THEORY OF PRESSURE SINTERING OF GLASS-CERAMIC MULTICHIP CARRIERS

A method of pressure sintering multilayer glass ceramic packages (MLC) that results in a hermetic product is described, Use of this process results in a reduced process time, and is achieved without the use of a die, which is commonly employed during pressure sintering. Complex g lass ceramic multichip substrates have been sintered this way to produ ce products with a flat edge contour and minimal distortion of the int ernal vias. In this article, we present a model that provides the fund amental basis for the pressure sintering approach to processing MLC. I n this semi-quantitative model, the mechanism of pressure sintering, i .e., the process of dimensional changes, is controlled by viscous how induced by sintering, and lamination flow created by an applied pressu re. The pressure sintering model is capable of predicting the how, den sity, and dimensional changes of a glass ceramic carrier during pressu re sintering in the absence of a die. Both the temperature and pressur e schedules are time dependent, and the pressure can vary over a range from zero, corresponding to free sintering, to large pressures (up to at least 800 psi) that closely simulate experimental data. The input parameters required for the model are the initial dimensions and initi al density of the substrate, the particle size, and the viscosity-temp erature-time relationship of the glass ceramic. This model has been ap plied to the sintering of glass ceramic substrates under a variety of pressure and temperature schedules, and the predictions of the model w ere found to agree well with experimental findings in terms of changes in the thickness, the density, and the edge bulge for substrates of d iffering thickness and cross-sectional areas, the latter of which is d efined as the product of substrate width and length. Moreover, we have obtained the scaling laws for substrate dimensions and the pressure s chedule required to maintain a hat edge (given a temperature schedule) for different substrate cross-sectional areas. These scaling laws are important for deducing useful information that can then be extrapolat ed to model fall sized substrates, based on data from experiments on s mall parts that are more amenable to experimentation.