Schwinger Phase in QED & Pair Production

• Schwinger phase is a term for a family of related phase structures in quantum electrodynamics, reflecting variations in topological sectors, CP-symmetry breaking, and gauge invariance.
• It encompasses phenomena like second-order quantum critical points at θ=π in the one-flavor Schwinger model and first-order transitions driven by fermion mass and chemical potential.
• Advanced numerical, lattice, and quantum simulation methods are employed to capture these phase transitions and control pair production through precise phase manipulation.

“Schwinger phase” is not a uniquely standardized term in current arXiv usage. In the literature on $1+1$-dimensional QED, it most commonly denotes the phase structure of the Schwinger model under variations of the topological angle $\theta$, fermion mass, temperature, and chemical potential. In phase-space formulations of Schwinger pair production, the closest formal analogue is instead a gauge-phase factor implemented as a Wilson line in the Wigner operator. A distinct usage appears in strong-field QED, where relative phase shifts and carrier-envelope phases control the spacetime structure of the invariant electric field and hence the pair-production rate (Funcke et al., 2022, Bañuls et al., 2016, Kohlfürst, 2017, Banerjee et al., 2018).

1. Terminological scope

Across the cited literature, the expression is used in several technically different senses. In multiflavor lattice studies at nonzero chemical potential, “Schwinger phases” are distinct ground-state sectors selected by density or flavor imbalance. In one-flavor and two-flavor studies with a topological $\theta$-term, the relevant phases are vacuum sectors distinguished by CP or charge-conjugation realization, spectral gaps, and order parameters such as the electric field. In DHW/Wigner treatments of inhomogeneous pair production, the closest object to a “Schwinger phase” is the gauge-link phase factor that renders the nonlocal correlator gauge invariant (Bañuls et al., 2016, Funcke et al., 2022, Ohata, 2023, Kohlfürst, 2017).

Usage	Setting	Technical meaning
Schwinger-model phase	$\theta$-term, mass, temperature, chemical potential	Ground-state or thermal phase structure
Gauge phase factor	DHW/Wigner formalism	Wilson line ensuring gauge invariance
External phase control	Colliding-pulse Schwinger effect	Relative phase shift or CEP controlling pair yield

This multiplicity of usage implies that the term must be interpreted contextually. A plausible implication is that, in contemporary research writing, “Schwinger phase” is better treated as a family of related notions tied to Schwinger-model criticality and Schwinger-effect phase dependence, rather than as a single canonical object.

2. One-flavor Schwinger model at $\theta=\pi$

For the one-flavor Schwinger model, the central phase-structure problem is the special $\theta=\pi$ sector. In Euclidean form, the model is

$S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$

supplemented by the topological term

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$

At $\theta=\pi$, because $Q\in\mathbb{Z}$, CP is not explicitly broken. The continuum Hamiltonian density can also be written so that the $\theta$0-term appears as a background electric field,

$\theta$1

or, in Coleman’s parametrization, $\theta$2 (Ohata, 2023, Kaikov et al., 2024, Thompson et al., 2021).

At zero temperature, increasing the fermion mass drives a transition from a CP-symmetric phase with one vacuum to a CP-broken phase with two degenerate vacua. The transition is second order. A simple expansion of the bosonized effective potential gives

$\theta$3

while a more precise DMRG estimate gives

$\theta$4

The reported critical exponents,

$\theta$5

match the quantum Ising values $\theta$6 and $\theta$7, placing the quantum critical point in the $\theta$8D Ising / quantum Ising universality class (Ohata, 2023).

A complementary cut through the phase diagram is obtained by fixing $\theta$9 and varying $\theta$0. For sufficiently large $\theta$1, $\theta$2 lies on a first-order transition line: the two lowest energy levels cross, the ground-state energy is non-analytic, and its first derivative is discontinuous at the crossing. Below the endpoint at $\theta$3, the lowest levels do not cross and the spectrum remains gapped, so there is no transition. This resolves an apparent ambiguity: the same critical point is second order as a function of mass at fixed $\theta$4, yet it terminates a first-order line in the $\theta$5 plane (Kaikov et al., 2024, Angelides et al., 2023).

3. Finite-temperature scaling near the quantum critical point

Near the one-flavor quantum critical point at $\theta$6, the finite-temperature phase diagram is organized by the scaling variable

$\theta$7

with nonuniversal constant $\theta$8. The electric-field correlator,

$\theta$9

has long-distance asymptotics

$\theta$0

where $\theta$1 are nonuniversal constants and $\theta$2 are universal scaling functions of the quantum Ising chain (Ohata, 2023).

