Quantum Repeaters

In the entanglement-distribution lesson we saw that photon survival falls exponentially with distance. A quantum repeater defeats that exponential by chaining together the two primitives you have already met — entanglement swapping and purification — so that the cost grows only polynomially with distance. The repeater is the single idea that turns intercontinental quantum links from impossible into merely hard.

The exponential wall

Distribute a Bell pair directly over a fibre of length $L$ . The success probability per attempt is

p \sim 10^{-\alpha L / 10},

so the expected number of attempts — and the time to establish one pair — grows as $1/p$ , exponentially in $L$ . Doubling the distance squares the difficulty. No amount of patience makes $1000\ \text{km}$ practical this way.

Divide and conquer

Split the total span into $N$ shorter elementary links, each of length $L/N$ , separated by repeater stations. Over a short link, distributing a pair is cheap because $p$ is much larger. Once neighbouring links each hold a Bell pair, a repeater performs an entanglement swap, fusing two short-range pairs into one pair spanning twice the distance. Repeating the swap in a nested, tree-like fashion stitches $N$ short pairs into one end-to-end pair:

\underbrace{A\!-\!R_1}_{\Phi^+}\ \underbrace{R_1\!-\!R_2}_{\Phi^+}\ \cdots\ \underbrace{R_{N-1}\!-\!B}_{\Phi^+} \;\xrightarrow{\text{swaps}}\; \underbrace{A\!-\!B}_{\Phi^+}.

The chain uses $N \propto L$ elementary links arranged in $\log_2 N$ nesting levels. Crucially, with nested swapping the total time and resources to establish an end-to-end pair grow only polynomially in $L$ (polynomial in $L$ , with the overhead scaling polylogarithmically), instead of the exponential cost of direct transmission.

Why purification is mandatory

Swapping does not come for free. Each Bell-state measurement combines the imperfections of two input pairs, so fidelity drops with every swap. Without intervention, after a long chain of swaps the end-to-end pair would be too noisy to use. The repeater therefore interleaves entanglement purification: several low-fidelity pairs on a link are distilled into fewer higher-fidelity ones before the next swap. The Briegel–Dür–Cirac–Zoller protocol alternates swap and purify at each nesting level so that fidelity is restored faster than swapping erodes it.

Memories make it efficient

A subtle requirement is quantum memory at each repeater. Elementary links succeed probabilistically and at different moments; a node must hold a successfully distributed pair while it waits for the neighbouring link to also succeed, so the two can be swapped. Without memory, every link in the chain would have to succeed in the same instant — a coincidence whose probability is again exponentially small. The quality of repeater-based networks is largely set by how long these memories can store a qubit (the coherence time) relative to the classical signalling time between nodes.

Generations of repeaters

Researchers classify repeaters into three generations by how they fight loss and operational errors:

First generation — heralded entanglement generation plus purification; relies on two-way classical communication and memories.
Second generation — quantum error-correcting codes suppress operational errors, reducing the classical back-and-forth.
Third generation — codes handle both loss and operational errors, enabling one-way communication with no waiting for acknowledgements.

Each generation trades more on-node quantum processing for less dependence on slow classical round-trips, pushing achievable rates higher.

Takeaways

A quantum repeater turns an exponential problem into a polynomial one by segmenting the link, swapping to extend entanglement, and purifying to keep fidelity high — all backed by quantum memories that buffer probabilistic successes. The next two lessons make the two atomic moves concrete: first a distributed gate built from teleportation, then the purification step that keeps the whole chain alive.