Coupled Kerr-Nonlinear Mach-Zehnder Interferometers

• Coupled Kerr-nonlinear Mach-Zehnder interferometers are electronic devices that use capacitive coupling in quantum Hall edge states to generate conditional, nonlinear phase shifts.
• The device architecture splits electron wave-packets into superpositions so that selective Coulomb interactions in adjacent arms yield path-dependent phase entanglement and observable Bell violations.
• Realistic implementations account for many-body effects, pulse distortion, and self-capacitance, which reduce but still permit measurable entanglement with optimized Bell parameters exceeding 2.

Searching arXiv for the specified paper and closely related terminology to ground the article in current metadata. Coupled Mach–Zehnder interferometers, in the mesoscopic realization analyzed in "Two-particle entanglement in capacitively coupled Mach-Zehnder interferometers" (Vyshnevyy et al., 2012), are two electronic interferometers implemented in a quantum Hall edge-state geometry and linked through a finite interaction region where adjacent inner arms are capacitively coupled. Two wave-packets injected on demand are split into path superpositions; only the component pair traversing the interacting arms acquires an interaction-induced phase. The resulting path-selective phase entangles the outgoing orbital, or lead, degree of freedom and permits a Bell-inequality test based on particle-number correlations in the four output leads. This suggests an effective nonlinear phase element implemented by capacitive Coulomb coupling rather than by a material Kerr medium.

1. Device architecture and interaction model

The device consists of two electronic Mach-Zehnder interferometers in a quantum Hall structure. The upper interferometer uses incoming/outgoing leads $1'$ and $3'$, and the lower interferometer uses incoming/outgoing leads $2'$ and $4'$. The adjacent inner arms, labeled $1'$ and $2'$, are capacitively coupled over a finite interaction region. Each interferometer contains an incoming beam splitter, $A$ or $A'$, an outgoing beam splitter, $B$ or $B'$, and a magnetic flux, $3'$0 or $3'$1, which sets the relative phase between the two paths (Vyshnevyy et al., 2012).

The interaction is modeled by a quadratic charge Hamiltonian,

$3'$2

where $3'$3 counts excess charge in the interacting segment of lead $3'$4. The matrix $3'$5 contains self-capacitances $3'$6 and $3'$7, and the mutual capacitance $3'$8. In the idealized two-particle discussion, the self-terms are dropped and only the mutual coupling matters; in the full Fermi-sea problem, both self and mutual couplings contribute.

This geometry is notable because the coupling is neither global nor uniformly distributed across all amplitudes. It is confined to one pair of adjacent inner arms, so the interaction acts only on a specific branch of the two-particle superposition. The entangling resource is therefore a conditional phase generated by spatially selective Coulomb coupling.

2. Scattering formulation and path-selective coupling

The analysis is formulated in a scattering picture. Two wave-packets are injected on demand, one in lead $3'$9 and one in lead $2'$0. The incoming beam splitters divide each packet into two components, producing four path combinations. Only the component in which both particles simultaneously propagate through the interacting arms $2'$1 and $2'$2 experiences Coulomb coupling; the other path combinations propagate freely (Vyshnevyy et al., 2012).

For symmetric incoming splitters, $2'$3, the state before the outgoing beam splitters has the form

$2'$4

where only $2'$5 contains the interaction.

The physical content of this expression is that entanglement does not arise from a uniform phase shift applied to each interferometer separately. It arises from the relative phase between the interacting and noninteracting path amplitudes. In that sense, the coupling is intrinsically two-body: the relevant phase is generated only when both particles occupy the coupled arms.

3. Interaction-induced phase and entanglement structure

For the idealized two-electron problem, the interaction generates a path-dependent phase according to

$2'$6

With exponential interaction kernels,

$2'$7

the asymptotic phase becomes

$2'$8

where

$2'$9

The parameter $4'$0 is therefore the accumulated Coulomb phase during simultaneous traversal of the interaction region. For Lorentzian wave-packets, the overlap factor is

$4'$1

For infinitely narrow packets, $4'$2,

$4'$3

The two-particle density matrix is

$4'$4

with $4'$5.

