Introduction to negative resistance 1 devices

onant oscillator circuits use no tuning. They depend on the device dimensions and the average charge carrier velocity through the bulk material to determine operating frequency. Some NDR devices, such as the Gunn diode (later in this chapter), will operate in either resonant or unresonant oscillatory modes.

Transferred electron devices Gunn diodes)

Gunn diodes are named after John B. Gunn of IBM, who, in 1963, discovered a phenomenon since then called the Gunn Effect Experimenting with compound semiconductors such as gallium arsenide (GaAs), Gunn noted that the current pulse became unstable when the bias voltage increased above a certain crucial threshold potential. Gunn suspected that a negative resistance effect was responsible for the unusual diode behavior. Gunn diodes are representative of a class of materials called transferred electron devices (TED).

Ordinary two-terminal small-signal PN diodes are made from elemental semiconductor materials, such as silicon (Si) and germanium (Ge). In the pure form before charge carrier doping is added, these materials contain no other elements, TED devices, on the other hand, are made from compound semiconductors; that is, materials that are chemical compounds of at least two chemical elements. Examples are gallium arsenide (GaAs), indium phosphide (InP), and cadmium telluride (CdTe). Of these, GaAs is the most commonly used.

N-type GaAs used in TED devices is doped to a level of 1014 to 1017 carriers per centimeter at room temperature (373°K). A typical sample used in TED/Gunn devices will be about 150 X 150 jjuM in cross-sectional area and 30 jiuM long.

Two-valley TED model

The operation of TED devices depends on a semiconductor phenomenon found in compound semiconductors called the two-trolley model (for InP material, there is also a three-valley model). The two-valley model is also called the Ridley/ Wat kins/ Hilsum theory. Figure 22-6 shows how this model works in TEDs, In ordinary elemental semiconductors, the energy diagram shows three possible bands: valence band, forbidden band, and conduction band- The forbidden band contains no allowable energy states so it contains no charge carriers. The difference in potential between valence and conduction bands defines the forbidden band, and this potential is called the band/jap voltage.

In the two-valley model, however, there are two regions in the conduction band in which charge carriers (e.g., electrons) can exist. These regions are called valleys and are designated the upper valley and the lower valley. According to the RWH theory, electrons in the !ower valley have low effective mass (0,068) and consequently a high mobility (8000 cmW-s). in the upper valley, which is separated from the lower valley by a potential of 0.036 electron volts (eV)> electrons have a much Itigher effective mass (1.2) and lower mobility (180 cm2/V-s) than in the lower valley.

At low electric field intensities (0 to 3,4 kV/cm), electrons remain in the lower valley and the material behaves ohmicaUy, At these potentials, the material exhibits positive differential resistance (PDR). At a certain critical threshold potential (IV), electrons are swept from the lower valley to the upper valley (hence the name transferred electron devices). For GaAs the electric field must be about 3.4 kV/cm, so VIu,

Two Valley Theory Gunn Diode
22-6 Transferred electron diode operation.

is the potential that produces this strength of field. Beraus^ is (lie pmdiurt <if the electric field potential and device length, a 10-m,M sample of (¡aAs will have n threshold potential of about 3.4 V. Mosl GaAsTED devices operate at maximum dr poteit-tials in the 7- to 10-V range.

The average velocity of tamers in a two-valley seink-omlnrtoi <,such as tiaAs) is a function of charge mobility in each valley and tlie relative numbers ofelec unn^ in each valley. If ail electrons are in the lower valley then the material is in J ho hi.uh.eiit average veltxrity state. Conversely, if all electrons were in the upper valley, then «ho average velocity is in its lowest state Figure 112-7.A shows drift velocity as a funriiojj of electric field, or dc bias.

in the PDR region, the GaAs material is ohmic, and the drift velocity increases linearly with increasing potential. It would continue to increase until the saturation

Gunn Diode Bias Circuit Negative Resistance Region Gunn Diode

22-7 Drifl velocity vs electric fin It I, (lij /-vs-/l(Htl and (C) diode current v.s time.

Gunn Diode Bias Circuit

velocity is reached (about. 107 cm/s). As the voltage increases alx>ve threshold potential, which creates fields greater than kV/cm, more and more electrons are transferred to the upper valley so that the average drift velocity drops. Tlris phenomena gives rise to the negative resistance (NDR) effect. Figure 22-7R shows the NDR effect in the I-vs-V characteristic.

