eduKate Learning Manual
Science | Edge Cases Science | Physical World
Understand → Observe → Explain → Test → Transfer → Go Deeper

Landau–Zener–Stückelberg Interference

How Crossing the Same Avoided Crossing Twice Turns It Into an Interferometer

Quick Answer

Near a two-level avoided crossing, sweeping the detuning through zero at finite speed creates a Landau–Zener transition probability

P_LZ ≈ exp(−2πδ)

in a common convention, where δ depends on the square of the gap divided by sweep rate.

The first passage creates a coherent superposition of two paths through energy-state space. Between passages, the two components accumulate a relative dynamical phase. The second passage mixes them again. Final population then contains an interference term that oscillates with this accumulated phase.

Changing sweep amplitude, speed, hold time or drive frequency moves the phase and produces Stückelberg fringes.

The simple picture assumes a coherent two-level system and passages separated enough to treat each crossing as a Landau–Zener event. Strong dephasing, overlapping crossings, multilevel structure and arbitrary drive waveforms require more complete models.

Physics Reports — Nonadiabatic Landau–Zener–Stückelberg–Majorana Dynamics and Interference →

Physical Review Letters (13 August 2026) — Landau–Zener Tunnelling and Quantum Interference of Andreev States →

What You Will Learn

• What an avoided crossing is.
• Why slow and fast sweeps produce different transition probabilities.
• How one Landau–Zener passage acts like an amplitude beam splitter.
• What phase accumulates between passages.
• Why the second passage converts phase into population fringes.
• What Stückelberg oscillations measure.
• Why coherence is required.
• How dephasing washes fringes out.
• How geometric phase can modify interference in suitable paths.
• Why LZS interference differs from Bloch oscillations.
• Why it differs from Berry phase alone.
• Where the adiabatic-impulse two-level picture fails.

Part 1 — Start With an Avoided Crossing

Two uncoupled levels can cross as a control parameter changes.

Add coupling Δ and the crossing opens into an anticrossing: the instantaneous eigenvalues repel and remain separated by a minimum gap.

A standard two-level Hamiltonian is

H(t) = ½ε(t)σ_z + ½Δσ_x.

ε(t) is the tunable detuning and Δ sets the minimum gap.

Part 2 — Slow Sweep: Adiabatic Following

If ε changes very slowly compared with the gap-controlled timescale, a system prepared in one instantaneous eigenstate follows that branch through the avoided crossing.

In the diabatic basis this can look like a transfer from one bare state to the other.

The system has time to adjust to the changing Hamiltonian.

Part 3 — Fast Sweep: Diabatic Passage

If the sweep is extremely fast, the state has little time to follow the changing eigenvectors.

It tends to remain in the same diabatic state across the crossing.

Between these extremes, finite probability flows into both branches.

Part 4 — Landau–Zener Probability

For a linear sweep ε = vt through an isolated two-level anticrossing, Landau–Zener theory gives an exponentially controlled transition probability.

Faster sweep favours diabatic passage; larger gap favours adiabatic following.

The precise definition of P_LZ depends on which basis and transition is being counted, so every formula should state its convention.

Part 5 — First Passage as a Beam Splitter

In the intermediate regime, the first passage does not produce a classical choice of path.

It produces a coherent superposition of amplitudes on two branches.

This is directly analogous to a beam splitter creating two optical paths, except the paths are states in energy/Hilbert space rather than two physical arms in space.

Part 6 — Phase Accumulation Between Crossings

After the first splitting, the two branches generally have different instantaneous energies.

They accumulate a relative phase approximately

φ_dyn = (1/ℏ)∫[E₊(t) − E₋(t)]dt.

Each Landau–Zener passage also contributes an additional Stokes phase associated with nonadiabatic scattering at the crossing.

The total interference phase combines these contributions and, in some protocols, geometric phase as well.

Part 7 — Second Passage as Recombiner

Reverse the sweep or drive periodically so the system encounters the avoided crossing again.

The second Landau–Zener event mixes the two amplitudes.

If their relative phase is constructive for one final state, that population increases. Shift the phase by π and the same final state can be suppressed.

the second crossing translates invisible phase history into measurable population.

Part 8 — Stückelberg Fringes

Scan a control parameter that changes the phase between crossings:

• drive amplitude;
• drive frequency;
• maximum detuning;
• hold time;
• gap size.

Final-state probability oscillates, creating interference fringes in parameter space.

