Here is the abstract you requested from the imaps_2019 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.
|Ferrites in Transfer Molded Power SiPs – Challenges in Packaging|
|Keywords: ferrite cores, packaging, Transfer Molding|
|According to recent technology studies, largest power module markets are industrial motor drives segment and the fast growing Electric Vehicle / Hybrid Electric Vehicle (EV/HEV) market. Here, the trend toward power modules being over-molded with high- temperature compatible Epoxy Molding Compounds [EMC], especially in automotive applications, can be observed. Main advantages of molded devices are reduced material cost, enabling complex and compact module designs, and an improved thermal management. As for low power discrete devices, the encapsulation with EMC is almost standard; but also for high power modules, Transfer Molding is enjoying growing interest. Here, volume constraints and related high- power density challenges require the development of new power System-in- Packages [Power SiPs], where not only transistors and diodes, but also drivers, passives, coils and ferrites shall be integrated in one package. Today’s integration of ferrites into power modules, encapsulated with silicone gel, is typically done by consideration of electrical properties of the ferrite only. Following the trend with molded power modules, the mechanical properties of ferrites need to be considered as well. Encapsulating modules in a Transfer Mold process induces higher mechanical load onto module components compared to silicone potting. Previous investigations have shown, that integration of delicate components as ferrite cores into molded packages is not as trivial as integration of conventional SMDs or power semiconductors; the brittle ferrites tend to fracture during the encapsulation process, resulting in higher ferrite core loss. Present study identifies the main root causes for ferrite core cracking during manufacturing of molded Power SiPs. Investigations focus on Manganese- Zinc, as that composition is the main material mix for filters and transformers, which can be integrated in power modules. As technical datasheets of ferrite cores typically focus on electro-magnetic performance only, and are often lacking of mechanical or even thermo-mechanical material properties, the used ferrite cores are analyzed more detailed. Module application determines ferrite grain size - one ferrite property, which is investigated here regarding tendency to fracture. It will be shown, that ferrites consist of microscopic and macroscopic grains. While microscopic grains define the electrical use case, macroscopic grains have a significant influence on mechanical strength of the ferrite. Second on component side, core design itself is assumed to be one key factor for ferrite integrity. Therefore, a test vehicle, developed at Fraunhofer IZM, is used to assemble and to encapsulate ferrites with different grain size and design. Test vehicle is a symmetrical, PCB based package; capable to be assembled with three pairs of E-cores. Encapsulation of PCB itself is done with different kinds of high-thermally conductive EMC ( > 1 W/m K) in a Transfer Mold process. Process related material properties of these EMCs, like viscosity and reaction behavior, are determined to better understand process induced stress. A dedicated mold tool is applied to measure viscosity of EMC under process temperature (175 °C), enabling the simulation of flow behavior of liquid EMC during cavity filling. Ferrite integrity is observed along the full manufacturing process flow from ferrite mounting to PCB to module encapsulation and thermal storage at elevated temperature for Post Mold Cure. Non-destructive and destructive analysis identify risky process steps. Influence of soft and hard adhesive for ferrite attach onto PCB substrate is shown and confirmed in finite element modeling [FEM]. Fluidic simulations, based on Finite Volumes (FV), incorporating shear- thinning behavior of the EMC melt superimposed by time- and temperature- dependent conversion, allow to study the filling stage. This sheds light on the potentially asymmetrical flow around the ferrites, which promotes cracking. Furthermore, optimized process parameters, such as plunger speed, can be determined regarding symmetrical flow, as well as pressure and shear load onto the ferrites during encapsulation. Short shots, showing the filling behavior of mold cavity, are used to validate the model. Finite Element Analysis (FEA) is used to assess the risk of ferrite fracture by calculating the thermo-mechanical stresses induced along the process chain, e.g. by cooling down from encapsulation temperature to room temperature. From four-point bending experiments of ferrites with different grain structure, the stiffness as well as a critical fracture stress are determined. The fracture stress for both materials differ enormously, whereas the stiffness only differ a fraction. The EMC materials and the different glues are characterized for their visco-elastic behavior by means of multi-frequency dynamical mechanical analysis (DMA) and by thermo-mechanical analysis (TMA) for their coefficient of thermal expansion. As the reaction shrinkage of the EMC material most likely plays a significant role in stress development in the assembly, this effect cannot be neglected in the numerical simulations. By comparing measurements of the surface of pre- molded assemblies, and comparing these with the numerical simulation results, the shrinkage value is determined in an iterative manner. A sensitivity study by FEA is performed by studying different ferrite geometries, as well as different materials for ferrite attach and encapsulation. The relation between material/geometry combination and the risk for ferrite fracture will be discussed. Based on the results of the FEA, samples are molded and tested for ferrite fracture by optical non- destructive (ultrasonic microscopy, x- ray-computer tomography) and destructive analysis (cross sectioning/grinding). Applying these methods, it is still challenging to detect thin incipient ferrite cracks. But as every ferrite core crack influences the magnetic flow behavior, another method to detect core cracks might be the core loss measurement. Suitability of crack detection is investigated here – the results of core loss measurements according to Baumann are compared to the common, optical analyzing methods. In summary the paper will discuss the challenges in encapsulation of sensitive components as ferrite cores by Transfer Molding and provide information on optimization of packages with integrated ferrites. Finally, an outlook towards future investigations will be provided.|
|Tina Thomas, scientific engineer