In the narrow-packet limit, the maximal Bell value is

$4'$6

Full entanglement, and maximal Bell violation, occurs when

$4'$7

which gives

$4'$8

In this ideal limit, the output behaves as a maximally entangled Bell pair in the orbital or lead basis. For finite-width packets, however, entanglement is reduced because not all wave-packet components overlap in the interaction region, the interaction deforms the packet shape, and which-path information is stored in the wave-function shape. A common simplification is to identify the device directly with a two-qubit Bell-state generator; the more precise statement is that such a description is exact only in the idealized two-electron, narrow-packet limit.

4. Bell-test formulation in the output leads

The Bell test is expressed through particle-number correlations in the four outgoing leads. The Bell correlator is defined as

$4'$9

and the CHSH-like Bell parameter is

$1'$0

The local-hidden-variable bound is

$1'$1

Within this formulation, the outgoing beam splitters $1'$2 and $1'$3, together with the fluxes $1'$4 and $1'$5, play the role of the four analyzer settings in a Bell test (Vyshnevyy et al., 2012). For the ideal two-electron case, the optimal flux settings are

$1'$6

with $1'$7 and a suitable $1'$8, and the Bell inequality can reach the Tsirelson value $1'$9.

The significance of this construction is operational. Nonclassical correlations are not inferred from an abstract state-vector decomposition alone, but from a detection protocol formulated directly in terms of measurable output occupations.

5. Many-body effects: Fermi seas, electron-hole pairs, and timing

The realistic treatment includes the filled Fermi seas in the Hall edge channels. Once the interaction is placed on top of the Fermi sea, the injected electrons can exchange energy with the background, electron-hole pairs are generated, the outgoing state is no longer just a simple two-particle state, and information about the path is partly stored in unobserved many-body degrees of freedom (Vyshnevyy et al., 2012).

The reduction of the Bell signal has two identified mechanisms: decoherence, or entanglement leakage into environment-like degrees of freedom, and pulse distortion, which stores which-path information in the wave-packet shape. In this many-body setting, the ideal Bell-pair picture ceases to be complete. The state retains nonclassical correlations, but these correlations are partially distributed over additional electronic degrees of freedom.

A time delay between the injected packets weakens the effective interaction. For a delay $2'$0, the phase is suppressed roughly as

$2'$1

so the entangling interaction becomes weaker as the pulses separate in time.

Finite-width packets further reduce visibility. For broader packets, corresponding to large $2'$2, the particles are less likely to overlap in the interaction region, the effective entangling phase decreases, and the Bell violation shrinks. In the ideal two-particle setup, using the non-optimal settings optimized at $2'$3, violation persists only for $2'$4, whereas with full optimization it persists up to $2'$5.

In the Fermi-sea case, the relative strength of self- and mutual-capacitance is parameterized by

$2'$6

Because the relevant interaction phases and visibilities are altered by self-action and pair creation, the maximal Bell signal is lower than in the ideal two-particle case.

6. Optimized operating regime and observable Bell violation

The principal numerical conclusion of the realistic analysis is that Bell violation survives even after including Fermi seas, electron-hole excitations, and interaction-induced pulse distortion. Using the flux and angle settings optimized for the ideal two-particle problem, the Bell parameter reaches about

$2'$7

at

$2'$8

With full optimization of the Bell test through the Horodecki procedure, the violation is stronger: $2'$9 at approximately

$A$0

(Vyshnevyy et al., 2012).

These values establish that the interaction-induced correlations remain experimentally meaningful beyond the ideal two-electron limit. The authors emphasize that $A$1 corresponds to $A$2, that is, the mutual coupling must be stronger than the self-coupling in the relevant regime. A plausible implication is that practical operation requires engineering the interaction region so that the conditional inter-arm phase dominates over self-action and many-body broadening.

Taken together, the coupled-interferometer mechanism is defined by a specific sequence: two electrons are injected on demand into two coupled Mach-Zehnder interferometers, beam splitters create path superpositions, Coulomb coupling acts only when both electrons occupy the adjacent interacting arms, the selective interaction imprints a relative phase, this phase entangles the outgoing orbital state, and Bell violation is detected via particle-number correlations in the four output leads. In the ideal two-particle case, the setup reaches maximal entanglement and the Tsirelson bound; in the realistic many-body case, entanglement is reduced by electron-hole pair production and pulse distortion but remains sufficient for a measurable violation, with a reported optimized maximum of $A$3.