Extreme-temperature devices using AlN

Diodes and transistors with AlN channels deliver high breakdown voltages and operation at incredibly high temperatures.

BY HIRONORI OKUMURA FROM UNIVERSITY OF TSUKUBA

A NUMBER OF human activities are expanding into extreme environments, often motivated by resource exploitation. This has taken exploration in various directions, including deep underground, to great depths at sea and into deep space. In all these environments the temperature is extreme – it exceeds 300 °C on the surface of Venus, in deep-well drilling, and in the space inside an operating engine.

To find out more about all these environments demands the deployment of sensors. But the most obvious ones – that is, those based on silicon – are not up to the task, due to a relatively low operating temperature limit. This means that in order to enrich our lives from these environments, we need to develop extreme-temperature electronics.

Figure 1. The leakage current path and the thermal degradation points in a MESFET with a gate oxide.

When all forms of semiconductor device are operated at extreme temperatures, they face issues associated with materials, electrodes, gate oxides and packaging (see Figure 1). As the temperature increases, numerous electron-hole pairs are generated, due to excitation of electrons from the valence band maximum to the conduction band minimum. These electrons, which increase the intrinsic-carrier concentration (see Figure 2 (a)), are detrimental, as they increase the leakage current of the devices and prevent them from turning off. Options for reducing the leakage current include introducing semiconductor materials with a larger bandgap energy and lower intrinsic carrier concentrations (see Figure 2 (b)), or restricting current diffusion from areas other than the channel. Turning to a channel layer surrounded with high-resistivity layers that have low effective donor/acceptor concentrations and low defect concentrations can raise the device’s operating temperature. Another approach is to deploy devices with p-n junctions, such as JFETs and BJTs. In these cases, it’s also important to select refractory metals for the electrodes that have minimal reactivity with base semiconductors. In particular, titanium, vanadium, tantalum, molybdenum, tungsten, and platinum are better for this purpose than aluminium, magnesium, copper, silver, indium, and gold.

Why use AlN?
There are many semiconductor materials with a larger bandgap energy than silicon. They include SiC (3.3 eV), GaN (3.4 eV), Ga₂O₃ (4.7-5.2 eV), diamond (5.5 eV), and AlN (6.1 eV). The team at NASA, led by Philip Neudeck, have reported that SiC JFETs can operate at temperatures over 800 °C. While this is undoubtedly an impressive result, materials with even wider bandgaps promise to reach even higher temperatures. However, quite a few of them have significant drawbacks. GaN suffers from a high effective donor concentration of 10¹⁶ cm^-3; it’s not possible to form p-type Ga₂O₃ layers; and diamond starts to reacts with oxygen at around 700 °C. In stark contrast, AlN has no obvious flaws, and offers thermal stability and controllable doping. Due to these attributes, our team at the University of Tsukuba has been devoting all our attention to AlN for the development of extreme-temperature devices.

Historically, it’s been assumed that AlN is only good as an insulator. However, around 20 years ago Yoshitaka Taniyasu and colleagues at NTT demonstrated that this is not the case by growing electrically conductive AlN layers by MOCVD.

This team recorded an electron mobility of 426 cm² V^-1 s^-1 for silicon-doped AlN layers, for a dopant concentration of 3 x 10¹⁷ cm^-3. Building on this work, they went on to pioneer p-type AlN growth and demonstrate the first AlN LEDs with a wavelength of 210 nm and quasi-vertical AlN p-n diodes. These successes are to thank for the recent, rapid development of deep-UV LEDs based on AlGaN and AlN.

Figure 2. (a) An illustration of electron-hole pair generation at high temperatures. (b) The intrinsic carrier concentration of silicon, SiC, GaN, β-Ga₂O₃ diamond, and AlN as a function of reciprocal temperature.

As well as optical devices, the research community has investigated AlN Schottky barrier diodes and AlN/AlGaN HEMTs, to explore the potential benefits of a high critical electric field. Unfortunately, these devices suffer from a low carrier concentration, due to high ionization energies for the donors and acceptors – it is 0.3 eV for silicon and 0.6 eV for magnesium. Due to this, carrier concentrations for both these dopants are around two orders of magnitude lower than their concentrations, causing devices to have very small currents. To overcome this problem, our team, in working in partnership with researchers at MIT and Aalto University, have broken new ground by introducing polarisation-induced doping in N-polar AlGaN/AlN structures. Thanks to spontaneous and piezoelectric polarisations, this form of doping can increase current and lower contact resistivity. Using polarisation-induced doping, we have demonstrated the first N-polar AlN-based PolFETs and HEMTs with drain currents over 100 mA mm^-1. Such success has led us to view AlN as a practical semiconductor for optical and electrical devices.

To produce these devices, we have been able to draw on a number of material suppliers. High-quality AlN samples on 2-inch sapphire substrates can be purchased from Dowa Electronics Materials, and 2-inch bulk AlN is commercially available from Stanley and Asahi Kasei.