Angular-Recoil Backaction: Effects and Applications

• Angular-Recoil Backaction is the phenomenon where angular momentum exchanges from optical, mechanical, and particle scattering actively feed back on emitters and their environments.
• It spans diverse settings from modified optical forces via Green function corrections and photon-recoil in levitated systems to kinematic effects in dark-matter and nuclear processes.
• The topic integrates theoretical frameworks and computational methods, revealing actionable insights that impact force modulation, signal interpretation, and even astrophysical recoil dynamics.

Angular-recoil backaction denotes the class of effects in which angular-momentum exchange—optical spin or orbital angular momentum, mechanical rotation, recoil-angle kinematics, or radiated angular momentum flux—feeds back on the emitter, scatterer, background medium, or inferred observable. In the cited literature, the term appears in several technically distinct but structurally related settings: Green-function corrections to optical forces on Rayleigh particles, photon-recoil and spin-curl modifications in structured photonics, self-generated chiral rotation in whispering-gallery optomechanics, recoil-conditioned angular spectra in molecular photoionization and directional dark-matter detection, proton-photon correlations in neutron radiative beta decay, and the evolution of mass and angular momentum in analogue and astrophysical black-hole recoil problems (Abbassi et al., 2018, Hatifi, 22 May 2026, Sayyad et al., 23 Mar 2025, Shan, 2022).

1. Definition and conceptual scope

In nonfree-space optics, “backaction” means the modification of the field at the particle’s position caused by scattering of the particle’s own radiation from surrounding structures such as mirrors, substrates, or nano-holes. The key conceptual point is that the light’s linear and angular momentum are not exchanged only between the particle and a fixed incident field; instead the environment participates, radiatively feeds back on the particle, and modifies gradient, radiation-pressure, and spin-curl forces (Abbassi et al., 2018).

Within that optical framework, angular recoil refers to changes in spin-related forces and, by extension, torques when the environment redistributes spin and possibly orbital angular momentum. The 2018 optical-force formulation does not compute torques explicitly, but it keeps the spin-curl term in full generality and therefore captures linear forces associated with angular-momentum exchange at the force level (Abbassi et al., 2018).

A second usage appears in levitated optomechanics, where random photon recoils are treated as measurement back-action. There, the quantity $\kappa_i$ is the dimensionless kinematic coefficient that determines the energy transferred per photon recoil into motional axis $i$, and therefore measures back-action per detected photon. In that language, tailoring the angular distribution of scattering changes angular recoil and hence changes back-action itself (Sayyad et al., 23 Mar 2025).

A third usage is kinematic rather than dynamical. In directional dark-matter scattering, recoil angles feed back on the mapping from the halo velocity distribution to the observed recoil spectrum. The joint distribution $P(v,\eta \mid Q)$ is not separable, so angular recoil information changes the effective one-dimensional speed distribution inferred in a narrow recoil-energy window (Shan, 2022).

In weak-interaction recoil observables, the same logic appears as proton-photon correlations in neutron radiative beta decay: the emitted photon changes the proton recoil momentum and angular asymmetry, so the radiation field backacts on the recoil degree of freedom itself (Ivanov et al., 2013).

A distinct and explicitly speculative formulation proposes an “extended dynamical equation” $m a_i = F_i^{ext} + m\tau \dot a_i$ and interprets inertial backreaction through angular momentum flux threading a surface. This usage is conceptually separate from the electromagnetic and scattering formalisms above, but it likewise casts recoil as a response to angular-momentum surges (Pinheiro et al., 2012).

2. Electromagnetic dipoles, Green functions, and optical recoil force

For a Rayleigh particle at position $\mathbf{r}_p$, the induced dipole is written as

$$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$

with an effective dyadic polarizability

$$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$

Here $\mathbf{G}_s$ is the scattering dyadic Green function, so backaction enters as self-interaction via the environment. Even for a spherical particle, $\boldsymbol{\alpha}$ becomes a position-dependent tensor because $\mathbf{G}_s$ is anisotropic and spatially varying (Abbassi et al., 2018).

The total local field is

$i$0

and the corresponding time-averaged force becomes

$i$1

The first three terms are the modified gradient force, radiation pressure, and spin-curl force. The fourth term,

$i$2

is new and stems from the spatial variation of the backaction field itself. In free space, $i$3, so this term vanishes (Abbassi et al., 2018).

