Transistors
Bipolar Junction Transistor. An n-p-n transistor may be considered to be an n-p diode back-to-back with a p-n diode. A schematic band diagram for an unbiased n-p-n transistor is shown in Fig. 8.25. The three connections of the transistor are called emitter (E), base (B), and collector (C).
Figure 8.25. Schematic band diagram of an unbiased n-p-n bipolar junction transistor.
If the transistor is used for the amplification of a signal, the "diode" consisting of emitter and base is forward biased, whereas the base-collector "diode" is strongly reverse biased (Fig. 8.26(a)). The electrons injected into the emitter, therefore, need to have enough energy to be able to "climb" the potential barrier into the base region. Once there, the electrons diffuse through the base area until they have reached the depletion region between base and collector. Here, the electrons are accelerated in the strong electric field produced by the collector voltage (Fig. 8.26(b)). This acceleration causes amplification of the input a.c. signal.
One may consider this amplification from a more quantitative point of view. The forward biased emitter-base diode is made to have a small resistivity (approximately 10-3 O cm), whereas the reverse biased base-collector diode has a much larger resistivity (about 10 O cm). Since the current flowing through the device is practically identical in both parts, the power (P = I2R) is larger in the collector circuit. This results in a power gain.
The electron flow from emitter to collector can be controlled by the bias voltage on the base: a large positive (forward) bias decreases the potential barrier and the width of the depleted region between emitter and base (Fig. 8.19). As a consequence, the electron injection into the p-area is relatively high. In contrast, a small, but still positive base voltage results in a comparatively larger barrier height and in a wider depletion area, which causes a smaller electron injection from the emitter into the base area. In short, the voltage applied between emitter and base modulates the transfer of the electrons from the emitter into the base area. As a consequence, the strong collector signal mimics the waveform of the input signal. This feature is utilized for the amplification of music or voice, etc.
In another application, a transistor may be used as an electronic switch. The electron flow from emitter to collector can be stopped completely (or turned on) by an appropriate base voltage. This virtue is used for logic and memory functions in computers (see Section 8.7.12).
Figure 8.26. (a) Biasing of an n-p-n bipolar transistor. (b) Schematic band diagram (partial) of a biased n-p-n bipolar transistor. (c) Symbol used for a bipolar n-p-n transistor.
Figure 8.27. Schematic collector voltage-current characteristics of a transistor for various emitter currents. Ic = collector current, Ie = emitter current, and Vc = collector voltage.
The device shown in Fig. 8.26 is called a "bipolar transistor"; the current passes in series through n-type as well as through p-type semiconductor materials.
Some details need to be added about technical features of the bipolar transistor. In order to obtain a large electron density in the emitter, this area is heavily doped. In the p-doped base area, the drifting electrons are subject to possible recombination with holes. Therefore, the number of holes there has to be kept to a minimum, which is accomplished by light doping. (Light doping also reduces the unwanted injection of hole current into the base.) Recombination is further decreased by making the base region extremely thin, i.e., 10 5-10 7 m. A narrow base region has a beneficial side effect: it increases the frequency response. (The reciprocal of the electron transit time equals the highest possible frequency at which amplification can be achieved.) The doping rate of the collector area is in general not critical. Usually, the doping is light for high gain and low capacitance of the device. The voltage-current characteristics for a transistor are shown in Fig. 8.27.
In p-n-p transistors, the majority carriers are holes. The function and features of a p-n-p transistor are similar to an n-p-n transistor.
Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET). A field-effect transistor consists of a channel through which the charge carriers (e.g., electrons in Fig. 8.28) need to pass on their way from a source (S) to the drain (D). The conducting path (source, channel, and drain) is made of the same kind of semiconducting material only, e.g., n-type. (This is in contrast to the bipolar transistor shown in Fig. 8.26, in which the current passes in series through n-type as well as through p-type semiconductor materials.) Field-effect transistors are therefore designated as unipolar. The electrons that flow from the source to the drain can be controlled by an electric field which is established by applying a voltage to the so-called gate (G).
A periodic variation of the gate voltage varies the source to drain current in the same manner (quite similar to the way the electron flow between emitter and collector in a bipolar transistor is modulated by the base voltage).
Figure 8.28. (a) Schematic representation of an n-channel depletion- (normally on) type MOSFET. The dark areas symbolize the (aluminum) metallizations. The "oxide" layer may consist of SiO2, nitrides (Si3N4), oxinitrides (Si3N4-SiO2), or multilayers of these substances. This layer is about 10 nm thick. The gate voltage is applied between terminals G and B. Quite often the B and S terminals are interconnected. (b) Circuit symbol for n-channel depletion-type MOSFET. (c) Gate voltage/Drain current characteristic ("Transfer" characteristic). For positive gate voltages (dashed portion of the curve) the device can operate in the "enhancement mode" (see Fig. 8.29(c)).
The gate electrode is electrically insulated from the channel by a thin oxide layer which prevents a d.c. current to flow from gate to channel.
