Buffer Related Dispersive Effects in Microwave GaN HEMTs
Doctoral thesis, 2018
In applications such as mobile communication and radar, microwave power generation at high frequency is of utmost importance. The GaN HEMT offers a unique set of properties that makes it suitable for high power amplification at high frequencies. However, their performance is limited by trap states, leading to reduced output power and time variant effects. Furthermore, for good high frequency performance a high efficiency it is essential to limit the access resistances in the transistor. The GaN HEMT technology has long lacked a good ohmic contact with good reproducibility.
In this thesis, three buffer designs are considered; C-doped GaN, AlGaN back barriers and a thin GaN structure. The three designs are evaluated in terms of trapping effects using the drain current transient technique. For the C-doped GaN buffer, trapping at dislocations covered with C-clusters is believed to be the main factor limiting output power. Dislocations are presumed to play a major role for the trapping behavior of AlGaN back barriers and the thin structure as well. The maximum output powers for C-doped GaN, AlGaN back barriers and the thin structure are 3.3, 2.7, and 3.9 W/mm at 30 GHz. The output power is found to be limited by trapping effects for all buffer designs.
Moreover, a Ta-based, recessed ohmic contact enables a contact resistance of down to 0.14 Ωmm. The results also indicate that a highly reproducible process might be possible for deeply recessed contacts. An optimized AlGaN/GaN interface shows high mobility \textgreater2000 cm2/Vs without the use of an AlN-exclusion layer. The improved interface also decreases trapping effects and the gate-source capacitance at large electric fields compared to an unoptimized interface.
In this thesis, three buffer designs are considered; C-doped GaN, AlGaN back barriers and a thin GaN structure. The three designs are evaluated in terms of trapping effects using the drain current transient technique. For the C-doped GaN buffer, trapping at dislocations covered with C-clusters is believed to be the main factor limiting output power. Dislocations are presumed to play a major role for the trapping behavior of AlGaN back barriers and the thin structure as well. The maximum output powers for C-doped GaN, AlGaN back barriers and the thin structure are 3.3, 2.7, and 3.9 W/mm at 30 GHz. The output power is found to be limited by trapping effects for all buffer designs.
Moreover, a Ta-based, recessed ohmic contact enables a contact resistance of down to 0.14 Ωmm. The results also indicate that a highly reproducible process might be possible for deeply recessed contacts. An optimized AlGaN/GaN interface shows high mobility \textgreater2000 cm2/Vs without the use of an AlN-exclusion layer. The improved interface also decreases trapping effects and the gate-source capacitance at large electric fields compared to an unoptimized interface.
trapping effects
C-doping
AlGaN/GaN interface quality
recessed ohmic contacts
GaN HEMT
buffer design
Kollektorn, MC2, Chalmers University of Technology, Kemivägen 9
Opponent: Gaudenzio Meneghesso, Department of Information Engineering, University of Padova, Italy
Author
Johan Bergsten
Chalmers, Microtechnology and Nanoscience (MC2)
Low resistive Au-free, Ta-based, recessed ohmic contacts to InAlN/AlN/GaN heterostructures
Semiconductor Science and Technology,;Vol. 30(2015)p. 105034-
Journal article
Performance Enhancement of Microwave GaN HEMTs Without an AlN-Exclusion Layer Using an Optimized AlGaN/GaN Interface Growth Process
IEEE Transactions on Electron Devices,;Vol. 63(2016)p. 333-338
Journal article
Dispersive Effects in Microwave AlGaN/AlN/GaN HEMTs With Carbon-Doped Buffer
IEEE Transactions on Electron Devices,;Vol. 62(2015)p. 2162-2169
Journal article
AlGaN/GaN high electron mobility transistors with intentionally doped GaN buffer using propane as carbon precursor
Japanese Journal of Applied Physics,;Vol. 55(2016)
Journal article
J. Bergsten, M. Thorsell, D. Adolph, J.-T. Chen, O. Kordina, E. Ö. Sveinbjörnsson, N. Rorsman, Electron Trapping in Extended Defects in Microwave AlGaN/GaN HEMTs with Carbon Doped Buffers
This thesis deals with gallium nitride (GaN) high electron mobility transistors (HEMTs). The GaN HEMT is a transistor technology that is well suited for amplification of electrical signals at high frequencies and high power levels. Therefore, the GaN HEMTs are interesting for applications within i.e. mobile communication and security systems. In mobile communication solutions, a high operating frequency means that high data rates are achievable, and with a high power signal, a larger distance between sender and receiver is possible. Using GaN HEMTs in these systems will increase data rates and decrease power consumption compared to current technology. In radar applications, a large power signal means that the radar to detect objects that are further away. Furthermore, the high robustness of GaN HEMTs is highly valued in radar applications since it can survive jamming signals with high input powers meant to interfere or even damage the radar system.
GaN HEMTs are already commercially available and are to some extent used in the applications described above. However, some issues related to the technology limits a more widespread use. This thesis deals with some of these problems. For example, a long-standing issue for GaN HEMTs are so called trapping effects. These effects give the transistor a form of memory, meaning that its performance in the present is a function of what signals it has experienced in the past. These effects are due to different defects in the GaN material and can severely limit the performance of the transistors. Furthermore, GaN HEMTs are currently expensive to fabricate. A reliable fabrication process is of utmost importance to reduce the associated costs of the transistors.
In this thesis, trapping effects associated with defects in the material are characterized in order to understand their origin. With this information it may be possible to decrease the trapping effects in future transistors. Moreover, the results imply that a manufacturing process with high repeatability is achievable, which simultaneously can yield transistors with low loss. Overall, this leads to future communication and radar systems which offers higher performance while consuming less power at a reduced cost.
GaN HEMTs are already commercially available and are to some extent used in the applications described above. However, some issues related to the technology limits a more widespread use. This thesis deals with some of these problems. For example, a long-standing issue for GaN HEMTs are so called trapping effects. These effects give the transistor a form of memory, meaning that its performance in the present is a function of what signals it has experienced in the past. These effects are due to different defects in the GaN material and can severely limit the performance of the transistors. Furthermore, GaN HEMTs are currently expensive to fabricate. A reliable fabrication process is of utmost importance to reduce the associated costs of the transistors.
In this thesis, trapping effects associated with defects in the material are characterized in order to understand their origin. With this information it may be possible to decrease the trapping effects in future transistors. Moreover, the results imply that a manufacturing process with high repeatability is achievable, which simultaneously can yield transistors with low loss. Overall, this leads to future communication and radar systems which offers higher performance while consuming less power at a reduced cost.
Subject Categories
Computer Engineering
Aerospace Engineering
Other Electrical Engineering, Electronic Engineering, Information Engineering
Infrastructure
Kollberg Laboratory
Chalmers Materials Analysis Laboratory
Nanofabrication Laboratory
Areas of Advance
Nanoscience and Nanotechnology
Materials Science
ISBN
978-91-7597-726-3
Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 4407
Publisher
Chalmers
Kollektorn, MC2, Chalmers University of Technology, Kemivägen 9
Opponent: Gaudenzio Meneghesso, Department of Information Engineering, University of Padova, Italy