Rare-Earth Ions Just Got a Quantum Internet Speaking Fiber's Own Language

Quantum Technology · Quantum Networking

The internet works because a signal can be copied, boosted, and passed along a chain of routers without anyone caring what happens in between.

A quantum internet can't cheat that way — and Princeton physicists just took a real step toward building one that speaks the same language as the fiber-optic cables already under our streets.

Graduate students with Professor Jeff Thompson in the Princeton lab where the erbium-ion quantum network devices are built and tested. Credit: Sameer A. Khan/Fotobuddy, Princeton Engineering.

A Qubit Built to Speak the Internet's Own Language

Building a quantum computer is only half of a much bigger ambition. The other half is connecting quantum computers, quantum memories, and quantum sensors together into a genuine quantum network, one that could enable unhackable communication and let distant quantum processors work as a single, larger machine.

A team at Princeton University, led by first author Haitong Xu, has now demonstrated something that earlier quantum network prototypes struggled to pull off cleanly: precise, individual control over two interacting erbium ions and a nearby atomic nucleus, all packed inside the same tiny crystal, using a platform that emits light at exactly the wavelength modern fiber-optic networks already use.

The result, published in Nature Physics, doesn't build a quantum network on its own. What it does is demonstrate the fundamental building block — a working, addressable two-qubit system — using a material platform that could plausibly scale into something much larger.

↑ Back to Table of Contents

Why Quantum Computers Need a Quantum Internet

Quantum technologies exploit the strange rules of quantum mechanics to do things ordinary computers and sensors can't. Central to all of it is the qubit, or quantum bit, which, unlike a classical bit, isn't locked into a simple 0 or 1 — it can exist in a superposition of both at once.

Quantum computers, quantum memories, and quantum sensors are all built around manipulating qubits in one location. But many of the most exciting applications people imagine for quantum technology — provably secure communication, distributed quantum computing, ultra-precise networked sensing — require connecting multiple quantum devices, or nodes, that might be sitting in different labs, different buildings, or different cities entirely.

That connective layer is what researchers mean by a quantum network, or eventually, a quantum internet. And building one turns out to be a fundamentally harder problem than simply scaling up a single quantum computer.

↑ Back to Table of Contents

The Photon Problem: Getting Quantum Information to Travel

One well-established approach to linking distant quantum nodes relies on atomic defects embedded in crystals — tiny imperfections that can be coaxed into emitting single photons. Because photons naturally travel through fiber-optic cable, they're a natural carrier for quantum information between distant locations.

But a photon alone isn't a network. Each node in the network also has to be able to receive that quantum information, store it reliably, and process it locally before passing it along or acting on it. That requires a second capability entirely: precise control over multiple nearby quantum spins within the same node, not just the ability to emit a photon.

Getting both halves working together, in the same physical system, is where a lot of earlier quantum networking research has run into trouble.

↑ Back to Table of Contents

Why Rare-Earth Ions Are Different

The Princeton team's approach centers on erbium (Er³⁺) ions, a rare-earth element with a long history in optical technology — erbium-doped fiber amplifiers already boost signals across much of the world's undersea internet cabling today.

What makes erbium especially attractive for quantum networking is that its optical transition sits directly in the telecommunications band, around 1.5 micrometers, the same low-loss wavelength that carries ordinary internet traffic through fiber. Most other leading quantum-defect platforms emit in the visible spectrum, which degrades quickly over the long fiber runs that need to be converted to telecom wavelengths before they can travel far, adding complexity and loss at every conversion step.

A single erbium ion doesn't emit light strongly on its own, though, which is part of why this class of ion took longer to develop than some alternatives. Thompson's group, which led this research, previously solved that problem by coupling individual erbium ions to a nanoscopic silicon photonic cavity, boosting how efficiently they emit photons by roughly a thousandfold, and by identifying calcium tungstate, after testing hundreds of candidate host crystals, as an unusually clean, low-noise environment for hosting the ions.

↑ Back to Table of Contents

Two Ions, Nanometers Apart

In this new work, the team went a step further than emitting single photons from a single ion. They identified two Er³⁺ ions sitting just a few nanometers apart inside the crystal, close enough that their electron spins interact with each other magnetically, and set out to control both of them individually and use that interaction as a computational resource.

Using laser pulses, the researchers coherently controlled both the optical transitions and the electron spins of the two interacting ions. That's a nontrivial technical feat on its own: at nanometer separations, addressing one ion without accidentally disturbing its neighbor is exactly the kind of challenge that has limited how far earlier solid-state qubit registers could be pushed.

A diagram of the nanoscopic silicon waveguide used to capture and guide photons emitted by erbium ions in the Princeton team's devices. Credit: Image courtesy of the researchers, Princeton Engineering.

↑ Back to Table of Contents

Telling Identical Atoms Apart

Individually addressing two atoms of the exact same element, sitting only nanometers apart, sounds like it should be impossible with a laser beam whose spot size is fundamentally limited by the diffraction of light — and by ordinary optical standards, it would be.

