Two-Photon Collisional Shielding

• Two-photon collisional shielding is a method for ultracold polar molecules that uses two coherent radiation fields to create repulsive adiabatic potentials and cancel long-range dipolar attraction.
• It employs microwave or Raman optical schemes to dress rotational states, enabling precise tuning of elastic collisions and significant suppression of two- and three-body losses.
• The technique leverages adiabatic following and coherent state mixing to engineer both barrier formation and interaction compensation, with studies reporting up to a tenfold reduction in reactive loss.

Searching arXiv for the cited papers and closely related work on two-photon collisional shielding. Two-photon collisional shielding is a family of field-dressed collision-control schemes for ultracold polar molecules in which two coherent radiation fields engineer repulsive long-range adiabatic potentials that prevent colliding pairs from reaching lossy short range. In the microwave realization, the central mechanism is the two-photon component of double microwave shielding, where two microwave fields of different frequencies and polarizations jointly dress rotational states so that the colliding pair adiabatically follows a repulsive potential while long-range dipolar attraction can be canceled (Karman et al., 14 Jan 2025). In the optical realization, two-photon optical shielding uses a Raman $\Lambda$ scheme between ground rotational levels, mediated by an off-resonant electronically excited state, to generate analogous repulsive barriers while suppressing spontaneous emission through dark-state physics (Karam et al., 2022). Across these realizations, the central objective is the same: simultaneous suppression of short-range loss and retention of controllable elastic interactions.

1. Historical emergence and conceptual scope

Two-photon collisional shielding emerged from attempts to extend microwave collisional shielding of polar molecules into regimes with stronger control over loss channels and greater technical flexibility. A decisive development was double microwave shielding, which recently enabled evaporative cooling to the first Bose–Einstein condensate of polar molecules and was formulated as a two-field dressing problem in which two-photon processes during collisions are intrinsic to the dynamics (Karman et al., 14 Jan 2025). A parallel line of work mapped microwave shielding onto a two-photon Raman transition in the optical domain, explicitly targeting the suppression of photon scattering that limited one-photon optical shielding (Karam et al., 2022).

The term therefore covers two closely related architectures. In one, two microwaves of $\sigma^+$ and $\pi$ polarization dress the rotational ground manifold and engineer a repulsive short-range core together with compensation of the long-range dipolar interaction. In the other, two cw lasers or two coherent lasers in a Raman $\Lambda$ configuration couple ground rotational levels through an electronically excited intermediate and create an avoided crossing between attractive and repulsive pair potentials. The two architectures differ in microscopic implementation, but both rely on dressed-state adiabaticity, avoided crossings, and control of long-range anisotropic interactions.

A persistent theme in this literature is that shielding is not only a two-body barrier problem. The double-microwave analysis emphasizes that attractive long-range pockets can support field-linked bound states that mediate three-body recombination, so a shielding scheme can suppress two-body short-range access and still remain unstable against many-body loss unless those long-range bound states are removed (Karman et al., 14 Jan 2025). Optical Raman studies add a complementary point: suppression of short-range flux is not sufficient if the shielding mechanism itself creates substantial photon scattering or inelastic branching (Karam et al., 1 Aug 2025).

2. Microscopic mechanism of repulsive shielding

In the microwave framework, a rigid-rotor molecule with rotational constant $B_{\mathrm{rot}}$ and permanent dipole moment $d$ has lowest rotational states $\lvert J=0\rangle$ and $\lvert J=1\rangle$ separated by approximately $2B_{\mathrm{rot}}$. A near-resonant microwave with Rabi frequency $\Omega$ and detuning $\sigma^+$0 couples $\sigma^+$1 either to $\sigma^+$2 for $\sigma^+$3 polarization or to $\sigma^+$4 for $\sigma^+$5 polarization, producing dressed eigenstates that are superpositions of rotational and photon-number states. For single-field blue detuning, molecules prepared in the upper dressed state acquire synchronized oscillating or rotating dipoles whose time-averaged dipole–dipole interaction does not vanish. In the $\sigma^+$6 case near the $\sigma^+$7 transition, this interaction is repulsive in the relevant pair channel and creates an effective shield (Karman et al., 14 Jan 2025).

