Sometimes you can’t seem to go a week without hearing new lidar news. This is disconcerting, as lidar was the desperation application for lasers for a very long time. If you had made a new pulsed laser and couldn’t think what it might be good for, you figured out something that needed a measurement at distance and claimed your laser was useful for that application. At best, you’d give the laser to an atmospheric scientist and get them to measure the density of aerosols in the upper atmosphere.
OK, maybe that’s a bit cynical, but for a very long time, lidar research was an unvisited backwater port on the sea of laser physics. But new research demonstrates how much this has changed: a new device that has wide applications and will probably make a huge impact in optical communications. Yet the device is being sold as great for lidar, which it may well be.
Lidar is hard
It wasn’t that lidar was unattractive to engineers in the past. It’s more that lidar was unwieldy. Lasers were big, the pulse durations were too long, the collection optics to get the signals were large, the electronics were simple… but big. Lidar instruments were delicate. The idea of putting an expensive and breakable device in the grille of a nitrous-injected Honda Civic, driven by a hormonal boy racer, probably didn’t thrill many engineers.
Then powerful lasers got smaller, cheaper, and less delicate. The electronics required for optical communications could be readily adapted to lidar. Suddenly, lidar started to look good.
For lidar to work effectively, you need to efficiently combine a laser beam that’s reflected off an object with a reference beam. The internal routing and external optics form one system, and misalignment between the transmitting and receiving beam result in a directional miscalculation. The easiest solution is to use the same optics to send and receive signals. But this is expensive, requiring high-speed micro-mechanical or electro-optic devices. Much better to have some passive optical circuit that simply reroutes light depending on its function. To do that, we need a non-reciprocal optical device.
A typical non-reciprocal device is called a circulator (see sidebar). These are bulky devices that usually involve passing the light through a relatively strong magnetic field and a special material that is almost certainly looked upon darkly by engineers in charge of fabrication machines. In a circulator, light that enters port 1 will exit at port 2, while light that enters port 2 will exit port 3, and light that enters port 3 will exit port 1 (assuming only three ports).
However, for lidar, what you might want is more complicated. A high-power light pulse needs to go from port 1 to port 2, where it exits to the outside world. The signal that comes back enters port 2 and exits port 3. A weaker reference signal enters port 1 and exits port 3 to mix with the return signal. Under these conditions, the outgoing laser pulse and reflected laser pulse are automatically aligned to each other. The reference and the received pulses are also automatically aligned.
Making silicon refuse to reciprocate
Creating this sort of device is exactly what the researchers have achieved, though in rather artificial circumstances. But they have done it using standard silicon integrated optical fabrication techniques, which means that if it works as expected as a unit, it can be manufactured very cheaply.
The researchers relied on silicon’s nonlinearity to ensure that the device’s behavior would change depending on the brightness of the light in the device. Essentially, if light is bright enough, it can change the properties of the material it is traveling though (this is what happens when you set termites on fire with a magnifying glass). However, in the tiny gap between a material not being changed by light and it catching fire, the change can be useful. Optical engineers spend a lot of time throwing themselves and their expensive kit at this gap with only the occasional survivor emerging without their hair on fire.
The device is basically a pair of waveguides that form four ports, with only three in use. Ports 1 and 2 are directly connected, while port 3 is in a separate waveguide. The researchers created an oval racetrack resonator between the two straight waveguides. In addition, a small defect is placed in the wave guide between ports one and two. This defect acts as a mirror at low-light intensities. However, if the intensity is high enough, the optical properties of the defect are modified, and it allows the light pulse to be transmitted.
Light traffic ahead
A strong light pulse is sent into port 1, where a small portion is transferred to the ring via a process called evanescent coupling. The light in the ring loops around, transfers to the next wave guide, and exits port 3. The light from port 1 that exits port 3 is the reference pulse. But most of the strong light pulse continues on outside the ring, encounters the defect, and modifies it, which stops it from reflecting the pulse. Hence, the remainder of the pulse is transmitted out port 2 and on to the outside world.
However, light that enters from port 2 (light that the lidar would have received from the outside world) is traveling in the opposite direction. The light from port 2 still enters the racetrack but cannot exit at port 3 because it is traveling in the wrong direction. Under ordinary circumstances, it would exit from port 4. This is where the defect plays its role. But when the light intensity is low, the defect is still reflective. It reflects the pulse back in the direction it came from. On the way back, light is coupled into the ring and can exit from port 3, just like the reference signal from port 1.
Note, though, that the transfer of light between the waveguide and ring is reciprocal, which means that about half the light entering from port 2 must exit port 4, and half exits port 3. This sounds bad but actually isn’t. These types of devices often have high losses no matter how they are constructed, so I don’t see 50 percent loss as a huge issue. Indeed, considering that car lidar operates in the 100m range, where signals are relatively strong, it’s probably possible to cope with the additional losses.
The researchers did not demonstrate this with an actual lidar unit. Instead, all the light pulses traveled through optical fiber with varying lengths of delay. The signals they ended up with seem rather strong for a lidar return signal. Nevertheless, this is all made with standard processes, so connecting it up to a real-world system will not be a difficult task.