11 Impatt Diode

Dr. Ayushi Paliwal and Dr. Monika Tomar

  1. Introduction

 

IMPATT (Impact Avalanche Transit Time) devices are solid state oscillators or amplifiers for microwave and millimeter wave frequencies up to above 200 GHz, generally with high power output (for a solid state device). Efficiencies as high as 25% have recently been obtained with GaAII devices, but the traditional IMPATT is a silicon device with up to 15% efficiency. The basic principle involves a 1800 phase-delay of the current with respect to the voltage, which clearly will make the device exhibit a negative resistance. Shockley (1954) had introduced this general idea, not yet involving impact avalanching to produce the phase-shift. A number of different p-n-junction type structures are used today. During operation, the diode is undergoing (controlled) reverse break-down. The history of the development of the IMP ATT device has some similarities to that of the Gunn-device. Both were being developed in parallel in the early 1960’s, for example one Bell Laboratories group worked simultaneously on both devices (DeLoach, 1976). In the case of the IMPATT, there was also a theoretical prediction of one version of the IMP ATT operation, published in 1958 by W.T. Read (Read, 1958). The event described above, resulted in the first publication regarding an IMPATT type oscillator (Johnston, DeLoach and Cohen, 1965). Only a couple of months later, Read-type IMPATT operation was also observed (Lee et al., 1965).

  1. OPERATION OF IMPATT DEVICES-PHYSICAL DISCUSSION 2.1 The Impact Ionization Process

Reverse break-down is most conveniently first discussed for a p+-n diode, usually implemented with a p+-n-n+ structure, see Figure 1.

The corresponding energy band diagram for large reverse bias is given in Figure 2. The reverse current consists primarily of holes which diffuse up to the potential barrier from the n-side, and then quickly traverse the junction, accelerated by the very strong electric field there. A smaller number of electrons cross the junction in the opposite direction, smaller because the p-side is more highly doped, and because it is the minority carriers which constitute the reverse direction current. The holes will become energetic enough to be able to ionize an electron pair, somewhere in the depletion layer. This process requires about 1.5 times the bandgap energy, i.e. about 1.6 eV for silicon.

The impact ionization will be able to sustain an avalanche if the probability that a carrier will cause an ionization event on its transit of the depletion region is equal to 1.0 or higher, and under these circumstances, the process will start to grow rapidly (in time) in the manner of an avalanche. The avalanche condition can be expressed:

Here, a is the probability of impact ionization occurring in unit length of travel, or the “ionization rate”. The units are usually cm -1. To be more realistic, we must recognize that the ionization rates for electrons and holes are not the same. Also, the ionization rates depend very strongly on the electric field. A typical dependence is:

A generalized avalanche condition can be derived for the case of unequal rates for electrons and holes

 

This value for < a > can be used with very accurate results for GaAs.

 

Due to the strong dependence of a on electric field, we find that the avalanching occurs almost exclusively very close to the peak of the electric field (in Figure l). This region of the diode is termed the “avalanche region”. In a device with field distribution as in Figure l, the holes generated will drift to the left into the contact and will be out of the picture as far as device operation is concerned, whereas the electrons will drift to the right through the so called “drift region” of the device, see Figure 3.

 

 

2.2 Reverse Break-Down Voltage

 

It is convenient to observe that the maximum field of the triangular field distribution in a p-n-junction diode at breakdown is fairly independent of the doping and the type of junction. The explanation for this fact is again the very high E-field dependence of the ionization-coefficient, Q. Thus, when the field has reached a certain magnitude, then it becomes very easy to satisfy the avalanche condition when the integral is taken over typical junction widths, and the field for which this happens is basically a universal constant for the material. The reverse break-down voltage can now be derived in terms of the maximum E-field at break-down, Em. The results are for the standard cases of a one-sided (p+ – n) diode, and two-sided diode, respectively,

 

Note that the break-down voltage is inversely proportional to the doping in both cases. Typical break-down voltages in the 10 to 100 volt range are obtained if the doping is of the order of 1016 to 1017cm-3 (on the low side in one-sided junctions). These widths are 10-20% of the width of the entire depletion region, justifying the model, which assumes that the drift region occupies most of the depletion region.

