Physicists Build Quantum Entanglement by Working Around Photon Loss, Not Preventing It

Photons travel at the speed of light and interact weakly with their environment - qualities that make them attractive carriers for quantum information. They also disappear. In any real optical system, photons are lost with high probability before they can be detected. For anyone trying to build a large entangled quantum state from many photons, this is not a minor inconvenience. A single missing photon breaks the entire state, and attempting to detect the missing ones destroys everything already built.

This loss problem has kept photonic graph states - entangled structures used in quantum computing, secure communication, and quantum sensing - largely in the realm of theoretical demonstration rather than practical implementation. A team at the University of Illinois Urbana-Champaign proposes a protocol that addresses the problem by changing what counts as a setback.

Building After Confirmation, Not Before

The conventional approach to building photonic graph states tries to minimize loss and detect as few photons as possible until the state is complete. The Illinois team, led by Associate Professor of Physics Elizabeth Goldschmidt and Professor of Electrical and Computer Engineering Eric Chitambar, inverted this logic. Their protocol, published in npj Quantum Information, adds a photon to the growing graph state only after confirming that photon's existence through detection.

Detecting a photon normally destroys it - you cannot detect a photon and then incorporate it into a quantum state. The team worked around this by introducing a framework called virtual graph states that makes destructive measurement acceptable across a broad class of applications. By confirming a photon exists before incorporating it into the state structure, the protocol shifts the limiting bottleneck. Instead of being constrained by photon loss rates - which can be very high and are difficult to control - performance is now bounded by the coherence time of the spin qubits used to emit photons, which can be orders of magnitude longer.

"We changed our thinking from what would be the most useful end result to what can we do with the resources we already have," said Goldschmidt. "It took us a long time to realize that destructively measuring the photons would be okay for this wide set of useful circumstances."

What Virtual Graph States Actually Are

In a standard graph state, all qubits exist simultaneously in an entangled structure. In a virtual graph state, the individual qubits - encoded in photons - do not all exist at the same time. Some have already been measured and destroyed. What persists is the pattern of correlations between them, mediated by the spin qubits that emitted each photon in sequence.

"There's something almost counterintuitive about it," said Max Gold, a graduate student and co-lead author. "We are building up these correlations that can only be described by quantum systems across different photons. We have photons that do not ever exist at the same time in nature, and something mediating their interactions that is not the photons themselves. Even though we talk about it as a single state, not all the qubits in the state exist at one time."

The correlations this produces are real and usable. They can support the same quantum protocols that conventional graph states enable - secure computation, quantum key distribution, measurement-based quantum computing - as long as the application fits within what virtual graph states can provide.

Compatible with Hardware That Already Exists

The practical significance of the protocol is that it works on existing experimental platforms. Trapped ions, neutral atoms, and other emitter-based systems typically have low photon collection efficiencies - meaning most of their photons are lost before detection. That is exactly the scenario the emit-then-add approach handles.

"Our method is feasible in practice even for emitter-based platforms with traditionally low photon collection efficiencies such as trapped ions and neutral atoms," said Jianlong Lin, a graduate student and co-lead author. "It would be one of very few demonstrations of photonic graph states with practical uses."

The team proposes a specific near-term application: secure two-party computation based on repeatedly generating small graph states. This is achievable with current hardware. Other potential applications including measurement-based quantum computing and quantum sensing require more development, but the protocol provides a foundation for approaching them.

An Honest Assessment of Scope

The full version of the protocol - applicable to the broadest possible range of uses - requires non-destructive measurement of photons. That capability is not yet available in any existing system. The paper focuses its main contribution on what can be achieved with destructively measured photons, which is achievable now.

Goldschmidt is moving the work from theory toward experiment. Lin is beginning early-stage experimental work on the approach, while Gold is exploring additional theoretical applications of the virtual graph state framework. The researchers hope the protocol encourages the broader field to take hardware constraints seriously rather than building theories that assume losses away.

"A lot of the literature has ignored hardware limitations, and I hope this work encourages other people to think about what could be produced given the real constraints of real near-term hardware," Goldschmidt said.