Market forces are driving the use of smaller passive component packages and finer I/O pad pitch integrated circuits.
By Jayna Sheats
The burgeoning popularity of a wide variety of mobile electronic appliances has provided a strong impetus to shrink components and packages as much as possible and to save space and weight for aesthetic reasons. For example, a list of major uses of MEMS components whose volume is projected to reach $4.6 billion by 2012 includes game controllers, digital cameras, camcorders, MP3 and portable media players, personal navigation devices, remote controllers, and sports equipment, laptop computers, phones, and a few larger, non-portable items. In addition, increasingly large numbers of handheld electronic appliances can be found in medical and industrial arenas.
These market forces are driving the use of smaller passive component packages and finer I/O pad pitch integrated circuits. Three-dimensional stacking (component on component, as well as embedded in the package) can also reduce area. By themselves, these structures do nothing to reduce thickness, however, and stacking unthinned components increases thickness.
The thickness of silicon ICs arises largely from convenience in handling. Production and handling of 200 mm and larger diameter wafers requires a minimum thickness for mechanical reasons. The circuits themselves occupy only a few microns of thickness, and indeed today there is a huge amount of activity around thinning these wafers to make them more readily stacked and interconnected by through-silicon vias. A thickness of 25 μm is easily obtained; even thinner is possible with careful attention to process control.
Such thicknesses open up new possibilities for interconnects that were not previously feasible. Today, chips and passives are connected either by wirebonding or by the so-called flip-chip process. The former uses ultrasonic energy to weld wires that may be as small as 25 μm in diameter to the bond pads; a sophisticated mechanical apparatus holding a hollow tip moves the wire from one pad to another and cuts it after bonding. This process is used for the majority of connections made today, despite its slowness (several bonds/sec), because robust metal connections can be made at a high lateral density (down to about a 50-μm pitch for the state of the art).
Did you enjoy this article? Click here to subscribe to the magazine.