 |
 |
 |
|
|
|
In this section we will develop an understanding of the basic processes used in monolithic integrated-circuit fabrication. Silicon is the dominant material used throughout the integrated-circuit industry today, and in order to conserve space only silicon processing will be discussed. However, all of the basic processes discussed here are applicable to the fabrication of gallium arsenide integrated circuits (ICs) and thick- and thin-film hybrid ICs.
Germanium was one of the first materials to receive wide attention for use in semiconductor device fabrication, but it was rapidly replaced by silicon during the early 1960s. Silicon emerged as the dominant material because it was found to have major processing advantages. Silicon can easily be oxidized to for silicon dioxide. Silicon dioxide was found to be not only a high-quality insulator but also an excellent barrier layer for the selective diffusion steps needed in integrated-circuit fabrication.
Silicon was also shown to have a number of ancillary advantages. It is a very abundant element in nature, providing the possibility of a low-cost starting material. It has a wider bandgap than germanium and can therefore operate at higher temperatures than germanium. In retrospect it appears that the processing advantages were the dominant reasons for the emergence of silicon over other semiconductor materials.
The first successful fabrication techniques produced single transistors on a silicon die 1 to 2 mm on a side. Early, integrated circuits, fabricated at Texas Instruments and Fairchild Semiconductor, included several transistors and resistors to make simple logic gates and amplifier circuits. From this modest beginning, we have reached integration levels of several million components on a 7 mm x 7 mm die. For example, a one-megabit dynamic random-access memory (DRAM) chip has more that 1,000,000 transistors and more than 1,000,000 capacitors in the memory array, as well as tens of thousands of transistors in the access and decoding circuits. The level of integration has been doubling every one to two years since the early 1960s.
One-megabit RAMs are currently being produced with photolithographic features measuring between 1 and 2 microns (m m). MOS transistors with dimensions approximately ten times smaller (0.1 to 0.2 m m) have already been fabricated in research laboratories. So we still have at least a factor of 100 to go in terms of integrated-circuit density, provided manufacturable fabrication processes can be developed for these sub-micron dimensions.
Early fabrication used silicon wafers which had 1- and then 2-in. diameters. The size of the wafers has steadily increased to the point where 4-, 5-, and 6-in. wafers are now in production. Wafers with 8-in. diameters have been successfully produced by silicon wafer manufacturers.
The larger the diameter of the wafer, the more integrated-circuit dice can be produced at one time. Many wafers are processed at the same time. The same silicon chip is replicated as many times as possible on a silicon wafer of a given size. Processing costs per wafer are relatively independent of wafer size, so the cost per die is lower for larger wafer sizes. Thus there are strong economic forces driving the integrated-circuit industry to continually move to larger and larger wafer sizes.
Here we see a problem with the units of measure used to describe integrated circuits. Horizontal dimensions were originally specified in mil (1 mil = 0.001 in.), whereas specification of the shallower vertical dimensions commonly made use of the metric system. Today, most of the dimensions are specified using the metric system, although English units are occasionally still used. Throughout this site, we will attempt to make consistent use of metric units.