November_EDFA_Digital

edfas.org ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 23 NO . 4 34 layer of a multilevel metal trap, using the W Lα1 emission line from the vias (see Fig. 5). PURPLE PLAGUE MITIGATION Another ion trap device failure occurs at the interface of the gold (Au) wire bond attached to the aluminum (Al) wire bond pad on the ion trap chip. Wire bond failures are primarily induced by contamination/impurities, poor welding, or extreme thermal exposure. [15] It is well known that gold and aluminum form a brittle intermetallic compound (IMC); in the microelectronics community it is known as purple plague due to the purple color of the AuAl 2 phase. [16] Common failuremodes arewires breaking at the heel of the bond caused by consumption of the gold in the IMC (see Fig. 6), opens caused by full consumption of the Al surrounding the bond, and bonds lifting due to Kirkendall voids under the bond. As noted, because ion traps can experience elevated times and temperatures in their preparation for and use in a QIS experiment, care in device and packaging materials choices and the thermal processes imposed is required. While there are many techniques employed for mitigating the effects of purple plague in microelectronics, [15] limiting the exposure time to elevated temperatures has been found to be the most expedient for ion traps. Trapwire bond aging experiments show that baking at 200°C for longer than seven days greatly reduced the pull strength of the wire bond (see Fig. 8), agreeingwell with the lifetime predicted in Blish. [17] A better approach to mitigate the effects of purple plague for ion traps is to use a homogeneous material system such as Au wires and Au I/O pads on the device, taking advantage of the native Aumetallization of ceramic packages. In fact, Au coating of trap electrodes is preferred since Au does not support a native oxide that can support unpinned charges seen by the ion, unlike Al-½%Cu. Depositing a sufficiently thick diffusion barrier such as platinum(Pt) between the Au and Al-½%Cu layers can sig- nificantly slow the formationof purple plague. A 300 nmPt barrier has shown to significantly decrease the IMCgrowth rate (see Fig. 7b), which may increase the functional life- time of ion trap devices. The resulting increase in thermal budget latitude may open the path for other packaging process options, such as solder reflow, and assembly paradigms, for example heterogeneous integration of Fig. 6 SEM image of a goldwire separated fromthe RF bond pad in an ion trap. High temperatures from the UHV preparation bake or from RF dissipation during use accelerate Au-Al intermetallic growth. This can lead to the heel of the bondbreaking thus causing anopen electrode on the device. Fig. 7: (a) Purple plague formation around a Auwire bond on an Al bond pad after undergoing a 10 day bake at 200°C, a common temperature/time exposure of an ion trap in preparing for first use. (b) Au wire bond on a Au bond pad with sufficient platinumbarrier layer after 21 days at 200°C showing no observable evidence of intermetallic formation. The longer bake times allow for a trap to be used more than once. (a) (b)