This scaling form yields three asymptotic regions. In the thermally disordered region, $\theta$3 and low $\theta$4, the system lies close to the CP-broken zero-temperature phase, but thermal fluctuations destroy long-range order. In the quantum disordered region, $\theta$5 and low $\theta$6, the system is already disordered at $\theta$7. In the quantum critical region, $\theta$8, the correlation length is set by temperature and scales as $\theta$9. More specifically,

$\theta=\pi$0

with asymptotics that are exponentially large on the thermally disordered side, equal to $\theta=\pi$1 in the quantum critical region, and saturated at $\theta=\pi$2 on the quantum disordered side (Ohata, 2023).

At the critical mass $\theta=\pi$3,

$\theta=\pi$4

and the amplitude is

$\theta=\pi$5

The nonuniversal constants extracted numerically are

$\theta=\pi$6

A major consequence is that CP symmetry is restored at any nonzero temperature, entirely analogous to $\theta=\pi$7 symmetry restoration in the finite-temperature quantum Ising chain (Ohata, 2023).

4. Multiflavor, Dashen, and density-induced phases

The two-flavor Schwinger model at zero temperature exhibits a richer phase structure. For all $\theta=\pi$8, the vacuum is non-degenerate. At $\theta=\pi$9, however, there is a region in the $\theta=\pi$0-plane where charge conjugation $\theta=\pi$1 is spontaneously broken and the vacuum is doubly degenerate. The boundary of this broken region consists of two critical curves meeting at the origin, each described in the infrared by the $\theta=\pi$2D Ising CFT with central charge $\theta=\pi$3. At the origin, the low-energy theory is the $\theta=\pi$4 WZW model with $\theta=\pi$5. Along the $\theta=\pi$6-invariant line $\theta=\pi$7, the theory undergoes BKT-type logarithmic RG flow, and the gap is non-perturbatively small,

$\theta=\pi$8

so that

$\theta=\pi$9

This identifies a regime in which dimensional transmutation coexists with symmetry-breaking boundaries of Ising type (Dempsey et al., 2023).

A distinct two-flavor regime is the Schwinger-model analogue of the CP-violating Dashen phase. With $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$0 and $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$1 scanned around $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$2, matrix-product-state calculations find abrupt changes in the average electric field and in the lattice analogue of the pion condensate. The bipartite entanglement entropy peaks near

$S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$3

slightly above $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$4, and its logarithmic scaling with volume indicates that the transition is not first order. In the paper’s interpretation, a nonzero pion-condensate analogue is a CP-odd order parameter, so the transition is identified as a CP-violating Dashen transition (Funcke et al., 2023).

At nonzero chemical potential, multiflavor models support density-induced phase sequences. In the two-flavor massless case at zero temperature, the relevant control parameter is the isospin chemical potential, and the phases are labeled by the integer isospin number

$S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$5

The transitions are first order and occur at odd multiples of $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$6 in the rescaled variable $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$7. For nonzero mass, the tower of phases persists but the spacings are distorted: the $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$8 phase expands and the $S_E = \int d^2x\, \bar\psi(\slashed{\partial}+g\slashed{A}+m)\psi+\frac14 F_{\mu\nu}F_{\mu\nu},$9 plateau shrinks. In the three-flavor model, the analogous labels are flavor-number differences $$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$0, and abrupt jumps in these observables diagnose first-order transitions as the chemical potentials are varied (Bañuls et al., 2016, Funcke et al., 2022).

5. Gauge-phase factors and the distinction from Schwinger functions

In phase-space treatments of pair creation, the object closest to a Schwinger phase is the gauge-link factor in the covariant Wigner operator. The DHW formalism defines the gauge-invariant density operator through

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$1

with Wilson line

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$2

The corresponding covariant Wigner operator is

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$3

The cited paper explicitly states that it does not use the term “Schwinger phase” by name, but identifies this Wilson-line factor as the closest formal analogue because it encodes the electromagnetic phase accumulated along the straight path connecting the two spacetime points and is essential for gauge invariance (Kohlfürst, 2017).

This usage should be distinguished from the Schwinger function criterion in Dyson–Schwinger studies of QCD. There, the projected propagator is decomposed as

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$4

and the Schwinger function is

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$5

Because the ordinary Schwinger function is an integral transform of the spectral density, it can miss positivity violations. The proposed generalized criterion therefore uses even derivatives,

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$6

In the refined truncation studied, the deconfinement line obtained from this generalized Schwinger criterion overlaps with the Nambu-phase chiral transition line, and the common critical endpoint is reported at

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$7

Thus, “Schwinger function” and “Schwinger phase” refer to separate constructions, even though both arise in gauge-theory diagnostics (Gao et al., 2016).

