Below-Threshold Harmonics: Theory & Applications

• BTH is defined as harmonic emissions with photon energies below the ionization limit, produced through bound–bound transitions and resonantly enhanced processes.
• The generation mechanisms span perturbative multiphoton resonances to nonperturbative, sub-cycle dynamics similar to plateau HHG, with notable Stark shifts and polarization effects.
• These processes provide a pathway for precision spectroscopy and diagnostics in multi-electron systems by controlling laser intensity, ellipticity, and dynamic resonances.

Below-threshold harmonics (BTHs) are harmonic emissions whose photon energies lie below the ionization threshold of the generating medium. In the literature summarized here, BTH occupies several partially overlapping regimes: coherent, parametric odd-order harmonic generation in isotropic media at intensities too low to produce significant multiphoton ionization; resonant or near-resonant bound–bound emission in the multiphoton regime; and a nonperturbative, sub-cycle process whose microscopic origin can be strikingly similar to plateau high-harmonic generation (HHG). Across these formulations, BTH is distinguished by its sensitivity to discrete excited states, dynamic Stark shifts, orbital-energy gaps, polarization selection rules, and, in multi-electron targets, electron–electron energy transfer (Wang et al., 2019).

1. Definition and conceptual scope

A standard definition identifies BTHs as harmonics whose photon energies lie below the ionization threshold. In isotropic media under laser intensities below threshold for multiphoton ionization, BTH generation is described as the coherent, parametric production of odd-order harmonics. In the multiphoton ionization regime, it is also described as coherent emission of “low-order” harmonics produced by resonant or near-resonant excitation of bound–bound transitions followed by stimulated emission of harmonic photons. These descriptions emphasize that below-threshold emission need not require continuum recollision and may be dominated by discrete-state dynamics (Andrews et al., 2019).

The modern view is more differentiated than the older tendency to classify below-threshold response as purely perturbative. One line of work reports that BTHs have long been viewed as “perturbative,” yet recent experiments and ab-initio simulations show that they share a nonperturbative, sub-cycle origin very similar to plateau harmonics. Another line of work, focused on the multiphoton regime with low ionization probability, explicitly contrasts resonant BTH with plateau harmonics produced by the three-step recollision mechanism and attributes the emission to bound–bound pathways. These formulations are not mutually exclusive; rather, they indicate that BTH is a regime label based on emitted photon energy, while the microscopic mechanism depends on laser wavelength, intensity, ellipticity, and electronic structure. This suggests that “below threshold” is spectroscopic rather than mechanistic terminology.

Several characteristic signatures recur. First, discrete resonances can strongly enhance specific harmonic orders. Second, harmonic energies can track intensity-dependent Stark shifts. Third, the polarization and phase of BTH can exhibit features absent from standard plateau HHG, including harmonic ellipticity exceeding that of the driver, phase alternation by $\pi$ between successive odd orders, and mirrored polarization in aligned few-level systems. Fourth, in multi-electron atoms, particular below-threshold orders can be strongly suppressed rather than enhanced when electron–electron coupling opens a competing non-radiative channel (Emelin et al., 2024).

2. Microscopic mechanisms and theoretical descriptions

A central theoretical framework for BTH is the time-dependent Schrödinger equation (TDSE). For a single-active-electron description in one dimension, one study uses

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$

with

$V(x)=-\frac{1}{\sqrt{x^2+b}},$

where $b=0.4371\,\mathrm{a.u.}$ models He, and

$E(t)=f(t)\,E_0\cos(\omega t)$

with a 14-cycle flat-top envelope. In the multiphoton regime for helium driven by an elliptically polarized field, the TDSE is written

$$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$

with

$$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$

where $f(t)$ is a trapezoidal envelope with 3-cycle turn-on, 30-cycle flat top, and 3-cycle turn-off. In that work, both a two-dimensional soft-core model and a three-dimensional model potential are used, and the laser intensity is fixed at $I=10^{14}\,\mathrm{W/cm^2}$ so that the ionization probability remains only a few percent (Emelin et al., 2024).

Multi-electron BTH requires an explicitly correlated or effective many-electron treatment. A two-electron one-dimensional TDSE employs

$$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$

with

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$0

where

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$1

Here $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$2 corresponds to full coupling and $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$3 to the single-active-electron limit. The same study also uses full three-dimensional TDDFT with time-dependent Kohn–Sham equations,

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$4

where

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$5

The HHG spectrum is then computed from the dipole acceleration or equivalently the time derivative of the dipole,

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$6

These formulations place BTH at the intersection of few-level dynamics, SAE modeling, and genuinely correlated electronic motion (Zhu et al., 2015).

