AAU Energy
PhD Defence by Masaki Takahashi

Pon 101 - 1.001/online
19.11.2024 13:00 - 16:00
English
Hybrid
Pon 101 - 1.001/online
19.11.2024 13:00 - 16:0019.11.2024 13:00 - 16:00
English
Hybrid
AAU Energy
PhD Defence by Masaki Takahashi

Pon 101 - 1.001/online
19.11.2024 13:00 - 16:00
English
Hybrid
Pon 101 - 1.001/online
19.11.2024 13:00 - 16:0019.11.2024 13:00 - 16:00
English
Hybrid
Supervisor:
Stig Munk-Nielsen
Co-Supervisor:
Asger Bjørn Jørgensen
Assessment Committee:
Kjeld Pedersen (Chair)
Professor Layi Alatise, University of Warwick
Professor Dr.-Ing Thomas Basler, Chemnitz University of Technology
Moderator:
Szymon Beczkowski
Abstract:
Silicon carbide metal-oxide-semiconductor field-effect-transistors (SiC-MOSFETs) are expected to replace silicon (Si) power semiconductors, which are currently the mainstream. Exceptionally proficient in high-voltage, surpassing 10kV, SiC-MOSFETs are undergoing extensive research and development to facilitate novel applications, such as Megawatt wind power systems. Distinguished by their unique device and power module package structures, SiC-MOSFETs engineered for high voltages exhibit notable disparities from conventional 1.2- 1.7kV counterparts. Noteworthy differences include an approximately tenfold increase in edge length, over double the device thickness, and approximately threefold greater weight in the ceramic substrate compared to 1.2 kV SiC-MOSFETs. While these attributes ensure insulation integrity at voltages exceeding 10 kV, their ramifications on thermal characteristics warrant further exploration.
This study investigates the heat generation attributes of a half-bridge power module employing 10 kV SiC-MOSFETs and introduces a design methodology to enhance thermo-mechanical reliability. Typically, reliability design relies on "power cycle curves" characterizing failure capability under varied conditions. However, owing to the limited deployment of 10 kV SiC-MOSFETs in the market, extensive evaluation and design iterations involving numerous failure tests still need to be completed. Consequently, this research adopts a digital design approach, leveraging three-dimensional thermo-mechanical simulations with a digital twin of the power module, minimizing the necessity for physical prototypes.
Initially, the temperature of 10 kV SiC-MOSFET power modules were measured and compared with 3D thermal simulations to validate the accuracy of the thermal modeling. Then, the 10 kV SiC-MOSFET power module has a greater temperature gradient on the die surface due to their large edge area as a heat conductive structure.
Design optimization of the 10 kV SiC-MOSFET power module structure for power cycle (PC) capability were demonstration based on the thermal characteristics. One strategy encompassed enhancements in wire bonding layout to reduce the wire heat generation, thereby averting wire lift-off. As an initial vilification with a small number of samples, actual PC tests showed a 25 % enhancement in PC capability post-layout improvements, with the failure mode shifting from the wire bond to the solder layer. Additionally, based on thermo-mechanical simulations, a die structure was proposed to alleviate stress on the solder layer, culminating in a 350 % stress improvement compared to pre-enhancement levels, deduced from lifetime calculations based on three-dimensional thermal simulation results.
Moreover, to strengthen heat cycle (HC) capability, experiments revealed failure occurrences at the copper pattern on the DCB substrate, prompting the proposition of a pattern layout to curtail pattern shape deformation during thermal stress simulations. The efficacy of this layout in enhancing reliability was validated through practical demonstrations.
This research clarify that the thermo-mechanical brittleness caused by the large temperature gradient of 10 kV SiC-MOSFETs. They suggested important parameters for structural optimization as an initial design, especially for devices such as 10 kV SiC-MOSFETs for which many physical prototypes are not available. This design approach contributes to early-stage reliability design with minimal reliance on physical prototyping by utilizing theoretical considerations based on 3D thermo-mechanical simulations and reliability failure physics models.