Figure 22-7C shows the /-vs-T characteristic of the Gunn diode operating in the NDR region. Ordinarily, you would expect a smooth current pulse to propagate through the material. But notice the oscillations (i.e., Uunn's instabilities) superimposed on the pulse. This oscillating current makes the Gunn diode useful as a microwave generator.

For the two-valley model to work, several criteria must be satisfied. First, the energy difference between lower and upper valleys must be greater than the thermal energy (A» of the material; At is about 0,026 eV, so GaAs with 0,036 eV differential energy satisfies the requirement. Second, in order To prevent hole-electron pair formation the differential energy between valleys must be less than the forbidden band energy (i.e., the bandgap potential). Third, electrons in the lower valley must, have high mobility, low density of state, and low effective mass. Finally, electrons in the uppeT valley must be jusL the opposite: low mobility, high effective mass, and a high density of state. It is sometimes claimed that ordinary devices use so-called '"warm" electrons (i.e., 0.026 eV) and TED devices use ''hot" electrons greater than 0,026 eV.

Gunn diodes

The Gunn diode is a transferred electron device that is capable of oscillating in several modes. In the uriresonant transit-time (TT> mode, frequencies between 1 and 18 GHz are achieved, with output powers up to 2 W (most are on the order of a few hundred milliwatts). In the resonant limited space-charge (LSA) mode, operating frequencies to 100 GHz and pulsed power levels to several hundred watts (1% duty cycle) have been achieved.

Figure 22-8A shows a diagram for a Gunn diode, and an equivalent circuit is shown in fig, 22-8B. The active region of the diode is usually 6 to 18 long- The N* end regions are ohmic materials of very low resistivity (0.001 Q-cm), and are 1 to 2

Gunn Diode

22-8 (A) Gunn diode and (B) equivalent circuit.

chick. Tile function of the N* regions is to form a transition zone between the metallic end electrodes and the active region. In addition to improving the contact, the N* regions prevent migration of metallic ions front the electrode into the active region.

Domain growth

The mechanism underlying the oscillations of a Guttn diode is the growth of Ridley domains in the active region of the device. An electron domain is created by a 1 unching effect (Fig. 22-9) that moves from the cathode end to the anode end of

22-9 Domain formation and propagation.

the active region. When an old domain is collected at the anode, a new domain forms at the cathode and begins propagating. The probation velocity is close to (he saturation velocity (10T cm/s). The time required for a domain to travel the length (L) of the material is called the transit tune (Twhich is:


Tt = the transit time in seconds (s) L * the length in centimeters (cm) Vs = the saturation velocity (107 cm/s)

Figure 22-10 graphically depicts domain formation. You might recognise lltat the "domain" shown here is actually a dipole, or double domain. The area ahead of ihe domain forms a minor depletion zone, and in the area of the domain, there is a bunching of electrons. Thus, there is a difference in conductivity between the two regions, and this conductivity is different still from the conductivity of the rest of the active region. There is also a difference between the electric fields in t he two domain i>oles and also tn the rest of the material. The two fields equilibrate outside of the domain. The current density is proportional to the velocity of the domain, so a current pulse forms.

Depletion of Charge

Depletion of Charge

Excess Charge

Eléctricas Charges

22-10 Domain formation in semiconductor material-

Gttnn operating modes

Transferred electron devices (i.e., Gunn diodes) operate in several modes and submodes, These modes depend in part on device characteristics and in part on external circuitry. For the Gunn diode, the operating modes are stable amplification (SA) mode, transit time (TT) mode, limited space-charge (LSA) mode, and bias circuit oscillation (BCO) mode.

Stable amplification {SA} modes in this mode, the Gunn diode will behave as an amplifier. The requirement for S A-mode operation is that the product of doping concentration (N^) and effective length of the active region (L) must be less than 1012/cm2. Amplification is limited to frequencies in the vicinity of v/L, where v is the domain velocity and L is the effective length.