A further abstraction is the few-level model. In a driven two-level system (TLS) with bare states $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$7 and $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$8, the field-free Hamiltonian is

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$9

and the dipole coupling is

$V(x)=-\frac{1}{\sqrt{x^2+b}},$0

Diagonalization yields field-dressed energies

$V(x)=-\frac{1}{\sqrt{x^2+b}},$1

This model isolates purely discrete-state physics and predicts phase and polarization effects specific to the BTH regime. A plausible implication is that few-level descriptions are especially useful when bound–bound resonances dominate and continuum mediation is secondary (Schöpa et al., 28 Jul 2025).

3. Resonant enhancement, Stark tuning, and sub-cycle dynamics

A particularly detailed mechanism for BTH is provided by the transient ac Stark-shifted resonance picture. For an atomic level $V(x)=-\frac{1}{\sqrt{x^2+b}},$2, the transient second-order ac Stark shift is

$V(x)=-\frac{1}{\sqrt{x^2+b}},$3

and for the first excited state $V(x)=-\frac{1}{\sqrt{x^2+b}},$4 and ground state $V(x)=-\frac{1}{\sqrt{x^2+b}},$5 the Stark-shifted gap is

$V(x)=-\frac{1}{\sqrt{x^2+b}},$6

with $V(x)=-\frac{1}{\sqrt{x^2+b}},$7. When an electron is resonantly excited at time $V(x)=-\frac{1}{\sqrt{x^2+b}},$8 and recombines at $V(x)=-\frac{1}{\sqrt{x^2+b}},$9, the emitted photon energy is

$b=0.4371\,\mathrm{a.u.}$0

Because $b=0.4371\,\mathrm{a.u.}$1, the leading-order shift satisfies

$b=0.4371\,\mathrm{a.u.}$2

For harmonics H23 to H27 near $b=0.4371\,\mathrm{a.u.}$3 under a $b=0.4371\,\mathrm{a.u.}$4 laser, the photon energies shift linearly with intensity around $b=0.4371\,\mathrm{a.u.}$5, and the yields are resonantly enhanced at $b=0.4371\,\mathrm{a.u.}$6 (Wang et al., 2019).

The associated trajectory picture parallels Corkum’s three-step model but replaces tunnel ionization with resonant excitation. At $b=0.4371\,\mathrm{a.u.}$7, near the peak of $b=0.4371\,\mathrm{a.u.}$8, the transiently Stark-shifted gap matches $b=0.4371\,\mathrm{a.u.}$9 and the transition $E(t)=f(t)\,E_0\cos(\omega t)$0 occurs by absorption of $E(t)=f(t)\,E_0\cos(\omega t)$1 photons. The electron remains bound rather than propagating in the continuum, but acquires a dipole phase. At a later sub-cycle time $E(t)=f(t)\,E_0\cos(\omega t)$2, when $E(t)=f(t)\,E_0\cos(\omega t)$3, the Stark shift is smaller, the gap is larger, and recombination $E(t)=f(t)\,E_0\cos(\omega t)$4 emits a photon of energy approximately

$E(t)=f(t)\,E_0\cos(\omega t)$5

The process is therefore intrinsically sub-cycle and nonperturbative, despite remaining below threshold. The main differences from plateau HHG are the absence of free-electron excursion in the continuum, the use of a bound excited state, and sensitivity to the transient rather than cycle-averaged Stark shift (Wang et al., 2019).

In the multiphoton regime, resonant enhancement is instead formulated through field-free or weakly shifted multiphoton resonances. For helium at $E(t)=f(t)\,E_0\cos(\omega t)$6, energy conservation including dynamic Stark shifts is written

$E(t)=f(t)\,E_0\cos(\omega t)$7

Because dynamic Stark shifts are small at this intensity, observed resonances fall very close to field-free multiphoton conditions. In the three-dimensional model, a clear 4-photon resonance with the $E(t)=f(t)\,E_0\cos(\omega t)$8 manifold occurs at $E(t)=f(t)\,E_0\cos(\omega t)$9, a 4-photon resonance with $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$0 states emerges at $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$1, and six-photon resonances with $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$2 and $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$3 appear at lower $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$4 and $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$5, corresponding to peaks in the 5th-harmonic yield. This contrast between transient Stark-controlled resonances and near-field-free multiphoton resonances indicates that BTH spans both strongly dressed and weakly dressed bound-state dynamics, depending on the parameter regime (Emelin et al., 2024).