Double microwave shielding extends this by applying one $\sigma^+$8 field and one $\sigma^+$9 field with distinct frequencies $\pi$0 and detunings $\pi$1. The resulting upper dressed state contains both $\pi$2- and $\pi$3-dressed amplitudes. During a collision, the pair can exchange photons between the two dressing fields, with Floquet channels separated by the beat frequency $\pi$4; the theory identifies these channels as the dominant residual inelastic pathway under double shielding (Karman et al., 14 Jan 2025). The critical advantage is that the coherent superposition of the two dressing components allows the net long-range dipolar interaction to be tuned in sign and magnitude, including complete compensation.

That compensation is the essential collisional-shielding feature. Outside the repulsive core, the long-range dipolar interaction can be canceled, eliminating shallow long-range wells that otherwise support field-linked bound states. The short-range repulsive core remains, so the pair is blocked from accessing lossy short-range channels, while the long-range molecular pair no longer supports the bound states that mediate three-body recombination. This suggests that two-photon collisional shielding is best understood not merely as barrier formation, but as simultaneous engineering of the barrier and the asymptotic topology of the dressed pair spectrum.

In the optical Raman formulation, two-photon shielding follows the same logic in a different internal-state architecture. Two optical fields couple ground-state rotational levels of the electronic ground manifold through a far-detuned intermediate electronic state. The dipole–dipole interaction mixes pair states and creates a crossing of attractive and repulsive long-range PECs at a distance $\pi$5. Raman coupling opens an avoided crossing there; for blue two-photon detuning, the upper dressed branch becomes repulsive and acts as a reflection barrier (Karam et al., 2022). When the two-photon detuning is set to zero for isolated molecules, the scheme prepares a dark state at $\pi$6, suppressing intermediate-state population and spontaneous emission.

3. Hamiltonian structure and dressed-state formulations

The minimal single-molecule Hamiltonian for the microwave scheme is

$\pi$7

with the near-resonant Rabi frequency and detuning defined by

$\pi$8

In the rotating-wave basis $\pi$9, the effective three-level Hamiltonian is

$\Lambda$0

Its upper eigenvector $\Lambda$1 determines the dressed-state decomposition (Karman et al., 14 Jan 2025).

For two colliding molecules in the center-of-mass frame, the pair Hamiltonian is

$\Lambda$2

with anisotropic dipole–dipole interaction. For static dipoles polarized along $\Lambda$3,

$\Lambda$4

Under single-field resonant dressing,

$\Lambda$5

and off resonance the magnitude is reduced by $\Lambda$6 (Karman et al., 14 Jan 2025).

The double-field theory also yields an analytic second-order effective repulsive potential,

$\Lambda$7

which captures the leading $\Lambda$8 repulsion in addition to the $\Lambda$9 dipolar term and closely reproduces numerically computed s-wave adiabatic potentials near and away from compensation (Karman et al., 14 Jan 2025).

In the optical Raman treatment for NaK, the full dressed Hamiltonian for two molecules under two laser fields is written in a dressed molecular$B_{\mathrm{rot}}$0Fock basis and then reduced by adiabatic elimination of the electronically excited manifold when $B_{\mathrm{rot}}$1. The resulting effective three-block Hamiltonian contains AC Stark shifts and second-order Raman couplings of the form $B_{\mathrm{rot}}$2, generating the effective two-photon coupling between ground blocks (Karam et al., 1 Aug 2025). In the earlier two-channel optical-shielding formulation, adiabatic elimination yields

$B_{\mathrm{rot}}$3

and

$B_{\mathrm{rot}}$4

with adiabatic eigenpotentials

$B_{\mathrm{rot}}$5

These are the optical analogues of the microwave dressed potentials (Karam et al., 2022).

4. Adiabatic dynamics, nonadiabaticity, and loss suppression

Shielding requires the colliding pair to remain on the repulsive dressed adiabat. In microwave shielding, this is analyzed directly in terms of the adiabatic curves of the pair Hamiltonian. Single $B_{\mathrm{rot}}$6-field dressing exhibits narrowly avoided crossings that lead to nonadiabatic loss, whereas $B_{\mathrm{rot}}$7-only shielding removes these narrow crossings and enables strong two-body suppression. In double-field shielding at compensation, the initial s-wave channel is purely repulsive and exhibits no narrow avoided crossings for all collision orientations; fixed-orientation cuts show that both head-to-tail and side-by-side approaches are shielded (Karman et al., 14 Jan 2025).

