DARPA has long recognized the critical role transistors play in Department of Defense (DoD) systems – especially radio frequency (RF) systems ranging from radar and communications to signals intelligence and electronic warfare. In the 1990s, the DARPA MIMIC program advanced Gallium Arsenide (GaAs) transistor technology that enabled the radar and communication systems in use today. More recently, the development of Gallium Nitride (GaN) technology in the DARPA WBGS-RF, NEXT and MPC programs has advanced the ability to deliver high power RF signals at higher frequencies, bandwidths and efficiencies than earlier Silicon and GaAs technologies. However, while GaN transistors are now being adopted for many platforms, the use of the electromagnetic spectrum is evolving in new directions that motivate new directions in transistor technology.
The large number of commercial and military RF signals in use today has led to a complex and crowded electromagnetic environment. The demand for even more RF signals is growing and driving both an increase in the density of signals within frequency bands, as well as pushing the utilization of higher operating frequencies particularly into millimeter wave frequencies (i.e. frequencies above 30 GHz). As a result, it becomes extremely important to expand dynamic range in RF systems due to in-band interfering signals, which require the capture of small signals in the presence of large interfering signals. The problem of realizing high dynamic range RF systems is that for operation in a complex spectrum with large signal-to-noise ratios, the intrinsic linearity and power density of the transistor technology fundamentally limits the ability of transceivers to process RF signals efficiently with large bandwidth and high fidelity.
“The same basic transistor types have been dominant since their invention and we have been engineering the heck out of them for 50 years,” said Dan Green, a program manager in DARPA’s Microsystems Technology Office (MTO) and the overseer of the DREaM program. “We’ve gotten a lot out of that approach, but the focus on so few types of transistor technologies and just a few semiconductor materials also has fundamentally limited us in the RF world. With DREaM, we want to rethink all of that and imagine new possibilities.”
Because of the importance of transistors for the generation and processing of signals at RF, microwave, and millimeter-wave frequencies—for such applications as communications, surveillance, signal intelligence (SIGINT), electronic warfare (EW), and electronic countermeasures (ECM)—DARPA has launched its Dynamic Range-enhanced Electronics and Materials (DREaM) initiative in 2017. DREaM is intended to encourage the exploration of new semiconductor structures and materials, with the aim of more efficiently generating and processing high-frequency signals for defense-related applications.
Next generation wireless systems such as 5G cellular communications, front-haul and back-haul networks, military and automotive radar, and IEEE 802.11ad and 802.11ay WiGig are targeting a range of new capabilities including higher bandwidth, more connected devices, low latency, and better coverage. To address the wide bandwidth requirements, researchers are exploring the higher frequencies in the centimeter and millimeter wave bands where more spectrum is readily available. DARPA’s desire for new transistor technologies is based on the need to process a growing number of wireless signals at RF, microwave, and even millimeter-wave frequencies.The overcrowding of electromagnetic (EM) signals has been well publicized in commercial markets. But it is also taking place in defense-related applications, with communications systems, various forms of radar, jammers, EW, systems all competing for available bandwidth at different frequencies. For a given application, such as a portable military radio, other applications will appear as noise in a congested EM environment.
Unfortunately, while GaAs, and more recently, GaN technology has advanced the ability to reach higher power densities, it has not fundamentally changed the power requirements for the linearity performance in part because the transistor design leveraged canonical structures that have evolved slowly since the initial conception of the transistor. In order to address to core capabilities of transistor technology, a new approach to the materials and device structure will be required, said DARPA. Specifically, DARPA is interested in new material options and transistor architectures to enable breakthrough dynamic range in millimeter wave systems. Metrics targeted for improvement at the device level include RF power density, efficiency, and linearity.
By opening the way to such advances with the DREaM program, DARPA hopes to create new RF/mm-wave transistors that provide the foundational capability to address challenges associated with the increasing need to access, make use of, and manage the electromagnetic spectrum. As both commercial and military users look to higher, millimeter-wave frequencies and their available bandwidths, DARPA’s vision is for new transistor topologies that can provide the high gain and low noise figures at those higher frequencies to enable systems to operate effectively and with high efficiency, even when surrounded by the noise of the many other wireless applications.