4. Multi-electron suppression and the quenching effect

Below-threshold harmonics in multi-electron atoms can be suppressed rather than enhanced. The quenching effect is a pronounced suppression of particular low-order harmonics when multi-electron interaction is included. The key condition is an orbital-energy gap,

$$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$6

where $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$7 is the highest-occupied valence orbital and $$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$8 the next unoccupied orbital. The suppressed harmonic order is therefore

$$i\,\partial \psi(\mathbf r,t)/\partial t = \left[ p^2/2 + V(\mathbf r)-\mathbf r\cdot \mathbf E(t) \right]\psi(\mathbf r,t),$$9

Numerically, $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$0 at $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$1, giving quenching of the 7th harmonic in Ar; similar behavior is reported for the 5th harmonic in Xe (Zhu et al., 2015).

The suppression mechanism is formulated as competition between two recombination routes. In Route 1, radiative recombination emits a photon:

$$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$2

In Route 2, the recombination energy is transferred non-radiatively to another electron:

$$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$3

followed by

$$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$4

Because Route 2 is resonant when the returning electron energy matches the real orbital gap, it can dominate over radiative emission and deplete the population available for harmonic generation. The work explicitly describes this as analogous to fluorescence quenching owing to energy transfer between molecules in fluorescent material. This suggests that missing below-threshold lines can encode correlated recombination dynamics rather than an absence of microscopic excitation (Zhu et al., 2015).

The numerical evidence is explicit. TDDFT calculations for Ar, Kr, and Xe driven at $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$5, $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$6, with an 8-cycle trapezoidal pulse show deep dips of $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$7–$$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$8 orders of magnitude around the 7th harmonic in Ar and Kr and the 5th in Xe. Single-electron references exhibit no suppression. In a one-dimensional two-electron TDSE at $$\mathbf E(t)=E_0 f(t)\,[\cos(\omega t)\,\hat{\mathbf x}+\epsilon\sin(\omega t)\,\hat{\mathbf y}],$$9, the spectrum shows a $f(t)$0–$f(t)$1 order-of-magnitude dip at the 7th harmonic. When the interaction is scaled as $f(t)$2, the quenching grows continuously as $f(t)$3 and disappears at $f(t)$4. Scans over $f(t)$5–$f(t)$6 and $f(t)$7–$f(t)$8 show that the suppressed order remains fixed at $f(t)$9, indicating an atomic origin rather than a laser-field artifact (Zhu et al., 2015).

Anomalous ellipticity dependence provides an additional diagnostic. Ordinarily HHG yield falls exponentially with ellipticity $I=10^{14}\,\mathrm{W/cm^2}$0 of the driving field, but at the quenched harmonic the intensity initially rises as $I=10^{14}\,\mathrm{W/cm^2}$1 increases from $I=10^{14}\,\mathrm{W/cm^2}$2 and only then falls at large $I=10^{14}\,\mathrm{W/cm^2}$3. The stated explanation is that electron–electron transfer is less efficient when the electron misses the core. In experimental interpretation, this matters because missing BTH lines can otherwise be misassigned to propagation or phase-matching loss. The quenching framework therefore places electron correlation at the center of below-threshold spectral structure.

5. Polarization, ellipticity, phase, and spatial structure

BTH polarization can differ qualitatively from the driving field. In helium driven by an elliptically polarized laser, there exists a set of laser frequencies at which harmonic generation efficiency increases dramatically with simultaneous increase of harmonic ellipticity, and the harmonic ellipticity can exceed the laser ellipticity by 2 or more times in absolute value. The relevant selection rules are most transparent in the circular basis. For an $I=10^{14}\,\mathrm{W/cm^2}$4-photon transition to magnetic quantum number $I=10^{14}\,\mathrm{W/cm^2}$5, absorption of $I=10^{14}\,\mathrm{W/cm^2}$6 $I=10^{14}\,\mathrm{W/cm^2}$7 photons and $I=10^{14}\,\mathrm{W/cm^2}$8 $I=10^{14}\,\mathrm{W/cm^2}$9 photons gives

$$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$0

so only states with $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$1 can participate. The excitation rate scales schematically as

$$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$2

Because different $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$3 channels have different polarization properties, one channel can dominate and produce harmonic ellipticity exceeding $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$4. For the 3rd harmonic at the 1s$$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$53d four-photon resonance, the reported net harmonic ellipticity is $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$6 for $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$7, in excellent agreement with the full TDSE spectra, and in the two-dimensional model at $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$8 the 3rd-harmonic ellipticity reaches $$i\,\partial \Psi(x_1,x_2,t)/\partial t = \hat H\,\Psi(x_1,x_2,t),$$9 while $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$00. Peak yields at resonance depend only weakly on $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$01, unlike nonresonant plateau harmonics (Emelin et al., 2024).

