Demonstrates a fully differential photonic–electronic architecture that measures conjugate quadratures on chip and extracts high-entropy quantum randomness in real time.
Quantum random number generators (QRNGs) do exactly what their name suggests: they use quantum physics to produce truly random numbers. Random numbers are essential for modern life — they secure our online banking, protect private messages, enable fair simulations, and support scientific computing. The problem is that most “random” numbers used today come from algorithms. These are fast and practical, but because they follow a set of rules, they are never perfectly unpredictable. QRNGs solve this by turning to the fundamental randomness of nature itself. By measuring tiny quantum fluctuations, effects that are inherently unknowable in advance, they can generate numbers that are genuinely random. In this current iteration QRNG, Christian Carver leads the effort to demonstrates a compact and scalable method that moves this powerful technology closer to real-world deployment.
At the heart of the system is a small optical circuit on a chip. A stable laser acts as a reference light, while an “empty” quantum state — the vacuum — provides the randomness. These two signals are combined in a 90° optical hybrid, which splits and mixes the light into four paths that carry complementary pieces of information about the quantum noise. Each path is detected by photodiodes that convert light into electrical signals. Because technical noise from the laser can easily overwhelm the fragile quantum signal, the electronics are designed to subtract matching signals from one another. This differential measurement cancels shared noise while keeping the quantum fluctuations. The result is a clean electrical signal whose variations come directly from quantum randomness. That signal is then digitized and sent to an FPGA, where some real-time processing removes any remaining classical noise and converts the data into a stream of uniform random bits which can be used for cybersecurity.
One of the key advances in this work is that the chip measures two complementary quantum observables at the same time rather than just one. Combining them produces a phase-independent quantity that is directly linked to photon number, which makes the system more stable and increases the amount of usable randomness extracted from each measurement. Careful optical balancing on the chip, together with a fully differential transimpedance amplifier, strongly suppresses unwanted noise. This leads to a high common-mode rejection of 69 dB and a shot-noise clearance of 25.6 dB with clear signatures that the measured signal is dominated by quantum effects rather than classical interference.
In its current demonstration, the system generates random numbers at 8 Mbps using a 1-MSPS analog-to-digital converter — already fast enough to keep up with most practical needs — and the measured detector bandwidth suggests that, with quicker electronics, it could sprint toward 800 Mbps. Just as importantly, the output sails through the full NIST statistical test suite, meaning the numbers behave exactly like perfect randomness, with no hidden patterns waiting to be discovered.
The bigger picture here is pretty exciting. By putting a QRNG on a photonic chip, we make the whole thing smaller, cheaper, more energy‑efficient, and much easier to use outside a carefully controlled lab. That means truly random numbers could show up in real‑world tech, helping keep our data safer, powering portable quantum communication systems, and laying the groundwork for future quantum networks. And it’s not just about randomness: the same setup can be used for other quantum measurements and sensing tasks too. So rather than being just a faster random‑number machine, this chip is more like a versatile little engine for practical quantum technologies.