Operating in a crowded and contested spectral environment, high dynamic range transistors are required in transmitting and receiving circuits. Over the past decades, RF power transistors have been implemented on the well-established planar device topology, in which it is difficult to attain high power with high efficiency at mm-wave frequency. The DARPA Dynamic Range-enhanced Electronics and Materials (DREaM) program is developing advanced high power and high dynamic range transistor technology with 4× higher output power density and 100× better linearity compared to state-of-the-art transistors today. The program explores novel wide bandgap materials. multi-channel epitaxy, and new device innovations such as FinFET-like compound semiconductor device structures. This paper will provide an overview of recent device demonstrations with significantly higher linearity and power density at 30 GHz.
The fundamental capabilities of any transistor technology can be measured through several figures of merit. For linearity, the output third order intercept point (OIP3) is the common metric which captures the theoretical output power point where output parasitic harmonics are generated at an equal power level to the output fundamental signal. Transistors today typically obtain high OIP3 at the expense of direct current power (PDC). The metric of OIP3/PDC assesses the linearity of a transistor technology and has produced a 10 dB rule of thumb that seems to be semiconductor material independent. At low frequencies, circuit techniques can be used to increase the linearity of transmitters or receivers, but the circuit techniques become difficult to implement or are no longer feasible in the millimeter wave regime . DARPA is looking to create a new class of transistor technology that can surpass the 10 dB rule by 100X.
In addition to the demand for higher linearity transmitters, there is also a competing demand on transmitter size, weight and power (SWaP) that prevents simply scaling up existing technologies to meet total output power requirement. Thus, a transistor technology that has increased power density must be developed to meet this need. Increasing power density in the millimeter wave region is even more challenging because power density decreases as frequency is increased.
DARPA is looking to elevate the performance of transistor technology by increasing the power density at millimeter wave to 20 W/mm.
The DREaM program aims to overcome the above stated challenges and realize high dynamic range RF transistors for a diverse set of amplifier applications by developing non-traditional materials, integrating new device structures, and innovating transistor layout. To realize higher dynamic range than what is achievable today, the output power density and the intrinsic device linearity must be significantly improved. It is anticipated that new concepts will need to be combined in order to achieve the program objectives.
Several experimental research results encourage the potential for radical innovation. For example, a GaN high electron mobility transistor (HEMT) with nanowire channel has shown flat transconductance,4 which implies the possibility to improve the transistor linearity by
engineering the channel geometry. In contrast to the wideband gap GaN, a small bandgap carbon-nanotube field effect transistor (FET) has also demonstrated flat transconductance with a concurrent projected improvement in OIP3.5 In addition to these results, which suggest
opportunities to engineer linearity, significant new materials options exist to raise power density. A nitrogen-polar GaN FET recently achieved record 6.7 W/mm at 94 GHz,6 which demonstrates 3X higher output power density than Ga-face GaN by growing the crystal structure in a completely new manner. Furthermore, a two-dimensional electron gas density larger than ~2 x 1014 cm-2 was measured in a complex oxide system (a SmTiO3/SrTiO3 interface), which is about 10X higher than state-of-art GaN material system.
The examples highlighted here are intended to illustrate the range of emerging device possibilities to enable 4X higher output power density
and 100X better amplifier linearity compared to the state of the art are projected but, importantly, are not intended to limit or recommend any specific solution. Because these new materials and device concepts are dramatically different from those concepts currently employed in the industry, it presents an extremely challenging and risky development.
The DREaM program will develop new materials and integrate them into devices to make the next leap in RF transistor performance for future high-dynamic-range RF systems.
To achieve these technical goals with the DREaM program, Green hopes to push transistor technology futureward along two mutually reinforcing directions. One centers on new materials that can accommodate more electrical charge and voltage without degrading than can currently known materials. Advances on this materials front could open the way to the higher-power and more capable transistors Green seeks to develop with the DREaM program. Ultrawide bandgap materials (UWB) such as complex oxides, which include gadolinium titanate and strontium titanate, and even particular crystal variants of GaN, are among the many possible candidates for research in this area. Researchers engaging in the second technical area, which focuses on the linearity of transistor behavior over wide ranges of signal frequencies, will investigate unconventional transistor structures, among them nonplanar and filamentous ones, such as ones based on carbon nanotubes, as well as still-to-be-imagined geometries and layouts that are not constrained by the row-and-column transistor formats of today’s integrated circuits.
