Quantum randomness confirmed on a chip with 10-sigma certainty

Researchers at the Technical University of Denmark and collaborating institutions have demonstrated a quantum random number generator on a chip with a contextuality violation exceeding the classical bound by more than 10 standard deviations, establishing a high level of statistical certainty in its randomness. The system utilizes two integrated silicon photonic chips to generate random numbers from the interference of single photons, certifying their security without relying on entanglement.

This architecture achieves a generation rate of 21.7 ± 0.5 bits per second, with each measurement outcome containing at least 0.077 ± 0.002 bits of extractable genuine randomness. These results suggest a path toward practical, semi-device-independent quantum random number generators compatible with future photonic quantum networks.

Silicon Photonics Platform for Quantum Technologies

Researchers have moved beyond bulky laboratory setups by fabricating the core components of this generator onto two silicon photonic chips, signaling a significant step towards deployable quantum devices for secure communication and advanced computation. This architecture differs from many existing quantum random number generators, which rely on entanglement to prove the randomness of their output; instead, this system leverages a mathematical principle known as the Klyachko-Can-Binicioğlu-Shumovsky (KCBS) contextuality inequality.

The team, comprised of scientists from the Technical University of Denmark, Queen Mary University of London, IHP, Leibniz-Institut für innovative Mikroelektronik, Technische Universität Berlin, and the University of Florence, designed a system that generates randomness from the interference of single photons within a complex optical network. This network is built upon a heralded single-photon source combined with a reconfigurable interferometric mesh, both integrated onto the silicon chips.

The KCBS inequality is then used to test for a specific type of non-classical behavior called contextuality, where the outcome of a measurement depends on the other measurements being performed simultaneously. The researchers explain in their published work that “By placing limits on the correlations allowed in a non-contextual theory, the KCBS inequality provides a minimal and experimentally accessible test of quantum contextuality, suitable to certify QRNGs.”

The team highlights the compactness of the integrated photonic circuits as a key advantage, significantly reducing the physical size of the quantum system. “Our integrated semi–DI QRNG leverages the compactness of photonic integrated circuits, significantly reducing the footprint of the overall quantum system.” Although presented as a demonstration, the work suggests a pathway for building more practical and scalable quantum random number generators, potentially impacting fields ranging from cryptography to scientific simulations. The researchers state that their work “establishes an end-to-end photonic architecture for contextuality-certified randomness generation, integrating photon generation, programmable state manipulation, contextuality verification, and randomness extraction within silicon photonic platforms.”

KCBS Contextuality Inequality for QRNGs

The pursuit of truly random numbers has intensified alongside advances in cryptography and data security, driving innovation in quantum random number generators (QRNGs). Existing QRNG technologies span a spectrum of trust assumptions, from fully trusted devices relying on internal quantum processes to device-independent schemes leveraging the non-local correlations of entangled states. However, the latter often face practical limitations in speed and complexity.

A growing area of interest focuses on semi-device-independent (semi-DI) QRNGs, which strike a balance by relaxing some hardware requirements while still offering quantifiable security assurances. Recent work demonstrates a novel semi-DI QRNG built on silicon photonics, certified by the violation of a specific quantum principle: the KCBS contextuality inequality.

Semi-Device-Independent QRNG Approach

Unlike many existing QRNGs reliant on complex entanglement, this system certifies randomness without it, offering a potentially simpler path toward secure and practical devices. The team’s architecture integrates two silicon photonic chips, a heralded single-photon source and a reconfigurable interferometric mesh, to prepare and measure qutrit states, quantum bits leveraging a three-dimensional state space, suitable for testing a specific inequality. This approach hinges on violating the Klyachko-Can-Binicioğlu-Shumovsky (KCBS) inequality, a mathematical tool used to assess contextuality.

Contextuality, in quantum mechanics, implies that the outcome of a measurement depends not only on the measured property but also on the context of other compatible measurements. This result unambiguously confirms non-classical behavior, moving beyond the limitations of traditional pseudo-random number generators.