These fringes are a map of accumulated quantum phase, not merely a sequence of independent transition probabilities.

Part 9 — Why Coherence Matters

The two amplitudes must retain a stable relative phase from the first passage to the second.

Environmental dephasing, ensemble inhomogeneity and control noise randomize that phase.

As coherence decays, Stückelberg fringes lose contrast and the dynamics approaches an incoherent sum of transition probabilities.

Part 10 — LZS as Coherence Spectroscopy

Because fringe visibility depends on phase memory, LZS interferometry can measure more than transition probability.

It can reveal:

• energy gaps;
• dephasing times;
• drive calibration;
• dispersion relations;
• phase shifts from interactions or geometry.

The avoided crossing becomes both a control element and a diagnostic sensor.

Part 11 — Geometric Phase Can Enter

If the two branches trace paths through a multidimensional control space, the relative phase can include Berry/geometric contributions in addition to dynamical and Stokes phases.

That does not make LZS and Berry phase the same phenomenon.

Berry phase is one possible phase contribution; LZS interference is the beam-split/recombine protocol that converts total relative phase into population fringes.

Part 12 — LZS vs Bloch Oscillations

Bloch oscillations arise because constant force carries quasimomentum through a periodic energy band, reversing group velocity.

LZS interference arises because repeated nonadiabatic crossings split and recombine quantum amplitudes.

In multiband lattices under driving, both can coexist. The observed signature—periodic real-space motion versus phase-sensitive interband population fringes—determines which mechanism owns the explanation.

Part 13 — Multiple Passages

Periodic driving can cross the avoided crossing many times.

Then the system resembles a multi-pass interferometer. Amplitudes from many histories accumulate and interfere, producing complex resonance patterns often described using Floquet theory.

The two-passage Stückelberg formula remains the conceptual foundation but may be quantitatively insufficient.

Part 14 — Multilevel Systems

Real atoms, molecules, quantum dots and superconducting circuits can contain several nearby avoided crossings.

Amplitude can split into more than two branches, and crossings can overlap.

Then pairwise independent Landau–Zener events can fail. One needs multilevel numerical propagation or more advanced analytic methods.

Part 15 — 2026: Andreev-State Interference

In August 2026, Physical Review Letters reported Landau–Zener tunnelling and quantum interference involving Andreev states in superconducting junctions.

The work demonstrates how LZSM interference continues to operate as a phase-sensitive probe in modern mesoscopic quantum systems rather than being confined to textbook two-level atoms.

Failed Model → Better Model

Naive model	Why it fails	Better model
Each avoided crossing is an independent probabilistic jump.	Coherent amplitudes from successive passages retain phase.	Add amplitudes and phase before probabilities.
Final population depends only on PLZ.	Relative dynamical and Stokes phases control recombination.	Use interferometer picture.
Any oscillation under periodic drive is LZS.	Rabi, Bloch and Floquet resonances can produce oscillations too.	Identify repeated avoided crossings and phase-sensitive fringes.
Two-level formula works for every driven spectrum.	Overlapping crossings, dephasing and multilevel coupling can dominate.	Use full time-dependent Hamiltonian when needed.

How Do We Know?

• Calibrate the avoided-crossing gap independently.
• Measure single-pass transition probability versus sweep rate.
• Run a controlled double-passage protocol.
• Scan the time or detuning between passages.
• Look for oscillatory final populations rather than simple monotonic probability.
• Deliberately dephase the system and test fringe-contrast loss.
• Reverse sweep direction and compare Stokes-phase conventions.
• Vary drive waveform to test the adiabatic-impulse approximation.
• Include additional levels in simulations when nearby crossings exist.

Observation vs Inference

• Observation: repeated passages through an avoided crossing produce oscillatory final-state populations.
• Measurement: fringe phase changes systematically with time, detuning and drive amplitude.
• Inference: the first crossing created coherent amplitudes that accumulated relative phase before recombination.
• Model: Landau–Zener beam splitters plus Stückelberg phase accumulation.
• Boundary: multilevel, strongly overlapping, highly dissipative or arbitrarily driven systems can require full Floquet/numerical treatment.

Common Misconceptions

Misconception	Better model
Landau–Zener transition means the system “chooses” one path classically.	In coherent evolution it can leave the crossing in a superposition.
Two crossings just square the single-pass probability.	The amplitudes interfere with a phase-dependent cross term.
Fringes prove Berry phase specifically.	Total phase includes dynamical, Stokes and sometimes geometric components.
Loss of fringes proves the avoided crossing disappeared.	Dephasing can destroy interference while the gap remains.