The paper’s central optical claim is therefore not merely that backaction renormalizes $i$4, but that it introduces an explicit force term proportional to $i$5. In the circular nano-hole example, the main discrepancy between a conventional perturbative dipole approximation and the full Maxwell-stress-tensor result is attributed to this term rather than to higher multipoles (Abbassi et al., 2018).

At the dipolar level, recoil can also be written directly in source variables. For a time-harmonic electric-magnetic dipole, the time-averaged recoil force and recoil torque are

$i$6

These terms exist even in the absence of external illumination. The 2025 dipolar-recoil analysis re-derives the recoil force from the total Lorentz force on a set of charges and traces its ultimate origin to retardation effects arising from the finite speed of light in the mutual interactions between charges (Golat et al., 1 Dec 2025).

That paper also emphasizes a symmetry distinction: the recoil force is odd under time reversal, whereas the interaction force is even. This does not mean the microscopic Lorentz force law is time-odd; rather, the recoil term emerges after imposing retarded, rather than advanced, boundary conditions (Golat et al., 1 Dec 2025).

3. Structured photonics and optomechanical angular instability

In whispering-gallery optomechanics, the optical modes are two counterpropagating WGMs, $i$7 and $i$8, with angular dependence $i$9. A localized movable scatterer at angle $P(v,\eta \mid Q)$0 couples them through

$P(v,\eta \mid Q)$1

Each circulation-changing event transfers angular momentum $P(v,\eta \mid Q)$2 to the scatterer, and the optical torque is

$P(v,\eta \mid Q)$3

For reciprocal bidirectional pumping, the torque vanishes at rest, but rotation Doppler-shifts the two opposite scattering rates in opposite directions. For suitable detuning, this produces negative angular friction, destabilizes the nonrotating state, and selects one of two symmetry-related steady rotations. The threshold intracavity photon number scales inversely with the square of the WGM azimuthal index, $P(v,\eta \mid Q)$4, and the chiral state produces a direction-dependent weak-probe response as a Doppler splitting of the backscattered spectra (Hatifi, 22 May 2026).

In one-dimensional single-photon scattering from a trapped two-level system, recoil backaction appears as coupling between the internal transition and quantized center-of-mass motion through

$P(v,\eta \mid Q)$5

with Lamb–Dicke parameter

$P(v,\eta \mid Q)$6

In the Lamb–Dicke limit, the reflection spectrum has a single resonance at $P(v,\eta \mid Q)$7. When $P(v,\eta \mid Q)$8 grows, phonon sidebands at $P(v,\eta \mid Q)$9 appear, reflection becomes multi-peaked, and the outgoing photon becomes entangled with the motional state. The paper interprets this as recoil backaction of the motional scatterer on the photon’s coherent transport (Li et al., 2013).

Collective angular recoil also arises in free-space superradiant scattering with orbital-angular-momentum light. For atoms on a transverse ring pumped by an $m a_i = F_i^{ext} + m\tau \dot a_i$0 mode, the single-atom torque obeys

$m a_i = F_i^{ext} + m\tau \dot a_i$1

while the adiabatically eliminated multimode equations generate collective azimuthal forces proportional to $m a_i = F_i^{ext} + m\tau \dot a_i$2. The uniform azimuthal phase distribution is unstable; atoms are set in rotation and bunch in phase at harmonics determined by $m a_i = F_i^{ext} + m\tau \dot a_i$3 and the ring radius, with collective growth rates $m a_i = F_i^{ext} + m\tau \dot a_i$4 (Gisbert et al., 2021).

These three realizations share a common structure: angular momentum is exchanged one scattering event at a time, but in the driven-dissipative or collective setting the recoil feeds back on the mode structure itself and can therefore destabilize the nominally reciprocal state.

4. Recoil per scattering event: levitated nanoparticles and hard-x-ray molecules

For an optically levitated dielectric nanosphere, the recoil-heating rate along axis $m a_i = F_i^{ext} + m\tau \dot a_i$5 is

$m a_i = F_i^{ext} + m\tau \dot a_i$6

The coefficient $m a_i = F_i^{ext} + m\tau \dot a_i$7 is the energy transferred per photon recoil into axis $m a_i = F_i^{ext} + m\tau \dot a_i$8. In the Rayleigh limit, a plane-wave calculation gives the canonical coefficients

$m a_i = F_i^{ext} + m\tau \dot a_i$9

The 2025 Mie-resonance study generalizes $\mathbf{r}_p$0 to arbitrary incident and scattered angular distributions and shows that near Kerker conditions, where the scattered wavevector aligns with the incident one, the energy transferred per recoil event can be strongly reduced. For a linearly polarized Gaussian beam at $\mathbf{r}_p$1 nm, $\mathbf{r}_p$2 is reduced by more than $\mathbf{r}_p$3 and $\mathbf{r}_p$4 by more than $\mathbf{r}_p$5 near the first Kerker-like radii for silicon and diamond spheres. The paper interprets this as intrinsic suppression of back-action per scattering event (Sayyad et al., 23 Mar 2025).

