Abstract: Now a days, there is a demand of faster things with high power capabilities.
With the rise in wireless
technologies and 5G, there is a need for the ability to send the signal to a wider range which requires high gain
and high-power transistors. Also, wireless/RF communication systems, transistors need to be able to switch at
very high frequency. On the other hand, all the modern electronic devices from phones to laptops, toy drones to
electric cars all use rechargeable batteries which require frequent charging. People on the other hand are losing
patience and cannot wait for hours until their drone or car battery get recharged fully. And it makes sense to
have the ability to charge batteries faster. Ideally, we would want their electric car to be recharged in the same
duration as one would spend time refueling the gas operated cars. That is too far from reality, but still, we can
dream of something closer. For this fast charging, we need transistors with high power handling capability also.
The current silicon-based transistors are stable and reliable but don’t have the switching speed we would need
to handle very high RF signals. This gap so far has been filled by some compound heterojunction transistors.
There has been some out there like Gallium nitride (GaN) based, Gallium arsenide (GaAs) based, or Indium
phosphide (InP) based hetero junction transistors called High Electron Mobility Transistor (HEMT) which has
very high-power capability due to high density of carriers which are majority carriers which forms a two-
dimensional gas (2DEG) at the junction. The mechanisms that lead to the formation of 2DEG in the heterojunction
of compound semiconductor materials have been explored and explained. This 2DEG also gives this class of
transistor a faster switching capability. Because of these good features, HEMTs are being used in all sorts of
power supplies; car charging stations; satellite and wireless communication systems, radio telescopes etc.
1. Introduction
We know that the group-IV elements in the periodic table namely Carbon/diamond(C), Silicon (Si), Germanium (Ge),
Tin (Sn) are elemental semiconductor materials. Among these, silicon (Si) has been quite popular in the semiconductor
industry for manufacturing discrete devices to integrated circuits because this is one of those semiconductor materials
that has been studied quite extensively. Silicon has several inherent advantages, like more reliable, easier to
use/engineer, low cost, abundance of source of silicon on the earth’s crust and cost [1]. On the other hand, there are
compound semiconductor materials as well which are alloys formed by combining Group III-V and Group II-VI
elements. Some of the compound semiconductor materials, like Gallium arsenide (GaAs), Gallium nitride (GaN),
Indium phosphide (InP) etc., have been studied for quite some time and have been put into some sleek applications.
In this paper, we cover the basic structure, the special characteristic feature and the operation of High Electron
Mobility Transistor (HEMT)- one of the applications of compound semiconductor materials.
A High-Electron-Mobility Transistor (HEMT), also known as a heterostructure FET (HFET) or Modulation-doped
FET (MODFET), is another kind of Field-effect transistor. HEMT is the first transistor to incorporate an interface
between two compound semiconductor materials with different band gaps [1]. Unlike silicon MOSFET which uses
minority carriers in the doped region under the gate as the channel, HEMT uses the 2-dimensional electron gas (2DEG)
that is formed at the interface of two compound semiconductors with different band gaps as the channel. HEMTs have
a very high electron mobility, which is the reason for their exceptional performance in high-frequency applications
[2]. Also, high carrier density, low output resistance, low noise figure or HEMTs secure their place for high power
applications. The formation of 2DEG in HEMT along with the reasons for all these exceptional characteristics of
HEMTs will be discussed in later sections. Before getting too far away, let’s take a quick view on the history of
development of compound semiconductor devices and inventions and use of HEMTs.
2. Brief History of use of Compound Semiconductors and Invention of HEMTs
The compound semiconductors like Gallium nitride (GaN), Aluminium nitride (AlN), Indium nitride (InN) and their
alloys form a unique material system, termed as nitride semiconductors. Compared to other systems, nitrides cover a
much wider spectrum of bandgaps as can be seen in the bandgap-lattice constant plot [2] and this material system has
been researched by several researchers and research groups for over three decades now.
Figure 1: Bandgap-lattice constant plot [2].
