Optimisation of the conditions of thermosound microwelding superlarge integrated circuits

Abstract

When producing superlarge integrated microcircuits (SIM) the effect of the thermomechanical parameters on semiconductor structures must be minimised. Therefore, to produce wire joints between crystals and external leadouts of SIM, an increasing use is made of the method of thermosound microwelding, in which the process of thermal compression microwelding is activated by ultrasound oscillations. This reduces the temperature required for producing high-quality wire interjoints from 603-573 to 473-393K.' This method is also efficient when assembling crystals on adhesive compositions, whose volatile components in polymerisation of the adhesive are absorbed on the contact boards of the SIM. However, published work contains only a small amount of experimental data on the effect of the main parameters of the conditions of thermosound microwelding, such as the force acting on the working tool P, the temperature T, the time tc and power of ultrasound oscillations We, the mechanical strength of microwelded joints in shear on the crystal Fk and separation on the transverse F,, the reproducibility of the quality and kinetics of formation of welded joints, and the nature and depth hmm of interaction of welded microsections. The aim of this work was to determine the optimum conditions of thermosound microwelding for producing highquality and reliable wire joints. To produce wire joints between a crystal and the external leadouts of SIM, the experiments were carried out on ZL999.9T gold wire (TU 48-1-353-87) 30 |am in diameter with a mean tensile strength of F = 0.146N and a relative elongation of 8= 5-6%. These characteristics were inspected in a 12MP5/20-1 mechanical testing system with an accuracy to ±1% for F and to ±5% for 8. After melting the ends of the wire leadouts with an arc discharge into a sphere 85-90 nm, the leadouts were connected to aluminium contact boards of the SIM 1 \un thick. The latter were placed on an SiO2 substrate 0.5-0.6 |im thick in EM-4060 P automatic equipment. The source of ultrasound oscillations was represented by a mechanical system with an oscillation frequency of 64 kHz and a power of 6.3W. After producing an Au-Al microwelded joint on a crystal, the wire was connected to the external leadout of the outlet frame of 42N alloy with a local gold coating 3 |a.m thick to produce an Au-Au contact pair. The crystal was connected to a crystal holder through a current-conducting adhesive based on a phenol resin glue with currentand heat-conducting additions in the form of Ni and Si. To evaluate the extent of interaction of the Au-Al compounds and construct profile diagrams, the aluminium was pickled in a 50% NaOH solution. Visual inspection of the surface of the interaction zones on the gold sphere was carried out by scanning electron microscopy. Transverse sections of microwelded joints were also produced and examined in an REMP-4 scanning electron microscope. The electrical power supplied to the ultrasound transducer was determined from the results of direct measurements of current and voltage, and the phase shift between then in a measuring system as described in Ref.2. The concentration of surface organic contaminants was inspected by measuring the contact wetting angle of the examined surface by a droplet of deionised water (10 kg/m). The thickness of the residual oxide on aluminium contact boards of SIM crystals was measured in an LEM-2 laser ellipsometric microscope with an accuracy to ±0.5 nm over an area of 100 |im ( 240 joints for each microwelding regime. The reproducibility of the quality of microwelded joints was determined from the coefficient of variation of strength K,, equal to (am/Nm)