IAN LANG ELECTRONICS
Resistors on an IC.
There are two methods of producing resistors on an integrated circuit; one is to print them on film and lay them on top of the monolithic component, and this is known as a hybrid IC. This method is promising for the future, but is currently not without its problems. The other is to diffuse them directly and this is the method we are going to discuss now.
Such a resistor would look like this:
As you can see, it is an n layer sandwiched between two p layers, the lower surrounding p layer being the substrate. The n layer is diffused into the p usually as an epitaxial layer (more on this later) and the second p is diffused into the n. A layer of silicon oxide is deposited on top for insulation and that is followed by a layer of aluminium which is etched to form the connections.
The value of the resistor depends on the length of p type material between the aluminium contacts, and a figure of 10 0hms per micrometre is to be expected if the width of the p strip is 10 micrometres. The width of the strip can be made wider if a larger current is to be put through the IC.
The depth of the p material is largely irrelevant for resistors. It is diffused at the same time as the base region of the transistors and so the depth depends on the transistor design.
Because putting more or less components on an IC does not significantly alter the cost of producing it it is most economically produced by putting as many components as possible in a given space. Since the value of a resistor depends on its length, a maximum length is set, perhaps 50 kilohms. Resistors can be made to meander around in square patterns if a larger value is desired but this causes space issues and if not done carefully may cause stray capacitance too.
A pinch resistor has an n+ layer diffused into it to increase resistance but this type is difficult and expensive to make. In some cases a resistor can be made from the input resistance of a transistor with a small resistance R between emitter and ground; such a device uses less space and the resistance is hfeR ohms. Because it is not always possible to accurately predict the gain, the resistor made this way may not be accurate either.
IC resistors are not high precision resistors, even in the best circumstances it is not possible to make an individual resistor better than 5% accurate and a standard is 20%, but since they are all formed at the same time a resistor network can be made with any two equivalent resistors that is some 1% accurate and thus IC resistance is determined by comparator resistance networks and not the individual resistors themselves.