Today I’d like to present the turbine-powered optical isolator, a steampunk-sounding piece of hardware that only allows light to pass in one direction. That, I must admit, is a sentence I never thought I’d write.
It turns out that it’s very important to have components that allow the light to flow only one way. I personally managed to destroy more than one laser because some of the emitted light ended up returning to the hardware and doing major damage. Even if you don’t destroy anything, back reflections can still ruin experiments. So optical isolators, as they are called, are very important.
They are also large and expensive. To reduce cost and size, researchers have demonstrated an optical oscillator that: nothing but requires a working turbine to function. Despite being the least practical insulator ever, I still want one, maybe even two. Give me even three, ’cause I’m here anyway.
The insulator effectively relies on the Doppler effect to allow light to flow in one direction and not the other. You are probably familiar with the Doppler effect: that is why an approaching ambulance siren has a higher pitch than a retreating ambulance. Light is also subject to the Doppler effect: a retreating light source has a redder color than an approaching light source.
Note that it is the apparently movement that matters. A moving source and a stationary receiver are exactly the same as a stationary source and a moving receiver.
This is important because the critical piece of hardware is a small glass sphere called a whispering gallery mode resonator. You may have experienced a whispering resonator in gallery mode yourself, or at least heard of it. In some cathedrals, sound travels neatly along the walls. Put your head against a wall, talk quietly, and your friend on the other side of the cathedral will hear you clearly. (Or they would if the 10 billion other tourists didn’t try the same trick.)
The point is that the sound will travel along the walls through a series of grazing reflections without passing through the cathedral’s center space. Light does the same in one of these little glass spheres, which move in a ring around the edges.
Unlike a cathedral, the light does not start in the ball. Instead, the ball is placed within a few nanometers of an optical fiber, allowing the light to leak across the opening to the ball. In fact, this spill can be designed so perfectly that all the light goes into the ball.
Our ball is also a bit more selective than a cathedral: it only lets certain colors (or wavelengths) of light through the ring around its perimeter. The rule is this: the circumference of the ball must be equal to half an integer number of wavelengths.
That seems like a simple selection rule, except that light doesn’t measure things the way we do. When light moves from air to glass, it slows down (by about 30 percent), so the light focuses and shortens its wavelength. I won’t go into detail on how this happens, but in order for light to move through a material, it has to move electrons (which is what gives glass an index of refraction greater than one). Electrons are heavy and don’t like to move, which slows down the light. Light moving electrons are key to the story here as it brings the Doppler effect into play.
Are you gonna Go My Way?
To get an idea of how the Doppler effect changes things, let’s look at light entering a sheet of glass moving very quickly toward the light source.
The electrons in the glass see the light as bluer in color than we see. The electrons also react as if the light is bluer. Blue light has to work harder to shift the electrons (its refractive index is greater). As a result, the light slows down even more than if the glass were standing still.
From the perspective of the light, the glass plate appears to be slightly thicker and more wavelengths can fit in it.
If the glass moved with the light, the opposite would happen. The wavelength appears to be redder. The electrons react as if the light is redder (the refractive index decreases). And the light slows down less. From the perspective of the light, the plate also appears a bit thinner and fewer wavelengths fit in it.
turn me on
The researchers achieve the same effect by rotating the glass sphere. Light traveling in one direction through the fiber encounters glass moving towards it. The light has the wrong wavelength to fit into the glass resonator, so it continues down the optical fiber.
But light going the other way has just the right wavelength to fit into the resonator, so it gets sucked in. Once in the resonator, the light cannot return to the fiber without changing direction in some way. Instead, it remains in the ball until it is absorbed or lost to the outside world through imperfections in the glass.
To match the wavelength of the light, the researchers needed a turbine with fairly high speeds: between 120,000 and 400,000 revolutions per minute. This speed is fast enough to ensure that light traveling in the direction of rotation enters the ball and is lost to the fiber, while light traveling in the opposite direction cannot enter the ball.
The spinning also had a nice side effect. To get light into the glass sphere, the fiber has to be placed at exactly the right distance from the sphere. The resistance of the spinning ball creates a cushion of air that stabilizes the fiber at a fixed distance. The researchers can fine-tune the distance by increasing or decreasing the tension in the fiber. This seems to be more stable than for a non-spinning glass sphere, which is pretty cool.
And now for the downside
The insulator works, but only works for a certain color of light. Or more precisely, there is a range of colors that can only propagate in one direction through the fiber. There is a second set of colors that can only propagate in the opposite direction. All other colors can propagate in either direction.
The colors are also very precisely defined. For context, in an optical communication system, a laser with a very narrow color gamut is modulated very quickly to carry information. This modulation is not great; if the laser light were in the visible range, you probably wouldn’t be able to observe that the light color was less pure after modulation. However, the color impurity caused by modulation is enough to prevent the insulator from doing its job. The insulator would work very nicely for the light before modulation, and not at all after modulation. In other words, we cannot use this hardware on light that carries data.
Yet this is the beginning of a new approach to a problem that previously relied on large magnets and special crystals. So it’s nice to see one or two new ideas.
Nature2018, DOI: 10.1038/s41586-018-0245-5