Transit time (Gunn) mode The transit time (TT), or Gunn, mode is unreso-nant and depends on device length and applied dc bias voltage. The dc potential must be greater than the critical threshold potential (tV) Because of the Gunn effect, current oscillations in the microwave region are superimposed on the current pulse. Operation in this mode requires that the NaL product be 1012/cma to 1014/cms. Operating frequency F0 is determined by device length, or rather the transit time of the pulse through the length of the material. Because domain velocity is nearly constant, and is often close to the electron saturation velocity (Fs = 10" cm/s), length and transit time are proportional to each other. The operating frequency is inversely proportional to both the length of the device and the transit time. The length of the active region in the Gunn diode determines operating frequency. The frequency varies from about 6 GHz for an iS-jwn sample (counting 1.5 jwn for each N+ electrode region) to 18 GHz for a 0-|xm sample. The operating frequency in the TT mode is approximately:


Fv = the operating frequency in hertz (Hz)

Ktom = the domain velocity in centimeters per second (cm/s)

Leif " the effective length in centimeters (cm)

Operation in the TT mode provides efficiencies of 10% or less, with 4 to 6% being most common. Output powers are usually less than 1000 mW, although 2000 mW has been achieved.

Limited space charge (LSA) mode The LSA mode depends on shock exciting a high-4) resonant tank circuit or tuned cavity with current pulses from the Gunn diode. The LSA mode and its submodes are also called accumulation mode, delayed domain mode, and quenched domain mode. These various names reflect variations on the LSA theme. For the LSA mode, the N0 L product must be 1012/cm2 or higher and the NJF quotient must be between 2 x 105 and 2 x 104 s/cm3.

Figure 22-11A shows a simplified circuit for a LSA-mode Gunn oscillator, and Fig. 22-1 IB shows the waveforms. The circuit consists of the Gunn diode shwited by either a LC tank circuit (as shown) or a tuned cavity that behaves like a tank circuit. The criterion for resonant oscillation in a negative resistance circuit is sim-

Rf cjtafe

Rf cjtafe

Gunn Diode Bias Circuit Quenched Mode Gunn Effect

22-11 (A) Oscillator circuit for negative resistance mode, (B) characteristic waveforms of circuit, and (C) outpi it voitage.

ple: the negative conductance (-G * V~R) must be greater than or equal to the conductance represented by circuit losses:

At turn-on, transit time current puises (Fig. 22-12) hit the resonant circuit and shock excite it into oscillations at the resonant frequency, These oscillations set up


22-12 LSA output pulses.

a sine-wave voltage across the diode (Fig. 22-12) that acids to the bias [xjtential. The total voltage across the diode is the algebraic sum of the dc bias voltage and the RK sine-wave voltage. The dc bias is set so that ihe negative swiug of the sine wave forces the total voltage below the critical threshold potential, Vtil. Under this condition, the domain does not have time to build up and the spac e charge dissipates. The domain quenches during the period when the algebraic sum of the dc bias and the sine-wave voltage is below tV The LSA oscillation period (T = i/F) is set by the external tank circuit and the diode adapts to it The period of oscillation is found from:

T = the period in seconds (s) L the inductance in hriirj-s (H) C = the capacitance in farads (F) R = the low-field resistance in ohms (ft) V,u = the threshold potential Fb = the bias voltage.

The LSA mode is considerably more effirietif than the transit-time mode. The Î^SA mode is capable of 20% efficiency se thai you can produce at least twice as much RF output power from a given level of dc power drawn from the power supply as transit time operation. At duty factors of 0.01, the LSA mode is capable of delivering hundreds of watts of pulsed output power. The output power of any oscillat or

where or amplifier is the product of three factors: dc input voltage (V), dc input current (/), and the conversion efficiency (n, a decimal fraction between 0 and U:


= the output power n = the conversion efficiency factor (IM) V = the applied dc voltage I = the applied dc current

For the Gunn diode case, a slightly modified version used:

where n = the conversion efficiency factor (0-1) v the average drift velocity Kh * the threshold potential (kV/ein} M » the multiple VlitJVtk L = the length in centimeters Tit, * the donor concentration e = the electric charge (1.6 X 10™"' coulomb) A * the area of the device in cm-

Bias circuit oscillation (BCO) Gunn mode This mode is quasiparasitic in nature and occurs only during one of the normal (iunii oscillating modes {TT or LKA). if product F\ is very, very small then BCO oscillations cau occur at frequencies from 001 to 100 MHz.