6. Phase control in Schwinger pair production

In strong-field QED, phase variables control Schwinger production directly. For two counterpropagating focused laser pulses with relative phase shift $$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$8, the locally constant-field pair-production rate is written as

$$i\theta Q,\qquad Q=\int d^2x\,\frac{g}{4\pi}\epsilon_{\mu\nu}F_{\mu\nu}.$$9

where $\theta=\pi$0 and $\theta=\pi$1 are determined by the Lorentz invariants

$\theta=\pi$2

When the field is nearly purely electric, $\theta=\pi$3, this reduces to

$\theta=\pi$4

The relative phase shift $\theta=\pi$5 and the carrier-envelope phase $\theta=\pi$6 move electric-like and magnetic-like domains in spacetime. For linearly polarized pulses, the longitudinal structure depends on $\theta=\pi$7, whereas temporal distributions depend on both $\theta=\pi$8 and $\theta=\pi$9. For circular polarization, the RR configuration depends only on $Q\in\mathbb{Z}$0, while the RL configuration depends on the combination $Q\in\mathbb{Z}$1. Pair production is maximized when a strong electric spike remains near the center of the pulse envelope and is suppressed when the peak is displaced from it (Banerjee et al., 2018).

The same theme reappears in the DHW analysis of inhomogeneous few-cycle pulses, where the vector potential

$Q\in\mathbb{Z}$2

contains the carrier-envelope phase $Q\in\mathbb{Z}$3. The cited analysis reports that $Q\in\mathbb{Z}$4 yields mostly single-peak-like spectra, $Q\in\mathbb{Z}$5 produces the strongest and most symmetric interference, and intermediate values generate asymmetric double-peak structures. The paper characterizes this as a “double-slit experiment in time” when multiple strong temporal peaks are present (Kohlfürst, 2017).

In confining holographic backgrounds, phase structure enters through the background itself. For both confining D3-brane and D4-brane geometries, there are two critical electric fields: a lower threshold $Q\in\mathbb{Z}$6, equal to the confining string tension, below which no Schwinger effect occurs, and an upper threshold $Q\in\mathbb{Z}$7, at which the potential barrier disappears and the vacuum becomes catastrophically unstable. Between them, $Q\in\mathbb{Z}$8, pair creation proceeds by tunneling (Sato et al., 2013).

7. Numerical and quantum-simulation realizations

The modern phase literature on the Schwinger model is strongly shaped by sign-problem avoidance and low-energy state preparation. Finite-temperature studies near the one-flavor quantum critical point use first-principle Monte Carlo simulations of the lattice bosonized Schwinger model specifically to avoid the sign problem. Zero-temperature multiflavor phase diagrams at finite chemical potential and in the Dashen regime are mapped by matrix-product-state methods, which operate directly in the Hamiltonian framework and therefore do not rely on probabilistic weights. This is why these methods can access regimes described in the cited papers as sign-problem afflicted for conventional Monte Carlo (Ohata, 2023, Bañuls et al., 2016, Funcke et al., 2023).

Quantum-computing approaches translate the same phase questions into ground-state preparation, spectral tracking, and observable jumps. A variational quantum eigensolver for the three-flavor model uses a layered, charge-preserving ansatz and reproduces the first-order phase structure through discontinuities in flavor-number differences. For the one-flavor model with a topological $Q\in\mathbb{Z}$9-term, VQE studies with Wilson and staggered fermions identify electric field density and particle number as the observables most directly revealing the first-order transition, and report that additive mass renormalization is vital for accurate continuum extrapolation. A momentum-space VQE implementation on IBM Q Lima gives evidence for the critical point at

$\theta$00

in good agreement with the classical numerical result $\theta$01 (Funcke et al., 2022, Angelides et al., 2023, Thompson et al., 2021).

Adiabatic state-preparation methods offer a different route. By separately evolving the ground state and first excited state under a time-dependent Hamiltonian that sweeps $\theta$02, one can distinguish a level crossing from a persistent gap and thereby identify whether the system lies in the first-order-transition region or in the no-transition region. In a related dynamical setting, a digital Rydberg simulation of the $\theta$03 Schwinger model studies a negative-mass quench that produces resonant Rabi oscillations between the Dirac vacuum and mesonic states, leading to repeated near-zeros of the Loschmidt amplitude and hence multiple dynamical quantum phase transitions. This broad methodological landscape suggests that “Schwinger phase” is now as much a problem of controlled state discrimination and continuum matching as it is a problem of analytic classification (Kaikov et al., 2024, Pomarico et al., 28 Jul 2025).