 

2.3 Saturated Drift Velocity

 

The saturation velocity decreases as the temperature increases. It is important that the velocity is almost independent of the electric field for the high fields which exist in IMPATT devices – this means that we can regard the carrier velocity as essentially constant and equal to the saturation velocity anywhere in the drift-region. The saturation velocity of electrons and holes in GaAs is somewhat lower. The saturation velocity is essentially the same for electrons and holes. Its temperature-dependence is given by:

 

 

2.4 Displacement Current

 

Due to the large amounts of rapidly moving electric charge in an IMPATT device, it is necessary to take the displacement current into account. The avalanche gives rise to the “injected” current which is localised to the avalanche region. Due to the nature of the avalanche process, the injected current is concentrated toward the end of the first half period of the AC-voltage across the device. The time dependence of the injected charge is therefore similar to a short pulse. As this pulse of charge drifts through the drift region, we can model it with a constant charge which moves through a capacitor. In general, the total current is continuous, and is represented by a displacement current in any plane of the capacitor which does not contain the charge, as shown in figure 4. If the charge is moving with constant velocity, then the total current is also constant in time until the charge has traversed the capacitor. We can calculate the average current when the charge, q, moves across a capacitor of width, W, in a time at, as

 

  1. Diagram of p+-n Diode IMPATT Device Operation

 

We are now ready to put the above facts together in a diagram (Figure 5). We assume that the E-field consists of a DC component, adjusted so that reverse break-down just occurs, and an AC component, which is added uniformly across the device (compare the Gunn-device case, where the electric field could not be uniform).

 

The structure of the device is shown in Figure 6. The E-field and voltage are positive when they accelerate carriers in the positive z-direction, and the current is positive in the same direction as the electric field (Figure 6).

Referring to Figure 5, the AC electric field is shown in the lowest set of curves, as a function of time. In the top of the diagram, we see in the first “frame” the DC electric field distribution. As the AC field component grows positive, the avalanche is initiated. The growth of the charge produced by the avalanche is pictured in the second set of frames from the top. In the second frame (t = T/4), the AC field is at its maximum, but the maximum avalanche charge injection does not occur until the third frame (t = T/2), just before the AC voltage goes negative and drives the total voltage below the breakdown voltage. In other words, the avalanche continues to grow as long as the electric field is larger than Em. As the AC voltage becomes more negative, the avalanche injected charge drifts through the drift region, and the device current, shown in the next set of frames, is constant with time until the charge reaches the right contact. If we compare the two lowest sets of frames, we notice that the current has an approximately square wave-form, being off for the first half period, and constant and positive during the second half period, which represents the negative resistance which we sought.

 

The negative resistance arises as a result of three effects:

 

1) The delay of the maximum of the avalanche until it just turns off, i.e. by T/2 with respect to t =0.

 

2) The transit time delay of T /2 for the drifting charges in the drift region.

 

3) The effect of the displacement current, which produces a square current pulse from a short pulse of injected charge. This factor effectively speeds up the onset of the current in the contact by T /2.

 

Taking into account all of the three above effects, we find a total delay of T/2 of the current with respect to the voltage. Specifically, we see that the correct value of the transit time delay in the drift-region should be about T/2. We can also calculate the estimated frequency of operation as follows (note that there is a factor of two ratio of the transit times, in periods, for the IMPATT device compared with a transit-time mode Gunn device.):

 

As an example, a 10 GHz IMPATT device using silicon should have a drift region length of about 5 micron. We also find it useful to introduce the concept oftransit angle, defined

  1. DOPING PROFILES FOR IMPATT DIODES

A great variety of doping profiles have been used for IMPATT diodes and a few are illustrated in Figures 6 to 8. The recent trend has been toward structures with very narrow (higher efficiency) avalanche regions, such as the profile originally proposed by Read (1958). The “lo-hi-Io” structure has a uniform field in the avalanche region, while the narrow clump of (donor) charge allows the field to decrease to values well below Em immediately as the carriers enter the drift region, preventing any avalanche there. Another popular structure is the double-drift one, which allows both electrons and holes to drift through separate drift regions.

 

Figure 6: (a) Schematic of IMPATT diode, (b) Distribution of doping, (c) electric field and (d) ionization probability in a Read type IMPATT diode.

  1. Summary
  • Operation of IMPATT devices: impact ionization process, reverse breakdown voltage, saturation drift velocity and displacement current
  • p+n diode IMPATT device operation
  • Doping profiles for IMPATT diode