A related but much higher-energy mechanism appears in molecular-frame photoelectron angular distributions of $\mathbf{r}_p$6 at $\mathbf{r}_p$7 keV. There the photon momentum is approximately $\mathbf{r}_p$8 a.u., the photoelectron momentum approximately $\mathbf{r}_p$9 a.u., and the recoil vector is

$$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$0

The measured forward-backward asymmetry is the expected nondipole effect, but the observed asymmetry with respect to the polarization direction cannot be explained by the initial bound state or final continuum state alone. Simulations assign it to recoil imposed on the nuclei by the fast photoelectrons and high-energy photons, producing a propensity for the ions to break up along the axis of the recoil momentum. For maximum transverse recoil, the angular momentum is about $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$1, and the molecule can rotate by about $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$2 during the $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$3 fs Auger lifetime. The measured consequence is that photoelectrons are predominantly emitted in the direction of the forward nitrogen atom (Kircher et al., 2021).

These two cases show two different meanings of “angular-recoil backaction.” In levitated optomechanics it is back-action per photon, quantified by $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$4. In hard-x-ray molecular photoionization it is recoil-driven reorientation of the fragmentation axis, which then feeds back onto the measured angular distribution itself.

5. Directional recoil observables in nuclear and weak processes

For elastic WIMP–nucleus scattering, the recoil energy in terms of the angle $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$5 between the recoil direction and the incoming WIMP velocity is

$$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$6

or equivalently,

$$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$7

The recoil-angle differential cross section used in the 2022 analysis is

$$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$8

Because the factor $$\mathbf{p} = \boldsymbol{\alpha}\cdot \mathbf{E}_0(\mathbf{r}_p),$$9 vanishes at $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$0 and $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$1, maximal-energy-transfer configurations are kinematically allowed but not statistically dominant. The effective WIMP speed distribution in a narrow recoil-energy window is therefore not just the halo speed distribution truncated at $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$2; it is reweighted by the allowed angular range and the form factors. The resulting effective speed distribution,

$$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$3

is the paper’s explicit example of angular recoil feeding back on spectral inference (Shan, 2022).

The 2021 Monte Carlo study of directional dark-matter detection simulates the full angular recoil-direction and recoil-energy distributions for $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$4, $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$5, and $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$6 targets. It shows that form-factor suppression strongly reshapes the incoming-WIMP-frame recoil-angle distributions, especially for Ge and Xe, and that in Equatorial coordinates the hot spot of recoil flux is shifted relative to the theoretical WIMP-wind direction, whereas the hot spot of average recoil energy is much closer to that direction. In the companion annual-modulation study, the all-sky average value of the average recoil energy is largest in the advanced summer and smallest in the advanced winter for all considered laboratories and all three target–WIMP mass combinations (Shan, 2021, Shan, 2021).

In neutron radiative beta decay,

$$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$7

the photon momentum modifies the proton recoil,

$$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$8

The 2013 calculation shows that explicit proton-photon correlations are required for recoil observables. They do not invalidate the standard inclusive radiative-correction functions $$\boldsymbol{\alpha} = \left[ \mathbf{I} -\frac{i k_0^3}{6\pi\epsilon_0}\boldsymbol{\alpha}_0 -\boldsymbol{\alpha}_0\cdot \mathbf{G}_s(\mathbf{r}_p,\mathbf{r}_p) \right]^{-1}\cdot \boldsymbol{\alpha}_0.$$9 and $\mathbf{G}_s$0 for the neutron lifetime and integrated proton recoil spectrum, but they do contribute to the proton recoil asymmetry $\mathbf{G}_s$1 at order $\mathbf{G}_s$2. The corrected proton angular distribution becomes symmetric in the radiative part proportional to $\mathbf{G}_s$3, restoring the same $\mathbf{G}_s$4 symmetry found at tree level (Ivanov et al., 2013).

Across these examples, the general lesson is the same: recoil angles are not passive labels. Once form factors, radiation, or selection windows are included, they actively reweight rates, asymmetries, and inferred source distributions.