The workaround relies on a property rare-earth ions have that many other qubit platforms lack: extraordinarily narrow spectral linewidths. Even though the two erbium ions are chemically identical, tiny differences in their immediate local environment inside the crystal shift their optical and spin resonance frequencies by just a small amount — enough that the researchers could distinguish and individually control each ion using its own characteristic frequency, rather than relying on physically resolving their positions in space.

The team confirmed they'd found a genuinely interacting pair using a spectroscopy technique called double electron-electron resonance, which detects the magnetic coupling between nearby spins before any gate operations are attempted.

↑ Back to Table of Contents

Turning Two Qubits Into a Gate

Identifying an interacting pair of ions is one thing; turning that interaction into a usable quantum logic gate is another. The team used dynamical-decoupling sequences, carefully timed pulse patterns that protect a qubit from environmental noise, to convert the ions' magnetic dipolar interaction into a functioning two-qubit gate between their electron spins.

As Xu put it, describing the result directly: "We realized two-qubit gates between the electron spins." That gate let the researchers use one erbium ion's spin to perform repeated, non-destructive measurements of its neighbor's spin state — a capability called quantum nondemolition measurement, useful because it lets you check a qubit's state without collapsing the very quantum information you're trying to preserve.

To construct these gates with high fidelity, the team relied on a numerical optimal-control method known as GRAPE, which computationally searches for the precise pulse sequences that produce the cleanest possible quantum operations given the specific noise characteristics of their system.

↑ Back to Table of Contents

Borrowing a Nucleus for Memory

Electron spins are useful for fast control and interaction, but they tend to lose their quantum coherence relatively quickly. Nuclear spins, by contrast, are far more isolated from environmental noise and can hold quantum information for much longer, which makes them attractive as a memory element — if you can find a way to write to and read from them reliably.

The Princeton team used a nuclear spin located near the interacting ion pair as exactly that: an ancillary memory qubit. Using the same optimal-control approach, they built electron-nuclear gates that let them store quantum information from an electron spin into the nuclear spin, and later retrieve it.

Critically, the team showed that information stored in the nuclear memory survived repeated optical excitation and measurement of the electron spin sitting nearby — meaning the memory qubit didn't get scrambled just because its neighboring electron spin kept being probed. That combination hadn't previously been demonstrated in a rare-earth ion system, and it's an essential capability for more advanced network operations like entanglement swapping and distillation, the processes that let a quantum repeater extend entanglement across many linked nodes.

↑ Back to Table of Contents

Why the Telecom Band Matters So Much

It's worth pausing on why the telecom-band detail is such a big deal, rather than a minor engineering footnote. The world already has an enormous, continent-spanning investment in fiber-optic infrastructure tuned specifically to transmit light efficiently around the 1.5-micrometer wavelength.

A quantum network built on a platform that naturally emits at that same wavelength could, at least in principle, piggyback on existing fiber networks rather than requiring a dedicated new physical layer built from scratch, or expensive, lossy frequency-conversion hardware bolted onto every node.

The same Princeton group has already shown this isn't purely theoretical: in earlier work, they demonstrated spin-photon entanglement using a single Er³⁺ ion, sending an entangled photon out over 15 kilometers of deployed optical fiber and confirming the entanglement survived the trip. That's a modest distance compared with some other repeater technologies' demonstrations, but it was achieved without any of the added complexity of frequency conversion that those other approaches require.

↑ Back to Table of Contents

From Two Qubits to a Real Quantum Network

The system demonstrated here is explicitly a proof of concept: two electron spins and one ancillary nuclear qubit, not yet the dense, multi-qubit register a practical network node would eventually need.

The team's near-term priorities are concrete rather than speculative. Gate fidelities in the current system are primarily limited by how quickly the electron spins lose coherence, so identifying and reducing the microscopic sources of that decoherence is a stated next step. In parallel, the researchers plan to create denser, spatially patterned clusters of erbium ions, scaling from a pair of interacting spins toward larger registers capable of supporting more complex operations.

Combined with the group's earlier telecom-band spin-photon entanglement work, the eventual goal is a multiplexed quantum repeater node — a device capable of creating, storing, and managing several quantum connections simultaneously, which would meaningfully improve the reach and reliability of long-distance quantum communication.

↑ Back to Table of Contents

Manohar Insight

What strikes me most about this result is how much of it is really about patience with materials science, dressed up as a physics breakthrough. Getting to two individually controllable, interacting erbium ions took years of narrowing hundreds of thousands of candidate crystals down to one, and boosting a naturally weak photon emitter a thousandfold with nanophotonic engineering, before anyone could even attempt the quantum control demonstrated in this paper.

That's a useful reminder for how progress in quantum networking actually happens. It rarely arrives as a single dramatic leap; it accumulates through unglamorous, sequential fixes to noise, material purity, and photon collection efficiency, each of which quietly raises the ceiling on what's experimentally possible.

The choice to build all of this around the telecom band, rather than a more exotic wavelength that performs better in the lab but worse in the real world, also says something about where this field's ambitions actually sit. It's a bet that the fastest path to a genuinely useful quantum internet runs through compatibility with the fiber network we've already spent decades building, not around it.