10 Tunnel Diodes

Dr. Ayushi Paliwal and Dr. Monika Tomar

  1. Introduction

 

Quantum tunneling processes are discussed in most elementary treatments of quantum mechanics, and represent an excellent example of how the quantum mechanical nature of “particles” (such as electrons) leads to surprising consequences, in direct violation of classical physics concepts. The quantum characteristics are also precisely what makes tunneling devices work. The basic idea follows from the wave-nature of the electron; the electron wave-function can exist even in areas into which the electron cannot enter according to classical physics. A typical example is an electron incident on a potential barrier which represents a potential energy greater than the kinetic energy of the electron – the electron wavefunction will become an evanescent wave inside the barrier, in analogy with the behavior of an electromagnetic wave incident on a waveguide below cut-off. If the barrier is not too wide (of the order of 5-10 nm or less), the wavefunction will still have a measureable amplitude at the opposite edge of the barrier, and the electron then has a finite probability of being transmitted. In 1958, Leo Esaki, a Japanese scientist, discovered that if a semiconductor junction diode is heavily doped with impurities, it will have a region of negative resistance. The normal junction diode uses semiconductor materials that are lightly doped with one impurity atom for ten-million semiconductor atoms. This low doping level results in a relatively wide depletion region. Conduction occurs in the normal junction diode only if the voltage applied to it is large enough to overcome the potential barrier of the junction.

  1. Description about tunnel diode

In the tunnel diode, the semiconductor materials used in forming a junction are doped to the extent of one-thousand impurity atoms for ten-million semiconductor atoms. This heavy doping produces an extremely narrow depletion. Also because of the heavy doping, a tunnel diode exhibits an unusual current-voltage characteristic curve as compared with that of an ordinary junction diode as shown in Figure 1.

 

At zero bias, there are no states available for tunneling, while with increasing positive bias, more and more such states exist. We thus expect the current to rise with voltage. The current will have a maximum when the band edge on one side coincides with the Fermi level on the other, since for voltages beyond this point, the electrons will have fewer states to tunnel to. Ideally, the tunneling current should therefore go to zero, while in practise tunneling is still possible through a small number of states in the forbidden gap. At the normal cut-in voltage of the p-n-junction diode, thermionic excitation “over the top” of the potential barrier will occur, and the current will again rise, as in the normal 1-V-characteristic of a diode. The end result will be a characteristic as shown in Figure 1, with a region where the differential conductance is negative. The “valley” current, Iv, is determined by the excess current due to the gap states. It is in general advantageous to achieve a large ratio Ip / Iv.

 

Operation of tunnel diode depends on the tunneling effect. In an open circuit i.e. when no bias is applied, Fermi level EF in the p-side is at the same energy as the Fermi level EF in the n-side as shown in Figure 2 (a). It can be seen from energy band diagram of tunnel diode in zero bias mode, that there are no filled states on one side of the junction which are at the same energy as empty allowed states on the other side.

 

Thus there is no flow of charge in either direction across the junction and current is zero.

 

Figure 2: Energy band diagram of a heavily doped p-n diode (a) under open circuit condition and (b) with an applied reverse bias

 

Reverse Bias Mode: When reverse bias is applied, the height of the barrier is increased above the open circuit value EO. Hence, n-side levels must shift downward with respect to the p-side level as shown in Figure 2 (b). Now some energy states in the valence band of the p-side lies at the same level as allowed empty states in the conduction band of the n-side, thus the electrons will tunnel from p to n side leading to reverse diode current. If reverse bias is increased further then tunneling and reverse current increases as shown in Figure 1. When Tunnel diodes are used in the reverse direction they are called back diodes and can act as fast rectifiers with zero offset voltage and extreme linearity for power signals.

 

 

Forward Bias Mode: When forward bias is applied, the height of the barrier is decreased below the open circuit value EO. Hence, n-side levels must shift upward with respect to the p-side level as shown in Figure 3(a). It can be seen that there are occupied states in the conduction band of the n-side which are at the same energy level as allowed empty states in the valence band of the p-side. Thus electrons tunnel from the n-side to p-side leading to forward current. As the forward bias is further increased to VP (shown in Figure 1), the maximum number of electrons tunnel from occupied states of n-side to empty states of the p-side, leading to maximum peak current value IP (Figure 3(b)). But if the forward bias still increases (> VP), the tunneling current decreases (shown in Figure 1). In this condition the occupied states of n-sides are shifting towards upwards with respect to empty states of p-sides which reduce the tunneling of electron resulting in negative resistance condition (Figure 3(c)). When operated in the negative region used as oscillator. When forward bias further increases to VV (shown in Figure 1), there are no empty allowed states are present on one side of the junction at the same energy as occupied states on the other side, thus tunneling current drops to zero (Figure 3(d)). On increasing the forward bias voltage (> VV) the tunnel diode starts conducting as normal p-n diode giving rise to forward current again (shown in Figure 1).