Two types of MOSFETs are common; the depletion-type MOSFET depicted in Fig. 8.28(a) consists of high-doped source and drain regions and a low-doped channel, all of the same polarity (e.g. n-type). (The high doping facilitates low-resistance connections.) The n-channel MOSFET is laid down on a p-type substrate called the body.
The channel width is controlled by the voltage between gate and body. Specifically, a negative charge on the gate drives the channel electrons away from the gate and towards the substrate, similarly as is illustrated in Fig. 8.12. In short, the channel can be made to be partially depleted of electrons, i.e., the conductive region of the channel becomes narrowed by a negative gate voltage. The more negative the gate voltage (VG), the smaller the current through the channel from source to drain until eventually the current is pinched off (see Fig. 8.28(c).) For the above reasons, this device is called a depletion-type metal-oxide semiconductor field-effect transistor or "normally on" MOSFET.
Figure 8.29. (a) Enhancement (normally-off)-type n-channel MOSFET. For details, see the caption of Fig. 8.28. (b) Circuit symbol. (The broken line indicates that the path between S and D is normally interrupted.) (c) Gate voltage (VG)/drain current (ID) characteristic. VT is the threshold gate voltage above which a drain current sets in.
An alternative to the depletion-type MOSFET that we just discussed is the enhancement-type MOSFET. Figure 8.29 shows that this device does not possess a built-in channel for electron conduction, i.e., at least as long as no gate voltage is applied. In essence, there is no electron flow from source to drain for a zero gate voltage. The device is therefore called a "normally-off" MOSFET. If, however, a large enough positive voltage is applied to the gate, most of the holes immediately below the gate oxide are repelled, i.e., they are driven into the substrate, thus removing possible recombination sites. Concomitantly, negative charge carriers are attracted into this channel (called the inversion layer). In short, a path (or a bridge) for the electrons between source and drain can be created by a positive gate voltage. The metal-oxide semiconductor technology, particularly, the enhancement-type MOSFETs, dominate the integrated circuit industry at present. They are utilized in memories, microcomputers, logic circuits, amplifiers, analog switches, and operational amplifiers. They possess very high input impedances,15 thus minimizing Joule heating.
Depletion-type and enhancement-type MOSFET technologies that utilize n-channels (as depicted in Figs. 8.28 and 8.29) are summarized by the name "NMOSFET" (in contrast to "PMOSFET", which employs devices with p-channels). If both an n-channel and a p-channel device are integrated on one chip and wired in series, the technology is labeled "CMOSFET" which stands for complementary MOSFET. This tandem device has become the dominant technology for information processing, because of its low operating voltage (0.1 V), low power consumption (heat!), and short channel length with accompanying high speed. Alternative names for MOSFET are MOST (metal-oxide-semiconductor transistor) or MISFET (metal-insulator-semiconductor field-effect transistor).
A few words on device geometry, etc., of a MOSFET, as shown in Fig. 8.28, may be useful. In order to obtain a short switching time and a high-frequency response, the channel length has to be short. The highest possible frequency at which amplification can be achieved equals the inverse of the electron source-to-drain transit time. The width of the device has to be kept small in order to reduce the cross-sectional area and, thus, the power density. (This reduces the heat which needs to be removed.) As an example, the channel length may be about 1 mm, the device width may be a few micrometers, and the field oxide thickness may be near 0.05 mm. The doping of the p-area needs to be small to sustain a high resistance and thus, a high electric field 106 V/cm) across the junction without current breakdown. The metal layer is generally made of aluminum. Alternate materials are highly doped silicon, refractory metals such as tungsten, or silicides of refractory metals such as TiSi or MoSi.
* Junction Field-Effect Transistor (JFET). The JFET consists again of a channel through which the carriers (electrons in Fig. 8.30) pass from source to drain. This electron flow is controlled by an electric field which is established by applying a negative voltage to the p-doped gate, to stay within the example of Fig. 8.30. In other words, the p-n gate-to-channel diode is reverse biased. This reverse biasing increases the width of the depletion layer (see Fig. 8.19) thus causing the conducting channel to become narrower. (Close to the drain terminal, the p-n junction is more reverse biased which results in a wider depletion layer near the drain.) A zero bias voltage on the gate results in a maximal source-to-drain current. A reverse voltage on the gate depletes the source-to-drain electron flow. A very large reverse current eventually pinches the current off. Junction field-effect transistors are therefore said to be of the depletion or "normally-on" type.
Junction field-effect transistors can be used as amplifiers, exploiting the effect that a small change in the gate voltage causes a large change in the channel current. Since the gate-to-channel p-n junction is reverse biased, only a minute current flows in the gate/source circuit (Fig. 8.16). The input impedance15 is therefore high (but not as high as in a MOSFET).
Figure 8.30. (a) Schematic representation of an n-channel junction field-effect transistor. The dark areas symbolize the metal contacts (e.g., aluminum). (b) Circuit symbol for an n-channel JFET. Note: In a p-channel JFET the arrow points away from the channel.
