Inline Tilt Measurements of Sintered Dies by Optical Line Scanning as Quality Assessment Tool for Smart Production

Tilt of semiconductor dies is a common issue during assembly in power electronics as, e.g. die bonding during silver sintering caused by inhomogeneous thicknesses of applied die-attach material, problems with a homogeneous force application, uneven substrates etc. This tilt usually leads to reliability problems later during testing or operation in the field. The reason behind this is a gradient in porosity in the die attach, where higher porosity regions will show considerably lower reliability under thermo-mechanical loading conditions. Unlike delamination and voids, excessive porosity and consequently resulting weakened bulk or interface regions can often be overlooked by SoA failure analytical techniques like scanning acoustic microscopy (SAM) or pulsed infrared thermography (PIRT), since acoustic impedance and thermal conductivity respectively are no adequate measure to resolve porosity changes in that interval. To enable products with enhanced reliability requirements such as power modules for future electric autonomous vehicles, tilt as a production tolerance in such assemblies must be reliably detected and inked for binning. Ideally this should be done on every component resulting in a 100% inspection. In addition to tilt, also the warpage of the chip provides valuable information about residual stresses and thus the quality of the joint and potential conspicuous behaviour in the field. Ideally, either of the two should be available within a 100% inline inspection paradigm, based on a novel testing hardware featuring rapid and accurate readout capabilities for immediate feedback into production as well as a connection to a big data environment for traceability within a smart production paradigm. We have developed a rugged, in-line capable and contactless optical tilt and BLT sensor for integration into an electronics manufacturing line under industrial boundary conditions. We furnish characteristic performance indicators exemplified on a production batch of sintered power dies.