifted off of the gold film after aging. Analysis of these defective films by SEM, Auger, and TOF-SIMS suggested that residues, deposited from the plating bath, concentrated beneath the bond as the gold and aluminum reacted to form an inter-metallic compound during aging. A combination, etch and cleaning treatment was developed for defective substrates, which removed a sufficient amount of residues from the gold to pass the qualification test. INTRODUCTION Aluminum wires bonded to electro-plated gold films constitute the first level of interconnect in many inertial MEM sensor applications. This combination is preferred because aluminum wires are highly robust to deformation in high acceleration environments and can be attached to gold films at room temperature. This interconnect structure has been extensively studied and shown to be very reliable when used at temperatures below 200°C.1 One of the criteria for accepting production lots of gold plated substrates is that aluminum wires bonded to them and aged for 48 hours at 120°C, should break rather than lift off the gold film when pulled to destruction. Our experience with three different vendors has been that only about half of the lots pass this test. Initially, the bonds are strong and all wires break in these rejected lots. As aging progresses, however, the bond strength steadily diminishes until most bonds fail by lifting off the gold film. EXPERIMENTAL Cross sections of aluminum wires bonded to gold films appear in Figure (1). Al Wire

Al Wire

Au Au

Ni

5 µm

Ni

Figure 1a. Cross section of an aluminum wire bond on gold metallization from an accepted lot.

5 µm

Figure 1b. Wire bond from a sample that failed the 48-hour age at 120°C qualification.

B5.18.1

Figure (1a) is from a sample that passed qualification and Figure (1b) is from one that failed. There are no apparent differences between these bonds with respect to gold or nickel thickness or deformation of the aluminum wire. Energy Dispersive X-ray (EDX) spectra of these cross sections also suggest no explanation for their different performance in the aging test. Samples of gold film from failed lots were examined using time-of-flight secondary ion mass spectroscopy. Spectra from a typical lifted wire bond site appear in