The standard Landau–Zener estimate is

$B_{\mathrm{rot}}$8

or equivalently,

$B_{\mathrm{rot}}$9

depending on notation (Karman et al., 14 Jan 2025, Karam et al., 2022). Double-field microwave shielding reduces $d$0 both by removing narrow avoided crossings and by increasing the local gap through the $d$1 repulsive core. In the optical Raman setting, efficient shielding likewise requires adiabatic following at the long-range avoided crossing, so large effective Raman coupling and ultracold relative velocity are favorable (Karam et al., 2022).

The central observable for two-body loss in the microwave scheme is

$d$2

Coupled-channels calculations at $d$3 nK show that near compensation, $d$4 for double-field shielding can be lower than the single-field minimum and is only weakly sensitive to small ellipticities. Residual two-body loss is dominated by Floquet inelastic channels in which photons are exchanged between the $d$5 and $d$6 fields; these channels lie only a few MHz below threshold, set by $d$7, unlike the larger energy release characteristic of single-field loss (Karman et al., 14 Jan 2025). By expelling all two-body bound states from the initial adiabat over a multi-MHz window in $d$8, the same scheme suppresses three-body recombination because there are no long-range dimer states into which a third molecule can recombine.

Optical Raman shielding introduces a different balance of rates. In the NaK calculations of the far-detuned two-photon-assisted collision theory, the reactive rate in the absence of shielding is $d$9, while the minimum reactive rate in the strong-coupling diagonal region near $\lvert J=0\rangle$0 MHz and $\lvert J=0\rangle$1 MHz is $\lvert J=0\rangle$2, an approximately tenfold reduction (Karam et al., 1 Aug 2025). However, the same calculations show that inelastic rates can grow with coupling strength and can exceed elastic rates because of branching into dressed open channels at shifted energies. In the later strong-coupling EIT formulation, the shielding metric

$\lvert J=0\rangle$3

reaches $\lvert J=0\rangle$4, meaning elastic collisions are favored over inelastic and reactive ones by about a factor of two (Hovhannesyan et al., 12 Feb 2026).

These results define two distinct performance profiles. Double microwave shielding aims at simultaneous suppression of $\lvert J=0\rangle$5 and $\lvert J=0\rangle$6 together with tunable elastic interactions. Two-photon optical shielding, at least in current NaK studies, achieves reduced reactive loss and low excited-state participation, but its optimal regions remain narrower and more vulnerable to inelastic branching.

5. Interaction tuning, compensation, and universality

A defining feature of the microwave two-photon scheme is independent control over dipolar and contactlike interactions. The axial dipolar length is

$\lvert J=0\rangle$7

with an analogous in-plane component $\lvert J=0\rangle$8 generated by ellipticity. By varying $\lvert J=0\rangle$9 at fixed $\lvert J=1\rangle$0, the axial dipolar interaction can be tuned smoothly through compensation, $\lvert J=1\rangle$1, to large positive or negative values. The s-wave scattering length is extracted from the zero-energy $\lvert J=1\rangle$2 matrix as

$\lvert J=1\rangle$3

or, in the explicitly indexed form,

$\lvert J=1\rangle$4

Away from compensation, $\lvert J=1\rangle$5 exhibits resonances whenever the long-range well supports successive bound states; inside the no-bound-state window, $\lvert J=1\rangle$6 can be tuned positive $\lvert J=1\rangle$7 negative while keeping $\lvert J=1\rangle$8 suppressed (Karman et al., 14 Jan 2025).

The same paper characterizes the scheme as universal across many polar molecules with rotational spectra allowing $\lvert J=1\rangle$9 microwave coupling, moderate magnetic fields that decouple hyperfine spectators, and permanent dipoles ranging from approximately $2B_{\mathrm{rot}}$0 to $2B_{\mathrm{rot}}$1 D. NaCs, RbCs, NaK, NaRb, and KAg are presented as examples. Shielding performance improves with larger dipole moment $2B_{\mathrm{rot}}$2 and larger reduced mass $2B_{\mathrm{rot}}$3, while depending only weakly on $2B_{\mathrm{rot}}$4 because the relevant dressed-state spacings are set by $2B_{\mathrm{rot}}$5 and $2B_{\mathrm{rot}}$6 (Karman et al., 14 Jan 2025). This establishes two-photon microwave shielding as a broadly applicable interaction-engineering tool rather than a species-specific resonance effect.

