In my younger daysâ€”about the time that Erik the Red was making a name for himselfâ€”I was really into electronics. Countless never-quite-working-as-expected circuits should have taught me the futility of telling electrons what to do. YetÂ my interest in electronics peaked with the construction of an electronically steerable phased-array antenna. This is where, by varying the timing slightly, numerous small antennas create a signal that can be sent in specific directions without moving any hardware.
Yes, my set-up did actually work, though not as well as I’d hoped. Anyway, what excited me about phased-array antennas isÂ that you could shape and steer an antenna’s radiation pattern by individually controlling the phase and amplitude of a string of individual emitters. It just seemed so cool. Later, when I moved on to optics, controlling the phases and amplitudes of individual lasers and combining them into a single, steerable laser beam… well, it was technically possible, but there was a vast gulf between our ideals and anyÂ practical implementation.
But recently, researchers have shown that phase control is possible in a device that is smaller than the wavelength of the light being controlled. Although a rather technical development, this is one key step along the road to high-capacity optical communications that don’t involve any fibers. Think mobile communications beyond 5G, or home Wi-Fi that actually doesn’t suck.
How hard can it be?
The idea that we can control the amplitude and phase of a bunch of emitters seems simple enough. Imagine that you want to do this with your Wi-Fi. The frequency of a Wi-Fi source is 5GHz, which means the wavelength isÂ 6cm. Now, let’s imagine that I have a set of 16 antennas arranged in a 4Ã—4 array. If I want to control the phase of the radiation going to each antenna element, then I need to ensure that each feed has the same length (to within, say, a 20th of a wavelength: 3mm). That is relatively easy to do.
Controlling the phase and amplitude is then a matter of ramping up or down the power to each antenna element while having some kind of variable capacitor in the circuit to apply a variable delay to each element. Although this sounds pretty easy, I point you to British Satellite Broadcasting’s fake phased-array antenna as a hint that phased-array antennas are not entirely trivial.
Now, let’s move to the optical regime. We canÂ go easy on ourselves and stick to the infrared, with a wavelength of about 10Î¼m. The feeds to our 4Ã—4 array need to have a length accuracy of about 500nm. That seems doable, but remember: with light, it’sÂ not just the length that matters. If the feed lines are just 40Î¼m long, then the refractive index of the feed lines cannot vary by more than a percent. The longer the feed lines, the more similar the refractive indices have to be. The manufacturing tolerances become more and more difficult to meet as the wavelength gets shorter, or the distance between emitters grows.
So, to sum up: a phased-array with lightÂ is doable and, in fact, is done on a regular basis under carefully controlled lab conditions. But it’s not easy. And we still haven’t reached the part when we change the phase and amplitude of the light to each array element yet.
Building a phase-shifter
Changing the phase of a light field is surprisingly easy: just make the light travel a little bit further to reach its destination or make it travel slower. This can be done by two methods: physically extending the path for the light to travel, or changing the refractive index of the material that the light travels through. The latter is usually more convenient, but most materials only let you change the refractive index a small amount. That means that any device that controllably changes the phase of light has to be quite longâ€”if the change in refractive index is small, you need to apply it over a longer distance to get the desired phase change. And it needs to be uniform over the entire length.
A better approach is to change the refractive index a large amount. To do that, the light needs to respond strongly to the electrons. That means using conductors, like aluminum or gold. Unfortunately, light that travels through a metal tends to be reflected or absorbed.
The exception is when light cooperates with the electrons in the metal to generate a surface plasmon polariton. In this case, the light and the electron motion combine to create a slowly moving wave that travels along the surface of the metal. If you were to calculate a refractive index, based on the speed of propagation, you come to a number in the order of 100 (glass is 1.3, and most normal materials lie in the range of one to five).
The key thing about the motion of a plasmon is that its speed of propagation depends very sensitively on the number of electrons available to move.
Graphene to the rescue
This is where grapheneâ€”a single layer of carbon atoms arranged in a honeycomb structureâ€”comes in. Graphene is a conductor, so it can support a surface plasmon. But it is not a metal; it is actually a semiconductor, like silicon. The difference between graphene and silicon is inÂ the energy gap between electrons that are free to move and electrons that are confined to nearby atoms: this is exactly zero for graphene, while it is about a volt for silicon. When an electric field is applied to graphene, the additional energy shoves more electrons into the conducting state, effectively speeding the plasmon on its way.
To get the full effect, though, the graphene needs to be quite sheltered from the outside world. If you just slap a layer of graphene on any old surface, then the surface plasmon radiates into the surface and leaves the building. You have adjusted the phase of… nothing. And that is a key point to this research: the scientists found a very inert substrate (boron nitride) to encapsulate the graphene. Having successfully enclosed the graphene in boron nitride, the researchers were able to generate surface plasmons that propagate with low losses for the entire length of the phase-shifting device.
The end result is a device that is about 600nm long but has a refractive index that varies from 80 to 160, allowing the incident light field to be shifted by a full cycle (a phase change of 360 degrees).
Since the heart of the device is graphene, the light has to have a rather long wavelength to generate a plasmon, so we are talking about infrared light, rather than visible. Indeed, the wavelength of the light in free space is more than 10 times longer than the entire phase-shifting device, which is mind-bogglingly cool.
(You might wonder how you can manage a full 360-degree phase shift if you seemingly can’t even fit an entire wavelength in the device. The answer is that the wavelength in the device is the wavelength of the plasmon, which is much shorter, so you can fit multiple wavelengths in the phase-shifter. That’s because the refractive index is about 100, which reduces the wavelength by about 100. That reduction makes the entire device much longer than the effective wavelength of the plasmon.)
What does all this have to do with antennas?
I started off talking about phased-array antennas, where the direction and shape of the radiation pattern is controlled by adjusting the phase and amplitude of the radiation emitted from each antenna element. And now, at least for the infrared, we have an effective phase-shifter. It’s not bulky, and it’s relatively robust. Does this mean that infrared phased-array antennas are on the horizon?
I’d sayÂ thatÂ yes, in the lab, they are. And, I think they will find applications relatively soon. For mobile, the 5G specification already allows for (and might even require) the ability to shape and steer antenna patterns so that a high-capacity link can be maintained between a moving mobile device and a fixed antenna. I think this is a clear sign that high-capacity links will require some sort of beam steering.
And for years, researchers have been thinking about using free-space light signals for high-capacity data links. To doÂ that, dynamically steering multiple laser beams is required. To do that, you need multiple element light-antennas, and you need to be able to control the phase and amplitude of each emitter. So, this isn’tÂ the next generation of wireless, but perhaps the generation after that.
Nature Photonics, 2017, DOI: 10.1038/NPHOTON.2017.98
from Ars Technica http://ift.tt/2ueMrS3