Generalized Sauter Model in Vacuum Decay

• The generalized Sauter model is a nonperturbative framework where spatial and temporal deformations of strong electric fields enable enhanced electron-positron pair creation.
• It employs both perturbative expansions and semiclassical worldline instanton methods to analyze composite field configurations and various assisting pulse shapes.
• Applications include laboratory analogues in optical lattices and semiconductors, offering insights into threshold behaviors and dynamic assistance mechanisms.

The generalized Sauter model denotes a family of extensions of the Sauter–Schwinger pair-creation problem in which the original smooth strong-field background is supplemented or deformed while the central nonperturbative mechanism is retained. In the literature, this label is used in several closely related senses: for spacetime-dependent Sauter-type electric fields such as a localized static profile plus a time-dependent assisting component; for composite fields consisting of a strong slow Sauter-like background and a weak fast pulse of variable shape; and for configurations in which a strong constant electric field is assisted by a weak transverse plane wave. Across these variants, the common structure is that a strong field remains the nonperturbative source of vacuum decay, while added temporal or spatial structure modifies the tunneling barrier, the threshold for enhancement, or the spectrum of observables (Linder et al., 2015, Schneider et al., 2016, Torgrimsson et al., 2017).

1. Terminology and canonical meaning

In its narrowest historical sense, the Sauter model refers to the smooth, exactly solvable electric-field profiles

$$E(t)=\frac{E_0}{\cosh^2(\omega t)}, \qquad E(x)=\frac{E_0}{\cosh^2(kx)}.$$

A generalized Sauter setting extends this to fields depending on both space and time, for example

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$

In that usage, the essential feature is a smooth, exponentially localized background for which instanton and tunneling methods remain applicable, now with explicit interplay between spatial localization and temporal assistance (Linder et al., 2015).

A second, equally standard usage treats the generalized Sauter model as a composite-field framework in which a strong Sauter-like slow background is combined with a weaker fast pulse of arbitrary shape. In Euclidean notation this is written through

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$

so that the strong field is Sauter-like while the assisting component is encoded in $h(\chi)$. In this formulation, “generalized” does not mean a single universal profile; it means a class of assisted Sauter backgrounds encompassing Sauter, Gaussian, cosine, and Lorentzian weak pulses within the same semiclassical machinery (Schneider et al., 2016).

2. Field configurations and geometric generalizations

For homogeneous time-dependent assistance, the basic field is typically written as

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$

with $E f_0(t)$ the strong slow part and $E\varepsilon f_1(t)$ the weaker faster part. Representative assisting profiles are the Sauter pulse $1/\cosh^2(\omega_{\rm fast} t)$, the Gaussian $\exp\{-(\omega_{\rm fast} t)^2\}$, and the sinusoid $\cos(\omega_{\rm fast} t)$. The distinction between these cases is not merely cosmetic in the time domain; it is tied to the large-frequency decay of the Fourier transform and therefore to which perturbative orders dominate the rate (Torgrimsson et al., 2017).

A more explicit spacetime-dependent extension is obtained by superimposing a weak transverse plane wave on a strong constant field. In the simplest geometry,

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$0

with temporal-gauge potential

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$1

The paper also considers the more general configuration

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$2

where $$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$3 and $$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$4. In this geometry, the enhancement is controlled by the component transverse to the strong-field direction, formally through the replacement $$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$5. The optimal incidence is therefore the perpendicular case $$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$6, and parallel polarization to the strong field gives the largest enhancement, whereas perpendicular polarization yields a smaller probability and introduces spin-factor dependence (Torgrimsson et al., 2017).

These constructions clarify that the generalized Sauter model is not restricted to purely temporal assistance. Spatial dependence may enter either through localized static backgrounds, as in $$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$7, or through genuine wave propagation, as in the transverse plane-wave extension. This suggests that “generalized” is best understood as a controlled deformation of the original Sauter background into a spacetime-dependent one, rather than as a single fixed field profile.

3. Dynamical assistance, thresholds, and dominant orders

The central physical effect in generalized Sauter backgrounds is dynamical assistance: a weak fast component can exponentially enhance pair creation produced by a strong slow component. For Sauter-like temporal assistance, the relevant combined Keldysh parameter is

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$8

and the threshold is

$$E(t,x)=\frac{E_1}{\cosh^2(kx)}+\frac{E_2}{\cosh^2(\omega t)}, \qquad E(t,x)=\frac{E_1}{\cosh^2(kx)}+E_2\cos(\omega t).$$9

For Gaussian assistance the onset scales instead as

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$0

while for a sinusoidal assisting field it scales as

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$1

The physical reason given for this separation is the asymptotic Fourier behavior of the weak pulse: Sauter-like profiles have exponentially decaying Fourier tails, Gaussian profiles decay faster than exponential, and a cosine injects discrete frequencies. That distinction determines whether the zeroth-plus-first perturbative orders already capture the leading exponent or whether higher orders become dominant (Torgrimsson et al., 2017).

The prefactor analysis shows that these threshold classes are not an artifact of exponent-only approximations. In the worldline-instanton treatment of the assisted problem, the fluctuation prefactor modifies the quantitative rate but does not change the qualitative pulse-shape dependence: Sauter and Lorentzian assistance exhibit a threshold approximately independent of $$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$2, whereas cosine and Gaussian assistance retain explicit $$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$3-dependence. Comparison with Riccati-equation numerics also shows that the semiclassical method can produce an unphysical dip just below threshold for small $$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$4, which is absent in the Riccati solution and is therefore identified as a semiclassical artifact (Schneider et al., 2016).

