If a semiconductor crystal contains no impurities, the only charge carriers present are thos produced by thermal breakdown of the covalent bonds. The conducting properties are thus characteristic of the pure semiconductor. Such a crystal is termed an intrinsic semiconductor.
If a semiconductor crystal contains n-type or p-type impurities, the conducting properties are chiefly due to the impurities. Such a crystal is termed an extrinsic semiconductor.
When a silicon wafer is heated to about 1200 degrees Celsius in an atmosphere of water vapour or oxygen a skin of silicon dioxide forms on the surface. This skin is a most effective seal against the ingress of moisture at room temperatures and has made possible the method of manufacture of planar transistors which is described below.
A crystal of n-type silicon, about 1 inch in diameter, is cut into slices about 0.008 inches thick. The slices are lapped and etched to approximately 0.003 iches thickness and, if required, an epitaxial layer can be formed on one surface. The slices are now heated in an oxidising atmosphere to acquire a protective coating of silicon dioxide. At this stage each slice has a sectional view similar to that shown in (a) below. Each slice yields ultimately up to 1,000 transistors and the next stage is to mark off the individual transistors. This is achieved by a photo-lithographic process: each slice is coated in a dark room with a photo-sensitive material (known as photo-resist) and is then exposed to ultra-violet light via a mask containing an array of apertures corresponding to the base areas of the 1,000 transistors. The slice is now developed to remove the photo-resist from these regions thus exposing the silicon dioxide coating. Next the slice is treated with an etch which removes the silicon dioxide from the exposed regions. The remainder of the photo-resist is now dissolved: the cross-section of the slice now appears as in (b) below which shows a gap in the layer of silicon dioxide defining the base area for a single transistor.
The slice is now exposed at a high temperature to a boron-rich atmosphere. The silicon dioxide coating protects the slice against diffusion of boron except at the exposed areas and here boron diffuses isotropically, i.e. horizontally under the protective coating as well as vertically into the crystal, thus forming a p-type base region. Other more precise ways of forming such a region have been developed, for example by ion implantation. This involves a sharply defined bombardment of the substrate by a beam from an ion gun which enables the active base area to be closely controlled in area and shape, a process which can be compared with precision etching. The slice is now returned to the oxidising atmosphere and a coating of silicon dioxide is formed over the base areas (and the rest of the slice) to give a cross-section similar to that shown in (c) below.
The emitter areas are now defined by a similar process of masking, photo-lithography, exposure to ultra-violet light, etching, etc., and the silicon dioxide is removed from the emitter areas to give a cross-section such as that shown in (d) below. The slice is now heated whilst exposed to an atmosphere rich in phosphorus. This forms an n-type emitter region by diffusion and the exposed area is again sealed by heating the slice in an oxidising atmosphere to form a layer of silicon dioxide. See (e) below.
Holes are now made in the silicon dioxide coating as shown in (f) to permit ohmic contacts to be made to the base and emitter areas, the position of the holes being again determined by a mask. Contacts are then made to the transistors by a process of evaporation: the slice is placed in a vacuum chamber in which aluminium is evaporated, e.g. from a hot filamant. This results in a deposition of a thin coating of aluminium over the entire face of the slice. Finally the aluminium is removed from the areas in which it is not required by a masking and selective etching operation. The slice is now divided up into individual transistors and connections are made to the base and emitter regions of each transistor as shown in (g) below. The base area of each transistor is sometimes of approximately annular shape surrounding a circular emitter area but in power transistors both base and emitter areas may be in the form of parallel strips.
Planar transistors lend themselves well to mass production. Planar technology revolutionised silicon transistor manufacture in the 1960s. The transistors are particularly robust and the protection of the silicon dioxide coating is such that even without sealing in cans the transistors will operate well under boiling water! Leakage currents are very low and the transistors can be designed to work at frequencies well over 1 GHz. In 1963 the process also made possible for the first time mass production of f.e.t.s although the pronciple of this type of transistor had been described by Shockley 11 years earlier.
When a conventional pn diode is forward biased some of the majority carriers crossing the junction are neutralised by combination with majority carriers of opposite polarity. Others remain (as minority carriers) and, when the applied voltage is reversed, return across the junction in the form of a substantial pulse of reverse current which takes a significant time to decay to the normal value of reverse current. This delay is a serious disadvantage in diodes required for operation at microwave frequencies or in high-speed switching.
A schottky diode uses a metal-semiconductor contact instead of a pn junction, and this gives a diode with superior reverse recovery. For example, in one form of construction a region of epitaxial n-type GaAs is grown on a GaAs substrate and a metallic layer is deposited on this. Ohmic connections are made to the substrate and the metallic layer. Only one type of charge carrier is involved in operation of the diode. When the metal is biased positively electrons from the n-region are attracted to it to neutralise the charge so giving rise to the forward current. When the metal is negatively charged electrons are repelled and there is no reverse current. There is no p-layer in which electrons could be stored and the resulting diode is highly efficient at frequencies as high as 20GHz.
The invention of the transistor is attributed to William Shockley
When a pn junction is reverse-biased the current is carried solely by the minority carriers, and at a given temperature the number of minority carriers is fixed. Ideally, therefore, we would expect the reverse current for a pn junction to rise to a saturation value as the voltage is increased from zero and then to remain constant and independant of voltage, as shown below. In practice, when the reverse voltage reaches a particular value which can be 100V or more the reverse current increases very sharply, again shown below, an effect known as breakdown. The effect is reproducible, breakdown in a particular junction always occuring at the same value of reverse voltage. This is known as the Avalanche effect and reverse-biased diodes known as Avalanche diodes(sometimes called - perhaps incorrectly - Zener diodes) can be used as the basis of a voltage stabiliser circuit. The junction diodes used for this purpose are usually silicon types.
In certain semiconductors, notably GaAs, electrons can exist in a high-mass low velocity state as well as their normal low-mass high-velocity state and they can be forced into the high-mass state by a steady electirc field of sufficient strength. In this state they form clusters or domains which cross the field at a constant rate causing current to flow as a series of pulses. This is the Gunn effect and one form of diode which makes use of it consists of an epitaxial layer of n-type GaAs grown on a GaAs substrate. A potential of a few volts applied between ohmic contacts to the n-layer and substrate produces the electric field which causes clusters. The frequency of the current pulses so generated depends on the transit time through the n-layer and hence on its thickness. If the diode is mounted in a suitably tuned cavity resonator, the current pulses cause oscillation by shock excitation and r.f. power up to 1 W at frequencies between 10 and 30 GHz is obtainable.
As its name suggests, this is a junction diode with a region of intrinsic semiconductor between the n- and p- regions. When such a diode is reverse-biased the intrinsic layer is depleted of carriers and the diode behaves as a capacitor. When it is forward-biased carriers are injected into the intrinsic region to give a forward resistance which varies linearly between, say, 1 ohm and 10kOhms with the current through the device. This property makes the diode useful as a modulator or switch in microwave systems and at frequencies between 1 MHz and 20 GHz.