Heralded Single-Photon Source & PIC Integration

The demand for provably secure random numbers is escalating, driven by advances in cryptography and growing concerns over data protection, and a new demonstration leverages integrated photonics to address this need with a novel approach to quantum randomness generation. This architecture, detailed in recent work, establishes a complete system for generating and verifying randomness directly on a chip, potentially paving the way for more compact and scalable quantum security solutions. This allows for the preparation and manipulation of qutrit states, quantum bits with three levels, necessary for testing a specific mathematical tool known as the Klyachko-Can-Binicioğlu-Shumovsky (KCBS) inequality.

This is a crucial distinction, as traditional pseudo-random number generators are entirely predictable given enough information. Through a detailed analysis, the researchers certified a conditional min-entropy of 0.077 ± 0.002 bits per experimental round. The use of two integrated silicon photonic chips is a significant step towards miniaturization; traditional quantum experiments often require bulky optical setups, but this integrated approach drastically reduces the overall footprint.

This compactness is not merely a matter of convenience, it’s essential for practical deployment in real-world applications, such as secure communication networks and data encryption. The system’s reliance on contextuality, rather than entanglement, offers a potential simplification in engineering complexity, and could lead to more robust and cost-effective QRNG designs.

Conditional Min-Entropy & Randomness Extraction

The pursuit of truly random numbers often begins with an assumption: that quantum mechanics, with its inherent unpredictability, provides the ultimate source. However, extracting usable randomness from quantum systems isn’t simply a matter of observation; it requires rigorous certification to ensure the numbers aren’t subtly biased or predictable, a challenge addressed by a new integrated photonic system demonstrating a pathway toward practical, secure random number generation.

Researchers have moved beyond relying solely on entanglement to prove randomness, instead leveraging a principle called contextuality, and achieving a remarkably high level of statistical confidence in their results. Unlike many established QRNGs, this architecture doesn’t require entanglement, potentially simplifying engineering challenges and reducing demands on maintaining delicate quantum states. This isn’t merely a marginal improvement; it represents a robust confirmation of genuine randomness, far exceeding the reliability of traditional pseudo-random number generators which, given enough information, are entirely predictable.

The team emphasizes that the present implementation is not a randomness-expansion protocol, clarifying that the system relies on a trusted random seed for measurement settings, a standard practice in certified randomness protocols. The focus, they state, is on demonstrating the feasibility of certification within a semi-device-independent framework, rather than maximizing the rate of random bit generation.

From this violation, they certify a conditional min-entropy per experimental round of H_(min) = 0.077 ± 0.002, derived via a tailored semidefinite-programming-based security analysis. Each measurement outcome therefore contains at least 0.077 ± 0.002 bits of extractable genuine randomness, corresponding to an asymptotic generation rate of 21.7 ± 0.5 bits/s.

Contextuality Violation & Non-Classical Behavior

This level of certainty, far surpassing typical pseudo-random number generators, confirms the genuinely non-classical behavior driving the system’s ability to produce unpredictable outputs. Researchers detailed the architecture, which leverages the principles of quantum contextuality to certify randomness without relying on the creation of entangled particles, a common requirement in many other QRNG designs. The core of this advance lies in the implementation of a Klyachko-Can-Binicioğlu-Shumovsky (KCBS) inequality test.

Unlike schemes demanding fully trusted hardware, this approach operates under semi-device-independent conditions, meaning it requires fewer assumptions about the internal workings of the system to guarantee security. This integration is significant, moving beyond bulky laboratory setups towards more compact and potentially deployable quantum devices. The experimental setup prepares and measures qutrit states, quantum units of information with three possible values, suitable for testing the KCBS inequality.

Single photons are guided through the reconfigurable mesh, where their properties are altered to create the necessary quantum states. Measurements are then performed using superconducting nanowire single-photon detectors and time taggers, recording the arrival of each photon. From this observed contextuality violation, the researchers have certified a conditional min-entropy of 0.077 ± 0.002 bits per experimental round. Each measurement outcome therefore contains at least 0.077 ± 0.002 bits of extractable genuine randomness, corresponding to an asymptotic generation rate of 21.7 ± 0.5 bits/s, demonstrating a practical speed for generating truly random numbers.