A glass chip written with a femtosecond laser has generated secure random bits at 42.7 Gbit/s, a record for source-device-independent quantum random number generation. Researchers at the University of Padua, Politecnico di Milano, and the CNR Institute for Photonics and Nanotechnologies report the work in Advanced Photonics, and the material of choice is ordinary borosilicate glass rather than silicon.

That choice is the story. Continuous-variable quantum communication, the family that includes quantum key distribution and quantum random number generation, depends on measuring the amplitude and phase of light. A coherent receiver mixes a weak quantum signal with a stronger reference beam and reads how the two interfere. Silicon photonics does this today, but silicon carries penalties: it is sensitive to polarization and tends to lose more light on the way through.

Glass behaves better. It is naturally insensitive to polarization, stable over time, and allows waveguides to be written in three dimensions with minimal loss. The femtosecond laser writes light-guiding paths directly inside the material, forming compact photonic circuits without a semiconductor fab in sight. That last point matters commercially. Rapid prototyping without expensive fabrication changes the cost curve for a field still dominated by custom lab hardware.

What is on the chip?

The team built a fully tunable heterodyne receiver, the core component for CV-QKD and CV-QRNG, by writing the optical circuit inside the glass. The device includes:

• Fixed and tunable beam splitters
• Thermo-optic phase shifters for electrical control
• Three-dimensional waveguide crossings
• Polarization-independent directional couplers

These elements let the quantum signal and the reference beam interact under control, enabling simultaneous measurement of two conjugate quadratures. The numbers are respectable by any platform's standard. Insertion loss sits at roughly 1 dB. The common-mode rejection ratio exceeds 73 dB, which indicates strong suppression of classical noise. Signal-to-noise performance stayed stable over at least eight hours of operation. The authors state these results match or exceed many silicon-based photonic receivers.

One chip, two jobs

The receiver's combination of low loss, tunability, and stability let it run two distinct quantum tasks without hardware changes. As a heterodyne detector, it supported a QRNG system that remains secure even if the incoming optical signal cannot be trusted, a property the field calls source-device-independence. The 42.7 Gbit/s secure generation rate is the record figure for this class of system.

The same chip then ran a QPSK-based CV-QKD protocol, where information rides on a four-point constellation of quantum states. Over a simulated 9.3-kilometer fiber link, the system reached a 3.2 Mbit/s secret key rate. The contrast between the two numbers is instructive. Randomness pours out at gigabit scale; secret keys trickle at megabit scale over metropolitan distances. Both figures matter, but they serve different layers of a quantum network.

Why glass, commercially speaking?

The study lists practical advantages that read like a deployment checklist:

• Environmental stability: glass is inert and resists temperature and mechanical change
• Low-loss fiber coupling: waveguide sizes closely match standard telecom fiber
• 3D design freedom: crossings and complex layouts without added loss
• Scalability: laser writing allows rapid prototyping without expensive semiconductor fabrication

The researchers note the platform could suit harsh environments, including potential use in space-based quantum communication systems. For network operators weighing quantum upgrades against existing fiber plant, the compatibility with standard telecom fiber is the detail worth underlining.

Where does this leave silicon?

The work does not retire silicon photonics, which retains its integration advantages. It does demonstrate that a glass platform can serve as the photonic front end for advanced CV-QKD without the polarization sensitivity and optical loss that constrain silicon implementations. The authors position glass-based integrated photonics as a route to close the gap between experimental setups and practical quantum networks.

The researchers suggest this approach could move quantum communication from controlled lab settings into real-world infrastructure, and toward global quantum networks. A record randomness rate on a chip that needs no fab is a reasonable opening move.