The phase of BTHs can also exhibit threshold-specific structure. In a driven TLS, the dipole expectation is

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$02

with Fourier component

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$03

For harmonics with $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$04, the phase alternates by $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$05 between successive odd harmonic orders, whereas for $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$06 it locks to a constant value. The phenomenological expression is

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$07

Thus $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$08, $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$09, $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$10 below threshold, while above-threshold harmonics share the same phase. In a four-level model composed of two uncoupled TLS subsystems aligned along orthogonal directions, this phase structure produces a polarization change: below both resonances, harmonics follow the driver polarization,

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$11

whereas above one threshold but below the other,

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$12

corresponding to mirrored polarization about one molecular axis. The model predicts analogous behavior for aligned systems with orthogonal transition dipoles (Schöpa et al., 28 Jul 2025).

Spatial structuring of BTH can be understood from the optics of vector polarization modes. In isotropic media below threshold for multiphoton ionization, rotational averaging yields harmonic intensity

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$13

and, after summing over output polarizations and introducing the local ellipticity $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$14,

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$15

For a Poincaré beam constructed by superposing two co-propagating Laguerre–Gaussian modes of opposite topological charge $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$16 with orthogonal polarization bases, the local ellipticity varies azimuthally with Schoenflies $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$17 symmetry. For an $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$18-derived input with

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$19

the $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$20th-harmonic transverse intensity becomes

$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$21

The number of bright filaments is $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$22, the ring radius is $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$23, and the angular spacing is $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$24. For $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$25 and $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$26, the azimuthal FWHM estimate gives $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$27 and arc-width $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$28. These results show that BTH is not only spectrally and temporally structured but can also be engineered into crown-like arrays of sub-wavelength filaments (Andrews et al., 2019).

6. Control parameters, observables, and implications

Intensity is a primary control parameter. Around $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$29 in the $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$30 helium calculation, H23–H27 exhibit both a linear red-shift with increasing intensity and resonant enhancement of yield. The H25 yield peaks sharply at $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$31, where the Stark-shifted gap exactly equals $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$32 at the excitation time. Neighboring H23 and H27 appear as side peaks separated by two-photon “steps.” Quantitatively, the H25 yield at $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$33 is $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$34 in arbitrary units, roughly twice its off-resonant value, and exceeds H23/H27 by a factor $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$35–$$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$36. The linear shift $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$37 provides a continuous tuning knob for sub-threshold photon energies over several tenths of an eV (Wang et al., 2019).

Frequency and ellipticity are equally consequential in the multiphoton regime. For helium driven at $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$38 with $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$39, sharp local maxima in the 3rd, 5th, and 7th harmonic yields occur at resonant frequencies, and at these peaks the harmonic ellipticity exceeds the driving ellipticity by factors of two or more. The peak yields depend only weakly on $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$40, which contrasts sharply with the strong ellipticity suppression of plateau harmonics. This indicates that resonant BTH can remain efficient even at high ellipticity and thereby supports coherent VUV-range radiation with tunable polarization. The stated applications include circular dichroism spectroscopy of chiral molecules and ultrafast magnetization dynamics (Emelin et al., 2024).

Ellipticity also functions as a diagnostic for competing microscopic channels. In the quenching problem, the anomalous increase of a quenched harmonic at small nonzero ellipticity is a stated “smoking-gun” for electron–electron quenching, because the non-radiative transfer pathway weakens when recollision misses the core. Suggested strategies to detect or mitigate quenching include varying ellipticity, using single-valence-electron ions, applying two-color or polarization-shaped fields to steer the return energy off resonance with $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$41, and employing pump–probe schemes to populate or deplete the $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$42 orbital before HHG. These proposals indicate that BTH spectroscopy is sensitive not only to resonance positions but also to correlated-state occupancy and recollision geometry (Zhu et al., 2015).

Phase and polarization measurements provide further observables. The few-level aligned-molecule model predicts that low orders should track the driver polarization angle $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$43, whereas higher orders can track $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$44 once one subsystem crosses threshold and the other remains below it. Interferometric phase measures such as RABBIT or two-source interferometry are proposed to reveal the $$i\hbar\,\frac{\partial \psi(x,t)}{\partial t} = \left[ -\frac{\hbar^2}{2m}\frac{\partial^2}{\partial x^2} + V(x) + x\,E(t) \right]\psi(x,t),$$45-jumps in the BTH region. More broadly, the accumulated results imply that below-threshold harmonics probe excited-state structure, symmetry, and transient dressing with a sensitivity that differs from plateau HHG. A plausible implication is that BTH should be treated as a distinct spectroscopic channel whose observables—yield, photon energy, phase, ellipticity, and spatial mode structure—are each governed by different combinations of bound-state resonance, dynamic dressing, and multi-electron coupling (Schöpa et al., 28 Jul 2025).