Optical implementations also modify the scattering length, but through narrower quasi-resonant structures. In the strong-coupling EIT study of bosonic $2B_{\mathrm{rot}}$7Na$2B_{\mathrm{rot}}$8K, the real and imaginary parts of the s-wave scattering length show resonances as the single-photon detuning $2B_{\mathrm{rot}}$9 is scanned near two-photon resonance, consistent with long-range quasi-bound levels induced by the lasers (Hovhannesyan et al., 12 Feb 2026). A plausible implication is that optical two-photon shielding may be especially useful where narrow control of near-threshold resonances is desirable, although the currently reported shielding windows are considerably tighter than the multi-MHz no-bound-state windows emphasized in double microwave shielding.

6. Optical Raman realizations, practical constraints, and relation to other schemes

Two-photon optical shielding was introduced as a way to reproduce the effective two-channel Hamiltonian and dressed potentials of microwave shielding while replacing near-resonant excited-state coupling by far-detuned Raman coupling (Karam et al., 2022). For $\Omega$0Na$\Omega$1K, the representative implementation uses the $\Omega$2 linkage, with the A/b $\Omega$3 manifold furnishing the intermediate state. The earlier proposal estimated $\Omega$4–$\Omega$5 MHz, barrier heights of approximately $\Omega$6–$\Omega$7K, and off-resonant photon-scattering probability per collision below $\Omega$8 under suitable Raman conditions (Karam et al., 2022).

The more complete two-photon-assisted collision formalism expresses the dressed Hamiltonian for two colliding molecules and two laser fields in a symmetrized fully coupled basis, with partial waves up to $\Omega$9, $\sigma^+$00, and $\sigma^+$01, and with absorbing boundary conditions at short range to model universal reactive loss (Karam et al., 1 Aug 2025). In that model, adiabatic elimination is valid when $\sigma^+$02 and the two-photon detuning remains on the MHz scale; at smaller detuning, such as $\sigma^+$03 MHz at fixed $\sigma^+$04, the excited-manifold admixture grows and adiabatic elimination no longer reproduces the full dressed PECs accurately. The later strong-coupling EIT study instead focuses on the regime $\sigma^+$05, where the excited $\sigma^+$06 channel must be retained and shielding appears in narrow detuning windows a few MHz wide (Hovhannesyan et al., 12 Feb 2026).

Several common misconceptions are clarified by these results. One is that all optical shielding necessarily suffers from severe spontaneous emission. The Raman dark-state formulation directly contradicts that: when the single-photon detuning is large compared with the Rabi frequencies, the excited-state population scales as

$\sigma^+$07

and the photon-scattering rate as $\sigma^+$08, so dark-state operation suppresses scattering relative to one-photon optical shielding (Karam et al., 1 Aug 2025). Another misconception is that suppression of short-range two-body loss alone guarantees evaporative-cooling compatibility. Double microwave shielding shows that long-range field-linked bound states can still produce three-body recombination unless the long-range dipolar interaction is compensated (Karman et al., 14 Jan 2025).

In direct comparison with related methods, one-photon optical shielding produces repulsive barriers but continuously admixes an electronically excited state and therefore suffers from photon scattering and heating; microwave shielding avoids excited electronic states and has already enabled collisionally stable samples and evaporative cooling; two-photon optical shielding offers optical spatial control and reduced scattering, but in the NaK calculations reported so far it produces moderate shielding, narrow quasi-resonant features, and substantial sensitivity to inelastic channel structure (Karam et al., 2022, Hovhannesyan et al., 12 Feb 2026). This suggests that the present frontier of two-photon collisional shielding lies in choosing between complementary advantages: the broad and interaction-rich control of double microwave shielding, and the platform compatibility and dark-state optical flexibility of Raman shielding.