Low optical loss electrical isolation for multi-section monolithic GaSb-based photonic circuits

Imagine a world where light isn't just a beam from a flashlight, but a tiny, high-speed courier delivering data or sensing the air around us. This is the realm of photonic integrated circuits (PICs). Think of these chips as the "motherboards" for light. Just as a computer motherboard has different rooms for the processor, memory, and graphics card, a light chip has different sections: one to generate the light (the laser), one to carry it (the waveguide), and one to control or stop it (the switch or absorber).

For these light couriers to do their job, the different rooms on the chip need to be electrically isolated. Imagine trying to talk to a friend in a noisy room; if the walls are thin, their voice leaks into your conversation, and you can't focus. In a light chip, if electricity leaks from one section to another, it causes chaos, heating up the device and ruining the signal. The challenge is building a wall that stops the electricity completely but lets the light pass through without bumping into it or getting lost. This is especially tricky with a specific type of material called GaSb (Gallium Antimonide), which is great for creating light that travels in the "mid-infrared" range—a special part of the spectrum perfect for sniffing out gases or detecting diseases, but notoriously "leaky" when it comes to electricity.

---

This paper tackles a specific headache in building these GaSb light chips: how to build a wall between two sections of a laser that stops electricity from leaking but doesn't kill the light beam. The researchers found that the usual way of building these chips leaves the electrical walls too thin, allowing current to sneak through and mess up the laser's performance. To fix this, they tried a radical idea: dig a deep trench to separate the sections.

Think of the laser's path as a highway. Usually, the road is a wide, shallow ridge (like a gentle hill). To stop electricity, the team tried digging a deep strip trench (like a deep canyon) right in the middle of the chip. While this deep canyon was fantastic at stopping electrical leaks, it was terrible for the light. It was like trying to drive a car from a smooth highway into a deep, rough ditch; the car (the light) would bounce around, lose energy, and might not make it through at all.

The team's main discovery is a clever "ramp" or taper that connects the smooth highway to the deep ditch. Instead of a sudden drop, they designed a gradual transition where the wide ridge slowly narrows and deepens into the strip waveguide. This adiabatic taper acts like a smooth, winding ramp that guides the light gently from the shallow section into the deep trench without it crashing or scattering.

The results were measured and found to be quite impressive. They tested five different versions of their laser chips. The "baseline" version, with no special tricks, had an electrical isolation resistance of only 121 Ω (ohms), meaning electricity leaked through easily. When they tried just digging the deep trench without the smooth ramp, the light got so scattered that the laser wouldn't even turn on. However, their best design—the one with the deep trench plus the smooth taper—achieved an isolation resistance of 17.3 kΩ. This is about 17 times better than what was previously reported for similar GaSb devices.

In their experiments, this optimized device worked perfectly as a continuous-wave laser, producing about 8 mW of power and staying stable even when they applied a reverse voltage of -5 V to the absorber section. The light remained a single, clean beam, proving that the smooth ramp successfully guided the light through the deep trench. While a version with an even deeper trench achieved a higher resistance of ~40 kΩ, it failed to lase because the light loss was too high, confirming that the "smooth ramp" design is the necessary balance.

The paper explicitly rules out the idea that simply digging a deep trench is enough; without the taper, the optical loss is too severe for the device to work. They also found that simply etching a shallow "leakage prevention" layer wasn't sufficient on its own, as the high conductivity of the p-type layers in GaSb allowed too much current to sneak around the edges. By combining the deep etch with the tapered transition, they created a building block that allows for reliable, multi-section lasers. This success suggests a clear path forward for creating more complex, single-chip sensors that can operate above 2 µm, opening the door for better environmental gas sensors and biomedical tools.