Microwave solid-state devices

Passive devices

A PIN (p-type/intrinsic/n-type) diode is a pn diode that has an undoped (intrinsic) region between the p- and n-type regions. The use of an intrinsic region in PIN diodes allows for high-power operation and offers an impedance at microwave frequencies that is controllable by a lower frequency or a direct-current (DC) bias. The PIN diode is one of the most common passive diodes used at microwave frequencies. PIN diodes are used to switch lengths of transmission line, providing digital increments of phase in individual transmission paths, each capable of carrying kilowatts of peak power. PIN diodes come in a variety of packages for microstrip and stripline packages, and are used as microwave switches, modulators, attenuators, limiters, phase shifters, protectors, and other signal control circuit elements.

A Schottky barrier diode (SBD) consists of a rectifying metal-semiconductor barrier typically formed by deposition of a metal layer on a semiconductor. The SBD functions in a similar manner to the antiquated point contact diode and the slower-response pn-junction diode, and is used for signal mixing and detection. The point contact diode consists of a metal whisker in contact with a semiconductor, forming a rectifying junction. The SBD is more rugged and reliable than the point contact diode. The SBD's main advantage over pn diodes is the absence of minority carriers, which limit the response speed in switching applications and the high-frequency performance in mixing and detection applications. SBDs are zero-bias detectors. Frequencies to 40 GHz are available with silicon SBDs, and GaAs SBDs are used for higher-frequency applications.

The variable-reactance (varactor) diode makes use of the change in capacitance of a pn junction or Schottky barrier diode, and is designed to be highly dependent on the applied reverse bias. The capacitance change results from a widening of the depletion layer as the reverse-bias voltage is increased. As variable capacitors, varactor diodes are used in tuned circuits and in voltage-controlled oscillators. For higher-frequency microwave applications, silicon varactors have been replaced with GaAs. Typical applications of varactor diodes are harmonic generation, frequency multiplication, parametric amplification, and electronic tuning. Multipliers are used as local oscillators, low-power transmitters, or transmitter drivers in radar, telemetry, telecommunication, and instrumentation. See Varactor

Active devices

Transistors are the most widely used active microwave solid-state devices. At very high microwave frequencies, high-frequency effects limit the usefulness of transistors, and two-terminal negative resistance devices, such as transferred-electron devices, avalanche diodes, and tunnel diodes, are sometimes used. Two main categories of transistors are used for microwave applications: bipolar junction transistors (BJTs) and field-effect transistors (FETs). In order to get useful output power at high frequencies, transistors are designed to have a higher periphery-to-area ratio using a simple stripe geometry. The area must be reduced without reducing the periphery, as large area means large interelectrode capacitance. For high-frequency applications the goal is to scale down the size of the device. Narrower widths of the elements within the transistor are the key to superior high-frequency performance. See Transistor

A BJT consists of three doped regions forming two pn junctions. These regions are the emitter, base, and collector in either an npn or pnp arrangement. Silicon npn BJTs have an upper cutoff frequency of about 25 GHz (varies with manufacturing improvements). The cutoff frequency is defined as the frequency at which the current amplification drops to unity as the frequency is raised. The primary limitations to higher frequency are base and emitter resistance, capacitance, and transit time. To operate at microwave frequencies, individual transistor dimensions must be reduced to micrometer or submicrometer size. To maintain current and power capability, various forms of internal paralleling on the chip are used. Three of these geometries are interdigitated fingers that form the emitter and base, the overlaying of emitter and base stripes, and the matrix approach. Silicon BJTs are mainly used in the lower microwave ranges. Their power capability is quite good, but in terms of noise they are inferior to GaAs metal semiconductor field-effect transistors (MESFETs) at frequencies above 1 GHz and are mainly used in power amplifiers and oscillators. They may also be used in small-signal microwave amplifiers when noise performance is not critical.

