Photonic Qubits
Encoding a qubit in light, and why that's the natural platform for a quantum internet
By the end of this topic you'll be able to
Photonic qubits encode quantum information in properties of individual photons — commonly polarization (horizontal versus vertical), the photon's arrival time-bin (early versus late), or which of two possible paths it traveled.
Photons barely interact with their surrounding environment, or with each other, while in transit, which gives photonic qubits excellent coherence essentially for free as they travel — making this the natural platform for quantum communication, and for transmitting entanglement over fiber-optic cable in a future quantum internet.
This is exactly the same property that made photons the obvious choice for BB84 quantum key distribution: a photon's polarization can carry a qubit down a fiber-optic line for kilometers with comparatively little disturbance, which is what lets an eavesdropper's interference actually show up as a detectable increase in errors rather than being lost in the noise of the channel itself.
The same property that makes photons easy to transmit makes them hard to compute with: implementing a genuine two-qubit gate requires two photons to interact, and photons do not naturally interact with one another at all — unlike, say, two trapped ions, which couple through their mutual Coulomb repulsion.
Linear-optical approaches to photonic quantum computing get around this using measurement and probabilistic gate schemes rather than any direct photon-photon coupling: a two-qubit gate is implemented by interfering photons at a beam splitter and post-selecting on specific measurement outcomes, which succeeds only some of the time and has to be repeated until it does.
This probabilistic character is the central engineering challenge specific to photonics: a computation needs many such gates to succeed in sequence, and the overall success probability multiplies down fast unless the scheme is engineered carefully — an active area of photonic hardware research, alongside integrating the whole setup onto photonic chips rather than tabletops of mirrors and lenses.
Try It Yourself
A basic linear-optic two-qubit gate succeeds with probability q = 1/9 per attempt, and fails otherwise (a commonly cited figure for the simplest post-selected schemes). If a circuit needs 3 such gates to all succeed in a single pass, with no retries, what's the overall success probability?
- 1Each gate succeeds independently with probability q = 1/9.
- 2For all 3 gates to succeed together in one pass, multiply the individual probabilities: (1/9)³ = 1/729.
- 31/729 ≈ 0.00137, or about 0.14%.
About 0.14% — illustrating why real photonic architectures can't just hope a whole circuit succeeds in one shot; they need a way to retry failed gates cheaply, which is exactly what the derivation below addresses.
Reference
| Photonic qubit encodings | Properties of a single photon used to store a qubit | |
| Advantage | Excellent coherence over long distances, e.g. fiber-optic cable | |
| Challenge | Two-qubit gates need probabilistic, measurement-based schemes |
Quick Check
Why are photonic qubits particularly well-suited to quantum communication and networking?
Why is it hard to build a genuine two-qubit gate between two photons?