JFETs which use n-type semiconductors for the channel material, as depicted in Fig. 8.30, are appropriately called n-channel field-effect transistors. The reader may correctly suspect that ap-channel field-effect transistor uses holes as charge carriers, n-type semiconductors as gate materials, and a reversal of the polarities of all voltages for its operation. The arrow in the circuit symbol (Fig. 8.30(b)) for p-channel transistors points away from the gate.
Bipolar transistors in combination with JFETs are called "BIFETs." They are used in high-performance linear circuits. If a JFET structure employs a metal- semiconductor junction, often in combination with n-type GaAs, a "MESFET" device is created, which is used for amplifiers and logic circuits in the gigahertz range (see next section).
A MODFET (modulation-doped field-effect transistor) consists of a thin layer of aluminum-gallium-arsenide deposited on an undoped GaAs substrate. This device is even faster than a MESFET, because the absence of impurity atoms increases the distance that an electron or a hole can travel before a collision with a foreign atom occurs.
* Gallium Arsenide Metal-Semiconductor Field-Effect Transistor (MESFET). Users of computers demand still higher switching speeds than the present 10"9 s cut-off or cut-on times achieved with silicon technology. Gallium arsenide, with its almost sixfold larger electron mobility compared to silicon (see topic 4), seems to be the answer. A quick inspection of the relevant band diagrams (Figs. 5.23 and 5.24) indeed confirms that the curvature of the conduction band near r is larger for GaAs than the comparable band for silicon (close to the X symmetry point) which translates into a smaller effective mass and, thus, into the just-mentioned larger electron mobility for GaAs. However, the upper valence bands for both materials are almost identical and fairly flat. Thus, the effective masses of the holes for GaAs and silicon are rather large and their hole mobilities are consequently small (see also topic 4). A transistor that aims to exploit the higher electron mobility in GaAs should therefore utilize n-type GaAs only.
Figure 8.31 depicts a metal-semiconductor field-effect transistor (MESFET), which consists of an n-doped, thin GaAs active layer situated over a semi-insulating (Cr-doped) GaAs slab. Three metal contacts provide the source, the gate, and the drain areas. The gate metal forms, together with the underlying semiconductor, a Schottky barrier (see Section 8.7.2). If f M is larger than f S and the gate metal is negatively charged, a reverse bias results (Fig. 8.15(a)). The larger the reverse bias, the wider the depletion region. If the depletion region is caused to fill essentially the entire active layer, any attempted electron flow from source to drain is stopped (or pinched off). A small negative gate voltage (or no gate voltage at all) allows an almost unhindered source-to-drain electron flow. The device shown in Fig. 8.31 is therefore a depletion- (or normally-on) type FET (see also Fig. 8.28(c)).
For high-speed, low-power applications, however, the normally-off GaAs MESFET is even better suited. For this device, the active layer is made so thin that the depletion area between the metal and the GaAs (Fig. 8.15) fills the entire active layer.16 As a consequence, the active layer below the gate metal electrode is depleted of electrons without necessitating an applied voltage.
Figure 8.31. Schematic representation of a GaAs MESFET (Metal-semiconductor field-effect transistor). Source and drain metallizations (dark areas) are selected to form ohmic contacts with the n-doped GaAs. The gate metal forms, with the n-doped GaAs, a Schottky-barrier contact.
A positive gate voltage is then required to attract electrons into the depletion area, thus making it conductive. Given the above-described GaAs device, the speed, i.e., the response time of the source-to-drain current to a change in the gate voltage, can be further increased by decreasing the length of the gate, which is presently about 1 mm.
Several effects may, however, offset the superior electron mobility in GaAs. First, the time required to reach the breakdown voltage under the influence of a reverse voltage (see Fig. 8.20(c)) is only two and a half times faster than in silicon. As we know from Fig. 8.20(a), this breakdown electric field triggers a helpful self-ionizing avalanche that multiplies the number of electrons. Second, a transistor of any type can be made to switch faster by applying more power to it. This, in turn, increases the heat which needs to be dissipated. Now, silicon has a three-times larger thermal conductivity than GaAs (see topic 4). Thus, silicon switches can be made much smaller than those made of GaAs. Since the speed of a device also depends on the length the electrons have to travel, a very small silicon device may well switch as fast as a large device made of GaAs. Third, the electron drift velocity depends upon the electric field strength. At low field strengths, the GaAs drift velocity is indeed substantially larger than for silicon (Fig. 8.32). However, as the field strength increases, the drift velocity for silicon and GaAs becomes nearly identical. This has its reason in the extra and slightly higher energy states that silicon possesses near the X-symmetry point (Fig. 5.23), in which electrons can be scattered after they have collided with structural imperfections of the crystal lattice.
Knowing the facts presented above, it seems understandable why some leading semiconductor manufacturers have left the GaAs field. However, the pendulum may soon swing in the other direction, as suggested in the next section.
Figure 8.32. Average electron drift velocity as a function of electric field strength for GaAs and silicon.