The DREaM program will develop new materials and novel device structures to create RF/millimeter wave transistors that enable high dynamic range RF systems. Such RF systems fundamentally require either high transmitting power to increase the signal strength or high
linearity signal reception to minimize the spurs or noise in the spectrum. The DREaM program will dramatically increase the output power density at the transistor level as compared to present GaN technology. In addition, DREaM devices with intrinsically higher linearity will enable circuit and system designs with superior reception at much lower power consumption penalties.
Thus, DREaM technology will enable RF transceiver systems to achieve the same or better RF specifications as today while consuming much lower DC power, which will benefit systems from large phased array applications to small apertures on power-constrained platforms. Overall, the DREaM technology is anticipated to be foundational and impact a broad array of RF and millimeter wave (MMW) applications.
The goal of the DREaM program is to develop new transistor technologies that can achieve 4X higher output power density and 100X better amplifier linearity for a given direct current (DC) power in the millimeter regime (30 GHz) than is available today.
DREaM program seeks enhancement to intrinsic transistor performance, and does not seek investment in any circuit design techniques. The DREaM program is open to all possible materials and device structure approaches beyond those described above as long as the proposed
transistor technologies will meet the program metrics.
An ECE team awarded a DARPA $4.9M DREaM grant
The MSU ECE Department was selected by the Defense Advanced Research Projects Agency (DARPA) to perform research under the Dynamic Range-enhanced Electronics and Materials (DREaM) program. Under the project, “Development of Millimeter-Wave High Power Density Diamond-Collector Heterojunction Bipolar Transistors,” MSU will lead a team including UW-Madison, the Fraunhofer Center for Coatings and Diamond Technologies, and the University at Buffalo. The goal of DREaM is to explore material properties and device concepts that go beyond the seemingly inescapable performance tradeoffs between four key characteristics of RF transistors: (1) signal power, which determines an RF system’s range of operation, (2) power-added efficiency, which determines the size and weight of the power system required to run them, (3) the range of frequencies (bandwidth) of transistor performance, and (4) the intrinsic device linearity, a measure of the ultimate fidelity at which a receiver can amplify signals, including weak ones that otherwise would get lost in the cluttered spectrum. The MSU-led approach allows disparate emitter and base materials to be combined with a diamond collector to enable higher power and voltage gain (dynamic range) in contrast to deeply-scaled transistor concepts that result in greatly reduced operating voltages to obtain wide frequency bandwidth performance. The MSU team is led by John Albrecht, Professor of ECE, Timothy Grotjohn, Professor of ECE, and John Papapolymerou, MSU Foundation Professor of ECE.
HRL Moves To Next Phase Of DARPA GaN Project
HRL Laboratories, an R&D lab in California owned by Boeing and General Motors, has announced a significant step in the development of next generation of GaN transistors for communications, radar, and 5G wireless networks. The research team, led by principal investigator Jeong-Sun Moon, has successfully met and exceeded the performance metrics defined by the Dynamic Range-enhanced Electronics and Materials (DREaM) program, a Defense Advanced Research Project Agency (DARPA) effort to improve dynamic range in millimetre-wave (mm-wave) electronics.
HRL has demonstrated a low-noise GaN HEMT with a record linearity for such devices – the ratio between output third-order intercept power (OIP3) and DC power consumption (PDC). The OIP3/PDC of 20 dB at 30 GHz was achieved, at least ten times greater than conventional GaN HEMTs [International Microwave Symposium, 2019]. In parallel, HRL’s DREaM GaN transistors demonstrated state-of-the-art power-added efficiency (PAE) of greater than 70 percent at 30 GHz, a vast improvement over reported PAE of other mm-wave T-gated AlGaN/GaN HEMT devices [Electronics Letters, April, 2020].
References and resources also include: