[ad_1]
The relentless quest for better efficiency in energy conversion systems has accelerated the adoption of new materials such as Silicon Carbide (SiC) and Gallium Nitride (GaN) that address a multi-billion market consisting of environmental-focused applications such as electric vehicles (EVs), charging systems, renewables, etc. One of the main differentiating features of both technologies is the bandgap (or energy gap), expressed in eV, which is 3.2 eV and 3.4 eV for SiC and GaN, respectively, three times larger than that of mainstream Silicon. Among materials with bandgap well above 5 eV, namely UWBG, we find Diamond, Gallium Oxide, Aluminum Nitride and cubic Boron Nitride, see Table 1.
Why is the energy gap so important
It is well known that the energy gap represents the amount of energy that an electron of the valence band must acquire to jump to the conduction band where it is free to move under an electric field and make itself available for generating a current flow so devices such as diodes and transistors can be fabricated. A high-bandgap transistor can withstand higher electric fields because atomic bonds are strong. This feature results in reduced on-state resistance at high voltages vs. Silicon, which, in turn, minimizes conduction losses and helps improve efficiency. Such performance is related to the critical electric field parameter, see Table 1, which reaches the highest values for AlN and c-BN.
Recently, researchers have focused on gallium oxide, diamond, and AlN. All of them exhibit attractive attributes, but also unavoidable weaknesses that have so far impeded their commercial development. However, AlN stands out as a potential contender for other materials, thanks to recent technological advances at Nagoya University reported at the most recent IEEE International Electron Devices (IEDM) Event, held last December in San Francisco.
Where AlN is already used today
Aluminum nitride (AIN) is a non-toxic material already used for its high thermal conductivity and remarkable electrical insulation properties. Apart from its thermal expansion coefficient and electrical insulation capabilities, AlN ceramics are resistant to attacks by most molten metals, such as copper, lithium, and aluminum. AlN is a ceramic material composed of 65.81% Al and 34.19 N. Because of its properties, this ceramic has proven useful in many applications, such as optoelectronics operating at deep ultraviolet frequencies. Aluminum nitride is also used extensively in applications such as heatsinks and heat spreaders, electrical insulators, Si wafer handling, and processing, as package substrates (in place of highly toxic beryllium oxide and Al2O3 called Alumina), as dielectric layers in optical storage media, microwave packaging, etc.
AlN as a semiconductor material
All semiconductors are based on chemical doping of impurity elements to operate. When a doping material is inserted, n-type or p-type semiconductors can be created depending on whether such step produces an excess of negative charge carriers, electrons, or positive charges derived from a deficit of electrons, called holes. Nearly all successful devices available in the market are made up of such doped semiconductors, sandwiched together. The primordial semiconductor structure is a p-n junction with two terminals attached or a diode.
There are some compound semiconductors containing elements from group III and group V of the periodic table— for example, gallium nitride— that have an unusual but conveniently exploitable property. At the interface where two specific semiconductors meet, such as GaN and AlGaN, they can spontaneously generate a two-dimensional electron gas (2DEG) of extremely mobile charge carriers, even without chemical doping. Nitrogen has a higher electronegativity than Gallium and Aluminum causing a net charge displacement or an electric spontaneous polarization, that is, distinct regions of opposite charge. Furthermore, mechanical stress due to lattice mismatch causes additional polarization by the piezoelectric effect. In other words, such an effect generates charges by just straining the lattice, an alternative form of doping called polarization doping. The two types of polarizations concur to create a net positive charge. But for charge neutrality, the same amount of negative charge pops up at the interface which is exactly the high-conductivity 2DEG.
AlN junction and polarization-induced (Pi) doping
The paper mentioned above was written by a team of seven co-authors, some of them from the Nagoya University including Hiroshi Amano, who won the Nobel Prize in 2014 for inventing the blue LED. This paper describes the realization of a diode by implementing the technique of dopant-free distributed polarization doping in aluminum nitride or, more precisely, an alloy of aluminum gallium nitride (AlGaN) consisting of a mixture of AlN and GaN. The underlying doping technique is the unique polarization-induced (Pi) doping scheme, which gives rise to a high-mobility 2DEG without impurity dopants. Recently, a two-dimensional hole gas (2DHG) has also been reported in undoped GaN/AlN structures. In addition to the generation of two-dimensional carriers from a polarization discontinuity across a heterojunction interface, Pi-bulk or distributed polar doping (DPD) for three-dimensional electron gas and hole gas with a constant bulk concentration can also be obtained from a constant polarization gradient in linearly graded structures.
Like any other diode, the device has a p-doped region and an n-doped one, or a junction. For both regions, the doping was implemented with distributed polarization doping techniques. They achieved the different n-type and p-type polarizations by creating a gradient, in each of the doped regions, in the percentage of AlN versus GaN in the alloy. The big innovation stands in the fact that the doping is n-type or p-type depending simply on the direction of the gradient. The authors proved that the diode based on alloys of aluminum nitride is capable of withstanding an electric field of 7.3 megavolts per centimeter, about twice as high as what is possible with SiC or GaN. This value is impressive, but it is still far from a theoretical value of around 15 MV/cm as shown in Table 1.
Simplified steps of diode fabrication
After forming an undoped AlN layer and a high concentration n+-type Al0.7Ga0.3N layer on a high-quality AlN (0001) substrate using metal organic vapor phase epitaxial growth (MOVPE), the mole fraction (MF) of AlN was gradually increased from 70% to 95% in a layer 400 nm thick to form the n-type DPD region. Then, the MF was linearly decreased in two steps from 95% to 70% and 30% to form the p+-type DPD region. Finally, a high concentration p++-type GaN layer was achieved with Magnesium doping. Electrodes were formed on the top high-concentration p++-type GaN layer and the bottom high-concentration n+-type Al0.7Ga0.3N layer to fabricate a p-n junction diode.
The next step is fabricating a diode that has a layer of 100% AlN at the junction, rather than 95 percent. According to some calculations, a layer of AlN just 2 micrometers thick would be sufficient to block a voltage of 3 kV. Thermal conductivity could also be significantly improved with a higher grade of AlN. The ability to conduct heat is vital in power electronics applications, and the thermal conductivity of the AlGaN alloy is mediocre, below 50 W/mK (watts per meter-Kelvin). Pure AlN, as one can see in Table 1, exhibits 319 W/mK, not far from that of 4H-SiC.
Future developments
Having proven that an AlN vertical diode is feasible with the polarization doping processes, the next step is the implementation of a vertical transistor to compete with SiC MOSFETs or GaN HEMTs. According to IEEE Member Takeru Kumabe, a coauthor of the Nagoya paper, “ AlN-based vertical heterojunction bipolar transistors, which consist of two p-n junctions and exhibit good power and area efficiencies, are our targeted device, our dream, to be realized.” 1 Kumabe added that to make the dream come true, a better understanding is needed in terms of charge mobility, carrier lifetime, critical electric field, and intrinsic defects.
References
1 The New, New Transistor – IEEE Spectrum
[ad_2]
Source link
