Research, Explained

Almost every optical system eventually needs to make light stronger — to push a signal down a fiber, read out a faint sensor, or drive the next stage of a circuit. On a chip, that has always been the hard part. Compact optical amplifiers are notoriously power-hungry, and the efficient ones are bulky lab instruments.

We built an amplifier that breaks that trade-off. Instead of brute-forcing the gain, our device recycles energy through a resonant design: it traps the "pump" light and its second harmonic in the same tiny structure so the same photons do useful work over and over. The payoff is dramatic — the chip intensifies light roughly 100× (>17 dB of gain) on only a few hundred milliwatts of power, an order-of-magnitude leap over previous integrated amplifiers, while adding very little noise and working across a broad slice of the spectrum.

"Low power" is what makes it matter. An amplifier this efficient could run on a battery, which opens the door to optical technology inside everyday devices — biosensors, portable diagnostics, data-communication links, and compact light sources — rather than only on an optical table.

Even a "perfect" laser carries an irreducible fuzziness called quantum noise — the fundamental flicker that sets the floor on how precisely we can measure with light. "Squeezed light" is a special quantum state that pushes that noise below the standard limit in one variable (by paying the price in another), and it's the secret ingredient behind record-breaking sensors, including the gravitational-wave detectors at LIGO.

The catch: squeezed light has traditionally needed an entire optical table of mirrors and crystals. In this work I demonstrated a squeezed-light source built entirely on a thin-film lithium niobate chip, using an on-chip optical parametric oscillator. Critically, I also performed full quantum-state tomography on-platform — reconstructing the complete quantum portrait of the light to prove it was genuinely squeezed, not just quiet by accident.

This is the kind of result that connects two worlds: the fragile, table-sized physics of quantum optics and the manufacturable, scalable reality of integrated photonics. A chip-scale squeezer is a building block for quantum sensing, quantum-enhanced measurement, and continuous-variable quantum information.

Short-wavelength visible light — blue in particular — is in constant demand: for quantum computers that control trapped atoms, for high-density data storage, and for biology and displays. One powerful way to make it is second-harmonic generation, in which infrared light is sent into a nonlinear material that fuses pairs of photons into single photons of double the energy (and half the wavelength), converting invisible infrared into bright visible light.

Doing this efficiently at short wavelengths on a chip is genuinely hard — the materials are fussy and the engineering tolerances are tight. By carefully designing and poling thin-film lithium niobate waveguides, I demonstrated blue-light generation at 456 nm with a normalized conversion efficiency of ~33,000% / W·cm², among the highest reported on this platform at the time.

Beyond the headline number, this work built up fabrication and design know-how — how to make these tiny structures actually perform at short, demanding wavelengths.