First p-n junction GaN based LED was reported by Amano et al in late 1989 in which the high defect densities drove
the device performance and not the intrinsic material properties [3],[4]. The work done by Isamu Akasaki at Nagoya
and Mejo Universities and Shuji Nakamura at Nichia Chemical company in Japan during mid-1980s did go a long
way in mitigating such performance issues [2]. Using the nitride semiconductor (GaN based) first blue light LED was
invented in early 1990 by Isamu Akasaki, Hiroshi Amano and Shuji Nakamura who were awarded Nobel prize in
2014 in physics. This was the first blue LED with lowest forward voltage of 4V with 20mA [3]. In the same decade,
nitride semiconductors found their use in electronic devices as well in addition to optoelectronics.
On the transistor side, first HEMT patent was submitted on August 16, 1979, for Fujitsu Laboratories Ltd., Japan by
Takashi Mimura [5]. In this, Schottky gate placed on the n-type AlGaAs was used to control the electrons at the
interface of the single heterojunction consisting of undoped GaAs on the other side of the junction. The first
demonstration of depletion mode HEMT (D-HEMT) and enhancement mode HEMT (E-HEMT) were reported in May
1980 and August 1980 respectively [5]. The High-speed switching characteristics of HEMT devices became evident
in 1981 when Fujitsu demonstrated a ring oscillator switching at a delay as low as 17.1 ps [7]. Since the E-HEMT
requires only a single positive voltage, circuitry is simple. This feature gives E-HEMTs an advantage in such
applications as mobile phone power amplifiers and digital integrated circuits [7]. First HEMT prototype for satellite
communication was developed and was introduced it at the 1983 International Solid -State Circuit Conference and
was commercially used as a cryogenic low-noise amplifier at Nobeyama Radio Observatory (NRO), Nagano, Japan
in 1985 [5]. Since then, it has found several applications ranging from amplifiers for satellite communications to radio
telescope and all sorts of applications where high gain, high power density, high switching speed and low noise figure
is required.
Compound semiconductor materials whose properties have already been studied since 1970s. HEMT devices
incorporate heterojunctions formed between two different bandgap materials where electrons are confined in a
quantum well to avoid impurity scattering. The direct bandgap material GaAs have been successfully used in high
frequency operations as well as in optoelectronic integrated circuits owing to their high electron mobility and dielectric
constant [6]. AlGaAs is suitable barrier material for GaAs as its lattice constant is nearly the same and has a higher
bandgap than that of GaAs. GaN/AlGaN is another material combination that is being used for the fabrication of
HEMT as they have matching lattice constant [6].
3. Basic Structure of HEMT
The basic requirement of HEMT engineering is the formation of two-dimensional Electron Gas (2DEG) in a
quantum well which is accomplished by the appropriately choosing the material system of different bandgaps
with similar lattice constants to avoid stresses within the crystal lattice [7]. The structural cross section of HEMT
is shown in fig (2) below. In a HEMT, for 2DEG to form, barrier material should be of wide bandgap material
and the buffer layer should constitute a narrow bandgap material where both of these layers may have the same
doping type, n-type, forming heterojunction [7]. In general, the barrier layer doping is higher than buffer layer
doping. Any HEMT requires these two basic layers. Any other layers below or above that we find in some special
HEMTs are for better engineering control of the device performances.
Figure 2: Basic Cross-sectional view of a HEMT Device [7].
4. Formation of 2-D Electron Gas
From the law of entropy, when the barrier layer is brought in contact with buffer layer, to acquire the minimum
energy configuration, electrons diffuse from the wide bandgap barrier layer to narrow bandgap buffer layer and
this transfer of electrons continues till the diffusion is counterbalanced by the built-in electric field, similar to the
phenomena that occurs in a PN Junction [6]. This is when the Fermi level becomes same across the junction (i.e.
∆Efi =0). Under equilibrium condition, the conduction and valence band bend suitably to form a continuum
through the junction. In the channel region (junction), the bending of especially the conduction band facilitates
the formation a two-dimensional quantum well with a finite barrier heights and widths [7]. As we know, per Pauli
exclusion principle, no two electrons in that quantum well have the same sets of 4-quantum numbers. Thus, these
electrons present it the channel are confined to the quantum well and reside in different quantized energy levels
within the well forming a two-dimensional electron gas (2DEG). There are two mechanisms, Spontaneous
polarization and piezoelectric polarization, which aid in creating the 2DEG without any dopant needed. Let’s take
an example of GaN/AlGaN to understand these phenomena of creation of 2-dimensional electron gas.