6. Gravitational, analogue, and mechanical backreaction

In binary-black-hole mergers, recoil is the reaction to anisotropic gravitational-wave emission. For equal-mass, spinning binaries with partial spin–orbital-angular-momentum alignment, the “hangup-kick” effect amplifies the out-of-plane recoil beyond the pure superkick value. A phenomenological fit for the dominant kick amplitude is

$\mathbf{G}_s$5

with $\mathbf{G}_s$6, $\mathbf{G}_s$7, and $\mathbf{G}_s$8. For $\mathbf{G}_s$9, the fit predicts a maximum recoil of $\boldsymbol{\alpha}$0 at $\boldsymbol{\alpha}$1, larger than the earlier superkick estimate. Using accretion-motivated spin distributions, the paper concludes that the probability that a remnant black hole receives a total recoil exceeding the $\boldsymbol{\alpha}$2 escape velocity of large elliptical galaxies is ten times larger than the probability of observing a line-of-sight velocity above $\boldsymbol{\alpha}$3, and that the direction of large recoils is strongly peaked toward the angular momentum axis (Lousto et al., 2012).

An analogue-fluid realization replaces the Kerr background by a rotating draining bathtub flow. There, backreaction is not a small local correction but a significant global change in the background parameters, and the system exhibits a memory encoded in the total mass of the fluid. The shallow-water effective metric is

$\boldsymbol{\alpha}$4

so time dependence in $\boldsymbol{\alpha}$5 and $\boldsymbol{\alpha}$6 produces a dynamical metric. The wave-induced mass change is governed by

$\boldsymbol{\alpha}$7

while the background angular-momentum density is

$\boldsymbol{\alpha}$8

This is the analogue-gravity version of angular-recoil backaction: scattering waves modify the mass and angular-momentum content of the rotating background, and the effective horizon geometry evolves accordingly (Patrick et al., 2019).

A more speculative mechanical analogue appears in the proposal

$\boldsymbol{\alpha}$9

where $\mathbf{G}_s$0 is a “critical action time.” The same paper introduces an angular momentum flux

$\mathbf{G}_s$1

and torque law

$\mathbf{G}_s$2

In that framework, angular recoil is interpreted as an induced angular momentum opposing sudden changes of motion, by analogy with Lenz’s law. This usage remains distinct from the optical, nuclear, and gravitational formalisms discussed above (Pinheiro et al., 2012).

7. Algorithmic recoil backaction and formal caveats

Angular-recoil backaction also has an algorithmic meaning in perturbative QCD. In the Herwig7 angular-ordered shower, the evolution variable for the first emission satisfies

$\mathbf{G}_s$3

After a second emission, these equalities cannot all remain true simultaneously, so one must choose a recoil scheme. The proceedings compare three possibilities: transverse-momentum preserving, virtuality preserving, and dot-product preserving (Ravasio et al., 2019).

In the transverse-momentum-preserving scheme, each emission obeys

$\mathbf{G}_s$4

and recoil is absorbed by increasing the parent virtuality,

$\mathbf{G}_s$5

This preserves the factorized soft-collinear structure of well-separated emissions but tends to overpopulate hard tails in event-shape distributions (Ravasio et al., 2019).

In the virtuality-preserving scheme, one keeps

$\mathbf{G}_s$6

fixed, which forces the first emission’s transverse momentum to change according to

$\mathbf{G}_s$7

This can make $\mathbf{G}_s$8 negative, so in practice one must set $\mathbf{G}_s$9 in those regions. The formal consequence is that recoil backaction from a later emission can erase the physical transverse momentum of an earlier one, degrading the logarithmic accuracy of the angular-ordered shower (Ravasio et al., 2019).

The dot-product-preserving scheme is constructed to keep earlier soft emissions stable: $i$00 In the soft limit this reduces to $i$01, so the backaction of later emissions is parametrically suppressed. A phase-space veto

$i$02

then removes overly large virtualities without changing the soft-collinear region. The authors conclude that recoil is not a subleading implementation detail but part of the logarithmic structure itself (Ravasio et al., 2019).

This computational example clarifies a broader point already visible in the physical literature: whenever angular ordering is enforced in one set of variables but recoil reshuffles a different set, the recoil prescription feeds back on the effective angular structure. In optics this can create new force terms proportional to $i$03; in optomechanics it can destabilize a reciprocal state; in directional detection it changes the inferred velocity distribution; and in shower algorithms it changes the formal accuracy of the resummation.