Heterojunction bipolar transistors (HBTs) have been designed with much higher maximum frequencies than silicon BJTs. HBTs are essentially BJTs that have two or more materials making up the emitter, base, and collector regions (Fig. 1). In HBTs, the major goal is to limit the injection of holes into the emitter by using an emitter material with a larger bandgap than the base. The difference in bandgaps manifests itself as a discontinuity in the conduction band or the valence band, or both. For npn HBTs, a discontinuity in the valence band is required. In general, to make high-quality heterojunctions, the two materials should have matching lattice constants. For very thin layers, lattice matching is not absolutely necessary as the thin layer can be strained to accommodate the crystal lattice of the other material. Fortunately, the base of a bipolar transistor is designed to be very thin and thus can be made of a strained layer material. Combinations such as AlGaAs/InGaAs and Si/SiGe are possible. See Semiconductor heterostructures

Field-effect transistors (FETs) operate by varying the conductivity of a semiconductor channel through changes in the electric field across the channel. The three basic forms of FETs are the junction FET (JFET), the metal semiconductor FET (MESFET), and the metal oxide semiconductor FET (MOSFET). All FETs have a channel with a source and drain region at each end and a gate located along the channel, which modulates the channel conduction (Fig. 2). Microwave JFETs and MESFETs work by channel depletion. The channel is n-type and the gate is p-type for JFETs and metal for MESFETs. FET structures are well suited for microwave applications because all contacts are on the surface to keep parasitic capacitances small. The cutoff frequency is mainly determined by the transit time of the electrons under the gate; thus short gate lengths (less than 1 μm) are used.

Power devices consist of a number of MESFETs in parallel with air bridges connecting the sources. GaAs MESFET devices are used in low-noise amplifiers (LNAs), Class C amplifiers, oscillators, and monolithic microwave integrated circuits. The performance of a GaAs FET is determined primarily by the gate width and length. The planar structure of a MESFET makes it straightforward to add a second gate which can be used to control the amplification of the transistor. Dual-gate MESFETs can be used as mixers (with conversion gain) and for control purposes. Applications include heterodyne mixers and amplitude modulation of oscillators. See Amplifier, Mixer, Oscillator

The MOSFET has a highly insulating silicon dioxide (SiO₂) layer between the semiconductor and the gate; however, silicon MOSFETs are not really considered microwave transistors. Compared with the GaAs MESFET, MOSFETs have lower electron mobility, larger parasitic resistances, and higher noise levels. Also, since the silicon substrate cannot be made semi-insulating, larger parasitic capacitances result. MOSFETs therefore do not perform very well above 1 GHz. Below this frequency, MOSFETs find application mainly as radio-frequency (RF) power amplifiers.

A disadvantage of the MESFET is that the electron mobility is degraded since electrons are scattered by the ionized impurities in the channel. By using a heterojunction consisting of n-type AlGaAs with undoped GaAs, electrons move from the AlGaAs to the GaAs and form a conducting channel at the interface. The electrons are separated from the donors and have the mobility associated with undoped material. A heterojunction transistor made in this fashion has many different names: high electron mobility transistor (HEMT), two-dimensional electron gas FET (TEGFET), modulation-doped FET (MODFET), selectively doped heterojunction transistor (SDHT), and heterojunction FET (HFET). The HEMT has high power gain at frequencies of 100 GHz or higher with low noise levels.

A monolithic microwave integrated circuit (MMIC) can be made using silicon or GaAs technology with either BJTs or FETs. For high-frequency applications, GaAs FETs are the best choice. A MMIC has both the active and passive devices fabricated directly on the substrate. MMICs are typically used as low-noise amplifiers, as mixers, as modulators, in frequency conversion, in phase detection, and as gain block amplifiers. Silicon MMIC devices operate in the 100-MHz to 3-GHz frequency range. GaAs FET MMICs are typically used in applications above 1 GHz.

Active microwave diodes

Active microwave diodes differ from passive diodes in that they are used as signal sources to generate or amplify microwave frequencies. These include step-recovery, tunnel, Gunn, avalanche, and transit time diodes, such as impact avalanche and transit-time (IMPATT), trapped plasma avalanche triggered transit-time (TRAPATT), barrier injection transit-time (BARITT), and quantum well injection transit time (QWITT) diodes.

A step recovery diode is a special PIN type in which charge storage is used to produce oscillations. When a diode is switched from forward to reverse bias, it remains conducting until the stored charge has been removed by recombination or by the electric field. A step recovery diode is designed to sweep out the carriers by an electric field before any appreciable recombination has taken place. Thus, the transition from the conducting to the nonconducting state is very fast, on the order of picoseconds. Because of the abrupt step, this current is rich in harmonics, so these diodes can be used in frequency multipliers.

For microwave power generation or amplification, a negative differential resistance (NDR) characteristic at microwave frequencies is necessary. NDR is a phenomenon that occurs when the voltage (V) and current (I) are 180° out of phase. NDR is a dynamic property occurring only under actual circuit conditions; it is not static and cannot be measured with an ohmmeter. Transferred electron devices (TEDs), such as Gunn diodes, and avalanche transit-time devices use NDR for microwave oscillation and amplification. TEDs and avalanche transit-time devices today are among the most important classes of microwave solid-state devices. See Negative-resistance circuits

The tunnel diode uses a heavily doped abrupt pn junction resulting in an extremely narrow junction that allows electrons to tunnel through the potential barrier at near-zero applied voltage. This results in a dip in the current-voltage (I-V) characteristic, which produces NDR. Because this is a majority-carrier effect, the tunnel diode is very fast, permitting response in the millimeter-wave region. Tunnel diodes produce relatively low power. The tunnel diode was the first semiconductor device type found to have NDR. See Tunnel diode

Avalanche diodes are junction devices that produce a negative resistance by appropriately combining impact avalanche breakdown and charge-carrier transit time effects. Avalanche breakdown in semiconductors occurs if the electric field is high enough for the charge carriers to acquire sufficient energy from the field to create electron-hole pairs by impact ionization. The avalanche diode is a pn-junction diode reverse-biased into the avalanche region. By setting the DC bias near the avalanche threshold, and superimposing on this an alternating voltage, the diode will swing into avalanche conditions during alternate half-cycles. The hole-electron pairs generated as a result of avalanche action make up the current, with the holes moving into the p region, and the electrons into the n region. The carriers have a relatively large distance to travel through the depletion region. At high frequencies, where the total time lag for the current is comparable with the period of the voltage, the current pulse will lag the voltage. By making the drift time of the electrons in the depletion region equal to one-half the period of the voltage, the current will be 180° out of phase. This shift in phase of the current with respect to the voltage produces NDR, so that the diode will undergo oscillations when placed in a resonant circuit.

A Gunn diode is typically an n-type compound semiconductor, such as GaAs or InP, which has a conduction band structure that supports negative differential mobility. Although this device is referred to as a Gunn diode, after its inventor, the device does not contain a pn junction and can be viewed as a resistor below the threshold electric field (E_thres). For applied voltages that produce electric fields below E_thres, the electron velocity increases as the electric field increases according to Ohm's law. For applied voltages that produce electric fields above E_thres, conduction band electrons transfer from a region of high mobility to low mobility, hence the general name “transferred electron device.’’ Beyond E_thres, the velocity suddenly slows down due to the significant electron transfer to a lower mobility band producing NDR. For GaAs, E_thres is about 3 kV/cm. The Gunn effect can be used up to about 80 GHz for GaAs and 160 GHz for InP. Two modes of operation are common: nonresonant bulk (transit-time) and resonant limited space-charge accumulation (LSA).

Impact avalanche and transit-time diodes (IMPATTs) are NDR devices that operate by a combination of carrier injection and transit time effects. There are several versions of IMPATT diodes, including simple reverse-biased pn diodes, complicated reverse-biased multidoped pn layered diodes, and reverse-biased PIN diodes. The IMPATT must be connected to a resonant circuit. At bias turn-on, noise excites the tuned circuit into a natural oscillation frequency. This voltage adds algebraically across the diode's reverse-bias voltage. Near the peak positive half-cycle, the diode experiences impact avalanche breakdown. When the voltage falls below this peak value, avalanche breakdown ceases. A 90° shift occurs between the current pulse and the applied voltage in the avalanche process. A further 90° shift occurs during the transit time, for a total 180° shift which produces NDR. An IMPATT oscillator has higher output power than a Gunn equivalent. However, the Gunn oscillator is relatively noise-free, while the IMPATT is noisy due to avalanche breakdown.

A trapped plasma avalanche triggered transit-time (TRAPATT) diode is basically a modified IMPATT diode in which the holes and electrons created by impact avalanche ionization multiplication do not completely exit from the transit domain of the diode during the negative half-cycle of the microwave signal. These holes and electrons form a plasma which is trapped in the diode and participates in producing a large microwave current during the positive half-cycle.

A barrier injection transit-time diode (BARRITT) is basically an IMPATT structure that employs a Schottky barrier formed by a metal semiconductor contact instead of a pn junction to create similar avalanche electron injection.

A variety of approaches have been investigated to find alternative methods for injecting carriers into the drift region without relying on the avalanche mechanism, which is inherently noisy. Quantum well injection transit-time diodes (QWITT) employ resonant tunneling through a quantum well to inject electrons into the drift region. The device structure consists of a single GaAs quantum well located between two AlGaAs barriers in series with a drift region of made of undoped GaAs. This structure is then placed between two n⁺-GaAs regions to form contacts.