GaN- Based HEMTs have similar basic layered structure (a buffer layer and a barrier layer) as GaAs-based
HEMTs, but no intentional doping is needed in AlGaN/GaN HEMTs [6]. Gallium Nitride (GaN) has Wurtzite
(like that of Zinc oxide) crystal structure.
Figure 3: GaN/AlGaN crustal structure showing the spontaneous and Piezoelectric Polatization[9]
Due to higher electronegativity (tendency of pulling electrons towards itself) of Nitrogen compared to that of
Gallium, which creates a dipole, there is spontaneous polarization (PSP). Also, GaN crystal is non-
centrosymmetric (which means the atoms surrounding the N-atom in the crystal are not symmetrically distributed
about N-atoms). This non-centrosymmetric nature of the crystal lattice gives rise to some tensile stress which
gives rise to piezoelectric polarization (PPE) vector. Now, net polarization is the sum of the two polarizations
which results in net polarization charge densities equal to the sum of polarization charge densities of piezoelectric
polarization and spontaneous polarization [9].
σ( PSP + PPE)= σ(PPE) + σ(PSP) [9]
If the polarization induced charge density is positive (+σ), free electrons from n-doped region (from surface trap
charges if both sides are undoped, explained in separate section below) will tend to compensate the polarization
induced charge resulting in the formation of a 2DEG at the interface [9]. The saturation electron velocity in
AlGaN/GaN HEMT is 1-2E13 cm-2 compared to 2E12 cm-2 in Si(Sige) [1]. Other physical properties of
GaN(Gan/AlGan) has been compared with other systems in the table below [1].
Table 1: Physical Properties of AlGaN/GaN compared with other semiconductors [1]
In an N-doped AlGaN/GaN HEMT, the source is the n-doped AlGaN besides surface traps and give rise to even
more electron density at the channel.
Figure 4: Energy Band Diagrams before junction formation and (b) after junction formation in a HEMT [7].
Now let’s take a look at the channel formation with the aid of the energy band diagram.
I. N-Doped AlGan in GaN/AlGaN HEMT
We know that the band gap of AlGaN (is higher than GaN. The n-doped AlGaN will have its Fermi level
closer to the conduction band. Again, as discussed earlier, due to the non-centrosymmetric nature of the
crystal lattice, there is some polarization in AlGaN as well as in the GaN side. Due to this there is some band
bending at closer to the junction. As a result of band bending, the fermi level crosses the fermi-level, as a
result the free electrons from the valence band of the AlGaN (n-doped region diffuses into the GaN side
leaving the net Positive charge on the AlGaN side. The transferred electrons are now confined in a potential
well and are confined in a two-dimensional plane between the two layers forming a 2DEG [13] which is
shown with the aid of energy band diagram in fig. (5) below.
Figure 5: Energy bands bending in n-doped AlGaN/GaN HEMT forming 2DEG.
As the dopants are away from the dopant, they experience no hindrances in their mobility in the two-
dimensional plane [2]. HEMT transistors are able to operate at higher frequencies than ordinary transistors,
up to millimeter wave frequencies, and are used in high-frequency products/applications such as cell phones,
satellite television receivers, voltage converters, fast charging applications and radar equipment. They are
widely used in satellite receivers, in low power amplifiers and in the defense industry [1].
II. Undoped AlGaN in GaN/AlGaN HEMT
In an undoped AlGaN/GaN HEMT, there is bending of energy bands at the junction due to charges on either
side of the junction created as a result of spontaneous polarization. In theory, as a result of this polarization
charges (which are bound charges), there should be no free electrons to form a 2DEG. But in an undoped
AlGaN, the surface traps will donate and supply free electrons which will migrate to the junction and form
the 2DEG [13] which is explained with the aid of energy band diagrams in fig.(6) below.
Figure 6: Energy bands bending in undoped AlGaN/GaN HEMT forming 2DEG
5. Operation of HEMTs
As we have seen earlier, as in traditional CMOS FETs, a HEMT also has a source, a drain and a gate. The
difference is, as discussed in the earlier section, the 2DEG channel is normally being formed in HEMTs. Thus,
as soon as the source and drain see the potential difference, there is current flowing through the device without
any gate bias. Thus, HEMTs in general are normally ON device [2]. There are 4 modes of operation. This is the
operation in accumulation mode and designers would not like this as there is no control in this device. Thus, this
device is better used when operated in depletion mode. This happens when gate voltage Vg>Vth is applied, the
2DEG under the gate gets depleted and the channel is open and HEMT is off. Thus, by applying the proper
voltages at Gate, source and Drain, HEMT can be used as a high-power switch.
