📱 Try our Android app
Caltech brings fiber‑optic level loss to glass photonics on silicon chips
AUGUST 20, 2026Science

Caltech brings fiber‑optic level loss to glass photonics on silicon chips

New germano-silicate waveguides printed on standard wafers deliver fiber-like ultralow loss and outperform silicon nitride, especially at visible wavelengths.

August 20, 20262 min read1495 tags

Optical fiber underpins modern communication because its glass is extremely pure and its surface is ultra-smooth, so light propagates with ultralow loss. A Caltech team has now brought this fiber-like performance directly onto silicon wafers, opening the door to far more efficient integrated photonics.

Fiber glass directly on chip

In Kerry Vahala’s lab, researchers developed a way to fabricate on-chip waveguides from germano-silicate glass – the same glass used in telecom fiber. Instead of pulling glass into long fibers on a spool, they lithographically pattern it onto standard 8- and 12-inch semiconductor wafers.

The waveguides are laid out as tight spirals rather than straight lines. This gives light a very long optical path length while occupying only a tiny chip area, analogous to winding fiber on a spool but shrunk to the nanoscale.

Because the material is glass, these germano-silicate waveguides exhibit extremely low propagation loss and can be efficiently coupled to both optical fibers and semiconductor lasers – a key requirement for reducing the energy cost of large-scale server and data-center infrastructure.

Beating silicon nitride, especially in the visible

At near-infrared wavelengths, devices made on the new Caltech platform already match the performance of state-of-the-art silicon nitride photonic circuits, which are widely used today for low-loss on-chip optics.

In the visible spectrum, the advantage becomes dramatic: the germano-silicate platform surpasses silicon nitride by roughly a 20× reduction in loss. Combining such low loss with wafer-scale fabrication could enable new generations of silicon photonics, from energy-efficient data links to advanced optical and quantum information processing systems.


Source: ScienceDaily