Checkpoint Questions

1. What is an avoided crossing?
2. What determines whether passage is adiabatic or diabatic?
3. Why is one intermediate-speed crossing like a beam splitter?
4. What phase accumulates between passages?
5. What does the second crossing do?
6. What creates Stückelberg fringes?
7. Why does dephasing reduce visibility?
8. How can geometric phase enter without owning the whole effect?
9. How is LZS different from Bloch oscillations?
10. When does the two-level adiabatic-impulse model fail?

Answer Key

Open after attempting the questions

1. A level crossing opened into a finite gap by coupling.
2. The sweep rate relative to the squared gap and local level velocity.
3. It coherently divides amplitude between two branches.
4. Mainly the integrated energy difference plus Stokes and sometimes geometric contributions.
5. It recombines the amplitudes and converts phase into population.
6. Constructive/destructive interference as the relative phase is varied.
7. It randomizes relative phase between the two paths.
8. The control path can add a Berry contribution to total interferometer phase.
9. Bloch oscillations arise from periodic-band motion under force; LZS arises from repeated nonadiabatic level splitting and recombination.
10. When crossings overlap, more levels participate, coherence assumptions fail or the drive cannot be separated into isolated impulses and adiabatic intervals.

Primary Science Bridge

• one event can split a system into two possible wave paths;
• two paths can collect different timing/phase;
• meeting again can create cancellation or reinforcement;
• repeating the same boundary can reveal hidden memory;
• the same transition point can be both a divider and a detector.

Secondary and JC Bridge

Core idea	Higher-resolution route
Coupled levels	Avoided crossing
Sweep	Landau–Zener transition
Superposition	Amplitude beam splitting
Phase	Dynamical + Stokes phase
Recombination	Stückelberg interference
Driven systems	Floquet/multipassage dynamics

Unfamiliar Transfer Challenge

A superconducting qubit driven back and forth through an anticrossing shows diagonal interference fringes versus drive amplitude and frequency. Increasing environmental noise erases the fringes while the single-pass transition probability remains almost unchanged.

This pattern strongly supports a Stückelberg-interference interpretation: the beam-splitting event survives, while the phase memory between passages is being destroyed.

Deep Science Window — Mach–Zehnder in Energy Space

The mapping to a Mach–Zehnder interferometer is unusually precise: first avoided crossing = first beam splitter; two adiabatic branches = interferometer arms; energy-difference integral = path phase; second crossing = recombining beam splitter; final level population = output intensity. This representation lets optical interference intuition transfer directly into driven two-level quantum dynamics.

Deep Science Window — Repeated Boundaries Create Memory

A single transition boundary tells us the local crossing probability. Revisit the boundary before coherence is lost and the second outcome depends on what happened between encounters. The boundary has become part of a memory-sensitive sequence rather than an isolated event. This repeated-boundary principle appears widely in coherent control and dynamical systems.

Evidence Boundaries

• LZS interference ≠ single-pass Landau–Zener transition.
• Stückelberg fringes ≠ automatically Berry-phase fringes.
• Periodic oscillation ≠ uniquely LZS.
• Loss of fringes ≠ loss of the avoided crossing.
• Two-level adiabatic-impulse model ≠ exact for overlapping multilevel crossings.
• Coherent path language ≠ claim that the system followed one classical trajectory between crossings.

Manual Summary — KNOW → CONNECT → EXPLAIN → APPLY → CHECK

KNOW: avoided crossing, Landau–Zener probability, superposition, phase accumulation, Stokes phase, Stückelberg fringes.

CONNECT: first crossing to amplitude splitting, inter-crossing evolution to phase, and second crossing to interference.

EXPLAIN: how crossing the same avoided crossing twice converts a transition point into a quantum interferometer.

APPLY: distinguish coherent LZS fringes from ordinary driven oscillations or independent transitions.

CHECK: vary passage separation, dephasing, sweep rate and level structure before assigning the mechanism.

Research Sources and Further Reading

• Physics Reports — Nonadiabatic Landau–Zener–Stückelberg–Majorana Transitions, Dynamics and Interference
• Physical Review Letters (2026) — Landau–Zener Tunnelling and Quantum Interference of Andreev States