For the plane-wave-assisted constant field, the combined Keldysh parameter is

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$5

with threshold

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$6

The perturbative expansion is

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$7

where the zeroth term reproduces the Schwinger factor

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$8

and odd orders vanish in the setup considered. The higher-order terms contain an effective mass

$$f(\chi)=\frac{1}{\rho}\tan(\rho \chi)+h(\chi), \qquad \epsilon\ll1,\quad \rho\ll1,$$9

reflecting the fact that the transverse wave transfers momentum as well as energy. Because of this momentum-conservation penalty, the enhancement is weaker than in purely time-dependent assistance, even though it remains exponential above threshold. The strong field remains essential throughout: the plane wave alone cannot create pairs, so the process stays nonperturbative in $h(\chi)$0 rather than reducing to a purely multiphoton mechanism (Torgrimsson et al., 2017).

4. Analytical frameworks and semiclassical structure

The two main calculational strategies are perturbative expansion in the weak component and fully nonperturbative worldline instanton methods. In the perturbative approach, the strong field is kept exact while the weak field is inserted order by order. For homogeneous assistance, the probability is written schematically as

$h(\chi)$1

For a broad class of pulses with exponential Fourier tails, including the Sauter pulse, the sum of the zeroth- and first-order amplitudes already gives good agreement with nonperturbative results. For Gaussian and sinusoidal assistance, by contrast, higher orders may carry different leading exponentials and can dominate for appropriate parameters. The large-frequency Fourier tail is therefore the organizing principle for classifying generalized Sauter assistance (Torgrimsson et al., 2017).

The fully nonperturbative description is formulated through the imaginary part of the effective action,

$h(\chi)$2

and the Euclidean worldline saddle-point equations

$h(\chi)$3

In homogeneous time-dependent problems, the exponent can be organized by a function $h(\chi)$4, while the subleading fluctuation factor is encoded in derivatives of $h(\chi)$5. In the spacetime-dependent plane-wave problem, the instantons are not purely real in the parallel-polarization case and become effectively three-dimensional or four-dimensional depending on polarization. The saddle-point problem is then solved numerically by a discretized worldline instanton method using Newton–Raphson plus continuation. A recurrent conclusion is that agreement between perturbative and nonperturbative descriptions is obtained only when the perturbative sum is carried to sufficiently high order, because the dominant order need not be the first nontrivial term (Torgrimsson et al., 2017, Schneider et al., 2016).

5. Analog and simulator realizations

Generalized Sauter physics has also been reformulated as a controllable many-body problem in condensed-matter and cold-atom platforms. In the optical-lattice proposal, ultra-cold fermionic atoms in a one-dimensional bichromatic lattice with

$h(\chi)$6

realize a discretized Dirac Hamiltonian. The two split sub-bands play the role of particle and antiparticle branches, the filled lower band acts as the Dirac sea, and an external lattice potential $h(\chi)$7 mimics an electrostatic potential. Spontaneous pair creation is modeled by slowly turning on $h(\chi)$8, driving a bound state into the negative continuum at supercriticality, and then switching the potential off so that a particle-hole excitation remains. In this sense, the model is a quantum-simulation generalization of the Sauter–Schwinger mechanism rather than a new analytic Sauter profile (Szpak et al., 2011).

A semiconductor realization develops a quantitative correspondence between interband tunneling and the Sauter–Schwinger effect, including spacetime-dependent fields. In a two-band model near a direct gap at $h(\chi)$9, the effective parameters

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$0

map the semiconductor dynamics onto a Dirac-like form. For full spacetime dependence, the correspondence requires slowly varying fields, small spatial Fourier components, and matched band curvatures near the gap,

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$1

The paper explicitly treats generalized Sauter backgrounds such as

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$2

and concludes that spatial localization of the strong field lowers the threshold for dynamical assistance. This provides a laboratory analogue of the generalized Sauter model in which the same threshold and tunneling concepts can be explored at accessible scales (Linder et al., 2015).

6. Observables, basis issues, and recent extensions

Later work has shifted part of the emphasis from total asymptotic yields to intermediate-time observables. In scalar QED in $$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$3 dimensions, an intense temporal Sauter pulse

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$4

admits exact hypergeometric mode solutions. The associated vacuum polarization current is defined by

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$5

where $$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$6 is the real part of the pair-correlation function. A central result is that the adiabatic particle number is basis-dependent at intermediate times, whereas the polarization current is basis independent for the two adiabatic prescriptions compared. For scalar particles the current changes sign, does not settle to a constant after the pulse, and instead shows nearly undamped late-time oscillations around zero. This resolves a common misconception that the instantaneous particle number is the uniquely meaningful dynamical observable during a Sauter pulse (Sah et al., 29 Mar 2025).

A different extension keeps the field homogeneous in space but generalizes the temporal profile to finite-duration circularly polarized pulses and pulse trains with tunable carrier-envelope phase and helicity. The field envelope is

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$7

and the pulse is

$$\mathbf{E}(t)=E\,[f_0(t)+\varepsilon f_1(t)]\,\mathbf{e}_z, \qquad \varepsilon\ll1,$$8

Solving the Dirac equation with Feynman or anti-Feynman boundary conditions yields helicity-resolved pair-creation amplitudes, from which spirals, nodal lines, vortex streets, and Bell-type helicity entanglement in momentum space are extracted. The carrier-envelope phase can move or annihilate vortices, transform vortex streets into nodal lines, and switch the dominant Bell channel. This suggests that the generalized Sauter model has evolved beyond rate calculations into a broader framework for studying spin, topology, and entanglement in strong-field vacuum decay (Majczak et al., 11 Jul 2025).

The generalized Sauter model is therefore best understood as a technically unified but physically diverse class of strong-field backgrounds. Its unifying content is nonperturbative pair creation by a dominant field in the presence of additional temporal, spatial, or polarization structure; its main differentiators are the geometry of the background, the spectral decay of the assisting component, and the observables selected for analysis.