 

Tunnel diodes of the type described above can be fabricated from any of a number of semiconductor materials. They continued to be used for many years, but at the present time have been superceded by other devices in almost all applications. One primary use was as low-noise amplifiers, where the simplicity and low noise temperature of the device made it attractive. A disadvantage was that dopants in some diodes would slowly diffuse through the device, activated by the high electric field this made the reliability less than desired. The mechanical structure of the device also made it quite delicate. Tunnel diodes thus might have been relegated to the category of obsolete devices, were it not for new advances in fabrication techniques, which have recently made a new version of the tunnel diode feasible with considerably improved characteristics: the “resonant tunneling device”, or RTD.

  1. Resonant Tunneling Devices

The RTD was proposed by Tsu and Esaki (1973) but its realization as a microwave device had to await the advanced state of MBE and OMCVD material growth technology of the 1980’s. Chang, Esaki and Tsu measured a small negative differential conductance (Chang et al., 1974). A few years later, Sollner et al. (1983) demonstrated for the first time that this device can have a negative differential conductance which is large enough for practical device applications. The early work of Sollner’s group also made it clear, as was expected, that the intrinsic speed of the tunneling process was extremely fast – of the order of 0.1 picoseconds. Resonant tunnel diodes have been investigated intensely since 1983 as a result of the potential for development of very high frequency devices based on the RTD structure. Such devices are at the present time superior to the conventional tunnel diode.

 

3.1 Fabrication of RTDs

 

The barriers through which electrons tunnel in the RTD are formed by growing alternating, very thin, layers of semiconductors with different bandgaps, rather than relying on thin ~n junction depletion layers, as in the original tunnel diode. A typical structure is shown in Figure 4.

It consists of two layers of AlAs, (bandgap 2.16 eV) of thickness 1.5 nm, on either side of a 5.0 nm thick layer of GaAs, (bandgap 1.43 eV). Thicker layers of n-type GaAs, on either side of the above layers are intended to provide a low resistance path from the ohmic contact (at the top) and the highly doped substrate to the tunneling structure. Electrons in the GaAs layer will find themselves surrounded by two fairly high potential barriers (Figure 5 (a)), and will form quasi-stationary states, according to elementary quantum mechanics. The number of states, and their location, will depend on the thickness and height of the barriers, as well as the width of the “well”. Tunneling through the barriers will be enhanced when electrons outside the barriers have an energy equal to one of the energy levels in the well – theoretically a transmission probability of 1 is predicted (Figure 5(b)).

 

If the voltage is increased further, the energy of the electrons at the cathode no longer matches that of those inside the well, and the tunneling stops (Figure 5(c)) and is shown in the I-V-characteristic of Figure 5 (d) results. There is a useful analogy to the transmission of electron waves through the RTD structure with the transmission of electromagnetic waves through a Fabry-Perot resonator (used in optics or quasi-optics), or a microwave resonator such as a waveguide transmission cavity with coupling holes at either end, as indicated in the insert to Figure 5. In the microwave case, 100% transmission also occurs at resonance if the cavity is lossless. It is assumed in the RTD case (as for the electromagnetic case) that the electron wave-function is coherent through the tunneling process, i.e. the scattering probability inside the well must not be too high for the device to work.

device a. Equivalent circuit model

 

The first equivalent circuit model used for the RTD was the one which has been applied to the pn junction tunnel diode as shown in Figure 6.

A load with an inductive component has been added, and oscillations will occur at a frequency for which the total admittance of the circuit is zero. The oscillation frequency can easily be derived to be:

The highest frequency (and of course zero power out) is obtained for the extreme case of RL = 0 and G = 1/2Rs. This frequency is:

 

The above expression for fmax is the one usually given in the tunnel diode literature. The recent work on RTDs has demonstrated that the RTD can not be described by the simple equivalent circuit in Figure 6, however. Most importantly, a depletion region is formed “down-stream” from the double-barrier, and a substantial part of the terminal voltage may be dropped across this region, as shown in Figure 7.

 

 

A thin accumulation layer is also located on the opposite side of the double-barrier, the “cathode”. The width of the depletion layer can be varied by changing the doping and width of the n-type layer on the anode side.

  1. Summary
  • IV characteristics of tunnel diode and its energy band diagram under different biasing configuration
  • Resonant tunneling devices: fabrication and equivalent circuit model