The relentless demand for more spectrum and more bandwidth, along with unique application characteristics of electromagnetic radiation at different wavelengths, is leading to persistent efforts for designs operating at ever-higher frequencies. While 1 GHz to 10 GHz was considered a wild frontier just a few decades ago defined by cumbersome, almost magical active and passive components with very few commercially available devices for mass-market products, we are now routinely in the low single-digit GHz space with smartphones, Wi-Fi, and IoT devices. The 5G advance is pushing past that zone into the tens of GHz.
Today’s 5G cellular networks operate at 6 GHz and below with future 5G communications operating at millimeter wave frequencies spanning from around 24 GHz to 53 GHz. The next generation, dubbed 6G, will push frequencies even higher for even wider bandwidths and channel capacities.
But why stop there? Defense Advanced Research Projects Agency (DARPA) recently issued an announcement regarding the Electronics For G-Band Arrays (ELGAR) project. The agency wants industry to develop monolithic microwave integrated circuits (MMICs) and front-end receiver arrays able to operate in the sub-terahertz G-band frequency range between 110 GHz and 300 GHz for applications such as radio astronomy, remote security sensing, and telecommunications (Figure 1).
Figure 1 The spectrum chart shows the many orders of magnitude of electromagnetic energy; terahertz waves span 100 GHz to 10 THz zone in the broader area labeled as microwaves. Source: Rutgers University
DARPA has established some goals for MMIC devices in this band. That includes power amplifier efficiency of 30%, transmit-array areal power density of at least 34 W/cm2, and transmit-array efficiency of 24% at 220 GHz. The upper millimeter-wave G-band of 110 GHz to 300 GHz is, at first glance, an attractive, underused span of the electromagnetic spectrum, especially above 200 GHz where atmospheric absorption is low.
Due to a combination of historical circumstances and different application jurisdictions, the G-band designation has also been used for 4 GHz to 6 GHz, which is very different from the 100 GHz to 300 GHz spectrum now defined as G-band. There is some confusion as different applications—radar, radio astronomy and military—have traditionally used different letter designations such as X, S, K for the same frequency/wavelength bands. See TeraSense’s note on Radio Frequency Bands for some insight on this situation.
In recent years, the industry has increasingly used IEEE Standard 521-2002 titled “IEEE Standard Letter Designations for Radar-Frequency Bands” for the designations aligning with the International Telecommunications Union (ITU) definitions. However, many of the older designations are still in wide use, which can be a source of confusion.
I have no idea what the DARPA-initiated effort will yield, but it’s worth following the developments. While the G-band is already in limited but highly specialized use and does have non-MMIC components, it’s a very hard place in which to design real products. For example, RF isolators are standard components in the RF world used for separating RF energy traveling in two directions over the same path as transmit and receive signals.
Take the case of the Faraday rotation isolator, which is in use since the 1960s. It’s widely used in the higher-frequency regions and the short application note titled “Isolators Designed for Low-Insertion Loss” from Micro Harmonics Corp. gives you some perspective into the challenges of developing useful components for this range. Some of the isolators include diamond-layered heatsinks for enhanced thermal performance.
MMIC progress in G-band
As developments reaching into the G-band make MMIC-related progress while using various materials, process technologies, support components, and test and measurement instrumentation, I suspect the next push will be up and into the terahertz band. That “strange” area of 300 GHz to 3,000 GHz—that some say starts at 100 GHz and some say ends at 1,000 GHz—which falls between conventional RF and the optical and infrared regions, as discussed in the IEEE Spectrum article titled “The Truth About Terahertz”. While both RF and optical are governed by Maxwell’s equations, they exhibit very different characteristics and require very different components.
There is university research work being done now which is using the overlap between these two ranges to develop hybrid photonic devices that are far removed from discrete lasers, LEDS, and phototransistors. For example, a joint team from Osaka University, Japan and the University of Adelaide, Australia has developed a four-channel multiplexer that supports a 48-Gbps data stream with modulation centered at 350 THz (Figure 2). They used silicon as a base material along with standard fabrication techniques.
Figure 2 It’s a silicon-based four-channel multiplexer supporting 48-Gbps data stream with modulation centered at 350 THz. Source: University of Adelaide
Another development is from a team based at Princeton University. They have devised a programmable terahertz metasurface with fully integrated silicon chip tiles for dynamic beamforming (Figure 3).
Figure 3 This programmable terahertz metasurface uses fully integrated silicon chip tiles for dynamic beamforming. Source: Princeton University
If history is any guide to the leveraging of technology advances, I suspect that MMIC progress in the G-band will subsequently lead to more progress and relatively standard components and techniques in the terahertz band. Once that starts and the overlap zone between the terahertz and optical ranges blurs, some impressive developments may occur. Perhaps, the long-awaited optical processors to carry out neural-network calculations with photons instead of electrons will become a reality.
Do you have any experience or even interest in G-band developments? What about terahertz design, electro-optics, and photonics in that region and above? What do you think?
- Why 5G Impresses and Worries Me
- Terahertz Waves for Production-Line Inspection?
- Terahertz waves and acoustic modulation: A datalink “Odd Couple”?