Gunn diode applications

Gunn diodes are often used lo generate microwave RF power in divers^ applications ranging from receiver local oscillators, to polico speed radars, to microwave communications links. Figure 22-13 shows two possible methods for connecting a Gunn diode in a resonant cavily. Figure L!2-):.IA shows a cavity that uses a loop-coupled output circuit. The output impedance ol't his circuit is a Tunc t ion of loop size and position, with the latt er factor dominat ing The lonp positioning is a trade-off between maximum output power and oscillator frequency stability.

Figure 22-I3B shows a Gunn diode mourned in a section o) (lajyged waveguide. RF signal passes through an iris to be propagated through the waveguide lo the load, hi both Figs. 22-13 A ami 22-ViE, the exact ivsfiimm frequency of the cavity is set by a tuning screw insert ed into t he cavity spat o.

IMPATT diodes

The avalanche phenomenon is weli-fcnmvn in PN junction diodes. If a reverse bias potential exceeds a en tain fritical thv< slioid. then I he diode breaks down a fid the reverse current increases abruptly from low 'leakage" values to a very high value. The principal cause of this phenomena is secondary electron emission (i.e., charge carriers become so energetic as to be able to knock additional valence elec-

of this expression is (22-8)

Shorted Coaxial Cavity

Rf Output



Rf Output



Shorted Coaxial Cavity
Wiwguide Flange or Horn Aducfuu

22-13 (A) Gimn diode operated in a cavity and (B) Gunn diode in a cavity antenna trons out of the crystal lattice to form excess hole-electron pairs). The common zener diode works on this principie.

The onset of an avalanche current in a PN jmiction is not instantaneous, but rather, there is a short phase-delay period between the application of a sufficient breakdown potential and the creation of the avalanche current. In 1959, W. T. Read of Bell Telephone Laboratories postulated that this phase delay couid create a negative resistance. It took until 1965 for others (Lee and Johnson) at Bell Labs to create the first of these "Read diodes," Johnson generated about 80 mW at 12 GHz in a silicon PN junction diode. Today, the class of diodes of which the Read device is a member are referred to collective^ as Impact Avalanche Transit. Time (1MPATT) diodes. The name IMPATT reflects the two different mechanisms at work; Avalanche (impact ionization) and Transit time (drift).

Figure 22-14 shows a typical 1MPATT device based on a N+-P-l-P+ structure. Other structures are also known, but the NPIP of Fig- 22-14A is representative. The "+" indicates a higher than normal doping concentration, as indicated by the profile in Fig. 22-14C, The doping profile ensures an electric field distribution (Fig. 22-14B) that is higher in the P region in order to confine avalanching to a small zone. The I region is an intrinsic semiconductor that is lightly doped to have a low charge carrier density. Thus, the I region is a near insulator, except when charge carriers are injected into it from other regions.

The IMPATT diode is typically connected in a high-Q resonant circuit (LC tank or cavity). Because avalanching is a very noisy process, noise at turn-on "rings** the tuned circuit and creates a sine-wave oscillation at its natural resonant frequency (Fig, 22-14D). The total voltage across the NP1P structure is the algebraic sum of the dc bias and tuned-circuit sine wave.

The avalanche current (/(,) is iryected into the I region and begins propagating over its length. Notice in Fig. 22-14E that the injected current builds up exponentially until the sine wave crosses zero then drops exponentially until the sine wave

Impatt Diode
22-14 IMPATT diode characteristic waveforms.

reaches thp negative peak. This current pulse is thus delayed with respect to the applied voltage.

Compare now the external circuit current pulse in Fig. 22-14G with the tuned circuit sine wave (Fig. 22-14l>). Notice that transit time tn the device has added additional phase delay, so the current is 180" out of phase with the applied voltage, Tlvis phase delay is the cause of the negative resistance characteristic. Oscillation is sustained in the external resonant circuit by successive current, pulses reringing it. The resonant frequency of the external resonant circuit should be:


F ^ the frequency in hertz (Hz)

VA = the drift velocity in centimeters per second (cm/s)

L = the length of the active region in centimeters (cm)

IMPATT diodes typically operate in the 3- to 6-GHz region, although 100-GHz operation has been achieved. These devices typically operate at potentials in the 75-to lfiO-Vdc range. Because an avalanche process is used, the RF output signal of the IMPATT diode is very noisy. Efficiencies of single drift-region devices (such as Fig. 22-14A) are about 6 to 15%. By using double-drift construction (Fig. 22-15), efficiencies can be improved to the 20 to 30% range. The double-drift device uses electron conduction in one region and hole conduction in the other. This operation is possible because the two forms of charge carrier drift approximately in phase wit h each other.

Electric Field Distribcjuoa


J ]

Hole Drift Region

Avalanche Region

Eettroo Drift Region


22-15 PPNN double-drift IMPATT diode.

22-15 PPNN double-drift IMPATT diode.

TRAP ATT diodes

Gurrn and IMPATT diodes operate at frequencies of 3 GHz or above. Operation at lower frequencies (e.g., 500 MHz to 3 GHz) was left to transistor, which also limited available RF output power to a great extent, Gunn and IMPATT devices cannot operate at lower frequencies because it was difficult to increase transit time in those devices. You might assume that it is only necessary to lengthen the active region of the device to increase transit time. But certain problems arose in iong structures, domains were found to collapse, and sufficient fields were hard to maintain.

A solution to the transit time problem is found in a modified IMPATT structure that uses P+-N-N" regions (Fig. 22-16). The P+ region is typically 3 to 8 (iM, and the N-region is typically 3 to 13 jinn, The diameter of the device might be 50 to 750 depending on the power level required. The first of these diodes was produced by RCA in 1967. It produced more than 400 W at 1000 MHz, at an efficiency of 25%. Since then, frequencies as low as 500 MHz, and powers lo 600 W, have




Diode Structure

Trapatt Diode Waveforms


TRAPATT diode.


TRAPATT diode.

been achieved. Today, efficiencies in the 60 to 75% range are common. One device was able to provide continuous tuning over a range of 500 to 1500 MHz

The name of the P+-N-N- device reflects its operating principle: trapped plasma avalanche triggered transit (TRAPATT)—the method by which transit time is increased by formation of a plasma in the active region. A plasma is a region of a large number of disassociated holes and electrons that do not easily recombine. If the electric field is low, then the plasma is trapped. That is, the charge carriers are swept out of the N region slowly (see Fig, 22- 16B).

The output is a harmonic-rich, sharp rise-time current pulse (Fig. 22-16C), To become self-oscillatory, this pulse must be applied to a low-pass filter at the input of the transmission line or waveguide that is connected to the TRAPATT, Harmonics are not passed by the filter, so they are reflected back to the TRAPATT diode to trigger the next current pulse.

BARITT diodes

The BARITT diode (Fig. 22-17) consists of three regions of semiconductor material forming a pair of abrupt PN junctions, one each P+-N and N-P+. The name of this device comes from a description of its operation: barrier injection transit time. The BARITT structure is designed so that the electric field applied across the end electrodes causes a condition at or near punch-through. That is, the depletion zone is formed throughout the entire N region of the device. Current is formed by sweeping holes into the N region of the device. Under ordinary circumstances, the P+-N junction is forward-biased and the N-P+ is reverse-biased. The depleted N region forms a potential barrier into which holes are injected into the N region from the forward-biased junction. Utese charge carriers then drift across the N region at the saturation velocity of 10T cm/sec, forming a current pulse. There are three conditions for proper BARITT operation.

The electrical field across the device must be great enough to force charge carriers in motion to drift at the saturation velocity (107 cm/s); The electrical field must

Metal Electrodes

22*17 BARITT diode, be great enough to create the punch-through condition, and the electrical field must not be great enough to cause avalancJung to occur

The normal circuit configuration for BARITT devices is in a resonant IX) tank or cavity. At turn-on, noise pulses will initially shock-excite the resonant circuit into self-oscillation, and successive current pulses supply the energy to continue the oscillations. Figure 22-18 shows the relationship between the oscillatory sine-wave voltage, the internal current, and the external current. As in the other diodes, the output energy comes from tapping the energy in the resonant circuit-

Resonance Circuit Design

Was this article helpful?

0 0
The Ultimate Computer Repair Guide

The Ultimate Computer Repair Guide

Read how to maintain and repair any desktop and laptop computer. This Ebook has articles with photos and videos that show detailed step by step pc repair and maintenance procedures. There are many links to online videos that explain how you can build, maintain, speed up, clean, and repair your computer yourself. Put the money that you were going to pay the PC Tech in your own pocket.

Get My Free Ebook


Post a comment