Novel polymer-based packaging technologies for high power semiconductors

The NEWTON project intends to replace the existing ceramic hermetic packaging by a novel non-hermetic polymer packaging for high power semiconductors (thyristors). Polymer electrical packaging material candidates have been identified by reverse engineering, by literature review and by using Cambridge Engineering Selector (CES) software. The potential materials identified are polyimide (PI), polyetherimide (PEl), liquid crystal polymers (LCP), poly (phenylene sulphide) (PPS), and epoxy (EP). In addition, two design options (ceramic replacement and over moulded) were selected, out of four, based on all the known issues and the requirements. PEl, PI, LCP, PPS and EP were the selected materials for the ceramic replacement design option and PEl, PI and EP for the over-mould design option. Ceramic replacement option was ranked as the best choice on the basis of the device stability during injection moulding. After a data survey of different producers, a few specific grades for candidate materials were selected. Polymer packaging has to overcome certain identified issues such as non-hermeticity (moisture absorption and transmission), high thermal expansion, voids and micro-voids, weldlines, stress, aging etc., which could be the potential reasons for the device failure. Coatings could also be a good solution to avoid failure due to nonhermeticity. Electrical degradation and breakdown in polymers was facilitated by moisture absorption and transmission, ionic impurities, ionizing radiations, oxidation, poorly dispersed antioxidants, micro-structure defects, aging and low tensile strength etc. A computer generated component model, designed on the ceramic replacement option, was analysed through Moldflow software. The wall shear stress values were compared with the recommended values. The thickness of the model was increased until the wall shear stress values reached just near the recommended values for all the analysed materials. Hence, two models A and 8 were selected for further in depth analysis using pin point tunnel gates. Moldflow results were compared for both models (A and B) and for all the materials. This showed that the PI material and the model A were the best choices. Finally, the candidate materials were quantitatively compared based by reverse engineering, CES materials selection, literature review, electrical degradation, nonhermeticity and Moldflow results. The PI material was the final choice, followed by PEl, PPS, and LCP. Injection moulding was identified as the best processing method. The package quality had to be improved by combined approach of part design, MoldFiow simulation, mould design and controlled processing (plasticising, filling, cooling etc.). Therefore, after 1 the selection of the model 'A', it was further optimised by reducing the fin sizes. This new model was named as 'C' and was further analysed using a disc gate. The parameters and the dimensions of the sprue and the disc gate were selected and optimised based on the Moldflow results. A cooling process phase was designed based on the documented cooling principles. The Moldflow results showed that the designed cooling system was efficient and sufficient. Finally, based on the Moldflow results using a disc gate, a mould has been designed for the polyimide material and the model 'C'.