Figure 7: Four operation states of an GaN/AlGaN HEMT [14].
References
[1] B. Syamal and A. Kundu, “AlGaN/GaN HEMT Modeling and Simulation,” in Handbook for III-V high electron
mobility transistor technologies, S.l.: CRC PRESS, 2020, pp. 251–273.
[2] M. Chakraverty, “Motivation Behind High Electron Mobility Transistors,” in Handbook for III-V high electron
mobility transistor technologies, S.l.: CRC PRESS, 2020, pp. 1–44.
[3] S. Nakamura and M. R. Krames, "History of Gallium–Nitride-Based Light-Emitting Diodes for Illumination,"
in Proceedings of the IEEE, vol. 101, no. 10, pp. 2211-2220, Oct. 2013, doi: 10.1109/JPROC.2013.2274929.
[4] H. Amano, M. Kito, K. Hiramatsu, and I. Akasaki, ‘‘p-type conduction in Mg-doped GaN treated with low-
energy electron beam irradiation (LEEBI),’’ Jpn. J. Appl. Phys., vol. 28, pp. L2112–L2114, 1989.
[5] T. Mimura, "The early history of the high electron mobility transistor (HEMT)," in IEEE Transactions on
Microwave Theory and Techniques, vol. 50, no. 3, pp. 780-782, March 2002, doi: 10.1109/22.989961.
[6] M. N. A. Aadit, S. G. Kirtania, F. Afrin, Md. K. Alam, and Q. D. M. Khosru, ‘High Electron Mobility
Transistors: Performance Analysis, Research Trend and Applications’, Different Types of Field-Effect Transistors -
Theory and Applications. InTech, Jun. 07, 2017. doi: 10.5772/67796. High Electron Mobility Transistors:
Performance Analysis, Research Trend and Applications | IntechOpen
[7] R. Komaragiri, “Introduction to High Electron Mobility Transistors,” in Handbook for III-V high electron
mobility transistor technologies, S.l.: CRC PRESS, 2020, pp. 45–64.
[9] O. Ambacher, B. Foutz, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu, M. Murphy, A. J. Sierakowski, W. J.
Schaff, L. F. Eastman, R. Dimitrov, A. Mitchell, M. Stutzmann; Two dimensional electron gases induced by
spontaneous and piezoelectric polarization in undoped and doped AlGaN/GaN heterostructures. Journal of Applied
Physics 1 January 2000; 87 (1): 334–344. https://doi.org/10.1063/1.371866
[10] U. K. Mishra, P. Parikh and Yi-Feng Wu, "AlGaN/GaN HEMTs-an overview of device operation and
applications," in Proceedings of the IEEE, vol. 90, no. 6, pp. 1022-1031, June 2002, doi:
10.1109/JPROC.2002.1021567.
[11] McClory, John. (2008). The Effect of Radiation on the Electrical Properties of Aluminum Gallium
Nitride/Gallium Nitride Heterostructures. 176.
[12] D. A. Neamen, Semiconductor physics and devices : basic principles. New York, Ny: Mcgraw-Hill, 2012.
[13] M. H. Jewel, “Origin of 2DEG in GaN HEMT,” www.youtube.com, Mar. 22, 2020.
https://www.youtube.com/watch?v=KaAwKqYcfN0&t=193s (accessed May 09, 2023).
[14]“GaN Power devices - the HEMT,” www.youtube.com.
https://www.youtube.com/watch?v=UJqY00xPWmY&t=536s (accessed May 09, 2023).
Other Reference materials reviewed/watched:
(1) Hetrostructure p-n junctions - YouTube
(2) Gallium nitride (GaN): the future of power electronics - Bing video
(3) What is Gallium Nitride (GaN) ? GaN in Electronics - YouTube
(4) GaN Power devices - Physics of GaN devices - YouTube
(5) [11] McClory, John. (2008). The Effect of Radiation on the Electrical Properties of Aluminum Gallium
Nitride/Gallium Nitride Heterostructures. 176.
(6) [8] Piezotronic effect tuned AlGaN/GaN high electron mobility transistor - IOPscience
(7)
