Project Leader: Daniel Kiener

Semiconductor materials play a crucial role in many electronic applications, for example in the automotive and industrial market. Used as switches, they are designed to safely operate electrical loads (bulbs, valves …) even under fault conditions, e.g. short circuits occurring accidentally in the wiring harness of a car. Such fault conditions lead to substantial electrical power dissipation (up to 3 kW for time-spans of microseconds) in the switch, heating up the semiconductor device by more than 300 K. These temperature rises and thermal gradients lead to substantial mechanical stresses and may cause failure after many activations (“active cycles”), by plastic deformation and fracture of the micrometer-scaled layers.

Understanding the fatigue behavior of these films is of importance to guarantee long term reliability of the new material stacks in the metallization layers and the wafer backside under the influence of high operating temperatures. Therefore, in this project the main objective is to examine the thermo-mechanical fatigue behavior of thin films and interfaces in new high voltage switches and backside film metallizations to be used in novel 300 mm based power devices.

The film behavior will be studied using different in-situ heating and stress measurements of metalized wafers. This is realized using conventional resistive heating with a wafer curvature system and in-situ stress/curvature measurements. Hereby, it is important to understand the thermo-mechanical behavior of the material stack and the particular layers itself (diffusion barriers, gate dielectrics, thermal oxide..). The film stresses developing in a 5 µm thick Cu layer upon heating up to 400 °C are shown below.

Furthermore, since wafer curvature experiments typically take several minutes per cycle, a new system utilizing short laser driven power pulses was developed in this project. This ultrafast heating method enables characterization of thermo-mechanical fatigue behavior up to high cycles (> 100.000 thermal cycles) at switching cycles that aproach application frequencies.

Films will be investigated in terms of their specific microstructural changes (grain size, grain boundaries, interfaces), as well of their global properties (film stress, roughness, electrical resistivity) to develop a material model for fatigue life prediction and enabling specific material improvement. The exemplary microstructure evolution of a fatigued film at the exact same location is shown below, indicating significant grain growth.