betaFeedback

WorksPhysics and Astronomy

Orbital angular momentum radiation and polarization of relativistic electrons in magnetic fields

Abstract

While spin polarization from synchrotron radiation is well established, the polarization of orbital angular momentum (OAM) in such radiative processes remains elusive. We study radiation and polarization of relativistic electrons in a uniform magnetic field, focusing on OAM polarization radiation for vortex electrons which carry intrinsic OAM. The results illustrate that transition rates are asymmetric in the low-photon-energy regime, favoring OAM decrease, analogous to the spin-flip asymmetry in the Sokolov-Ternov effect. Under these conditions, synchrotron radiation can polarize the OAM. The characteristic relaxation time and stationary-state OAM distribution are obtained analytically. The polarization of spin about Pspin\mathcal{P}_{\text{spin}} reaches 92.38%92.38\%, while that of POAM\mathcal{P}_{\text{OAM}} can even approach almost unity for a large OAM; however, their polarization behaviors are different. For typical storage ring parameters, the OAM polarization time is orders of magnitude shorter than the spin polarization time. Thus, synchrotron radiation offers a mechanism for controlling vortex electron beams which carry OAM for high-energy accelerator applications.

Introduction

Synchrotron radiation, the electromagnetic emission by relativistic charged particles undergoing centripetal acceleration in magnetic fields, constitutes a cornerstone of modern high-energy physics and accelerator science. The fully relativistic quantum theory was developed by Sokolov, Ternov, and others [1, 2, 3], who predicted the remarkable effect of spontaneous spin polarization, whereby an electron beam circulating in a storage ring can become polarized up to ∼92.38%\sim92.38\% with spins antiparallel to the magnetic field. This Sokolov–Ternov effect, which has been extensively studied [4, 5], arises from the spin-dependent part of the synchrotron radiation rate and has become an essential tool for producing polarized beams in experiments [6, 7].

It should be noted that the Sokolov–Ternov spin polarization limit of 92.38% is obtained under the assumption of uniform and constant magnetic fields, where realistic accelerator effects, such as field inhomogeneities, orbital oscillations, and quantum fluctuations associated with discrete photon emission, are not explicitly taken into account.

The original formulation solved the Dirac equation for an electron in a uniform vertical magnetic field, obtaining the asymptotic polarization 8/(53)≈92.38%8/(5\sqrt{3}) \approx92.38\%, independent of energy [8]. In practical storage rings, however, the magnetic fields are inhomogeneous and particles traverse a distribution of orbits, which results in the depolarizing spin resonances that do not appear in the simple uniform-field model [9]. If the magnetic environment is artificially held constant but the electron energy is large, the achievable spin polarization Pspin\mathcal{P}_{\mathrm{spin}} is generally less than 92.38%, and this discrepancy can be substantial due to the onset of such resonances [10]. Furthermore, when the electron energy is not held constant but instead decreases because of radiative losses, the final polarization also falls below the theoretical maximum, a phenomenon first quantitatively addressed by Derbenev and Kondratenko, whose formalism incorporates the spin-diffusion processes arising from quantum fluctuations of the radiation [11]. In high-energy storage rings, depolarizing resonances (particularly those of a nonlinear nature) are the dominant factors limiting the actual polarization degree, driving it far below 92.38%. Meanwhile, it is worth noting that the use of Siberian snakes in high-energy storage rings provides a nontrivial spin precession which enables longitudinal polarization at interaction points while substantially suppressing depolarization [12].

Following the idealized framework established by Sokolov and Ternov, which captures the fundamental spin dynamics in a uniform magnetic field, the present work pursues a similarly focused investigation of the underlying quantum-electrodynamic processes for orbital angular momentum (OAM). The additional complexities encountered in practical storage rings, such as depolarizing resonances and energy fluctuations, are not within the scope of this study.

Vortex beams, characterized by helical phase fronts and a well-defined OAM, represent a general class of structured wavefields applicable to both light and matter waves. Optical vortex beams have been extensively investigated, leading to the development of structured light and a variety of applications in areas such as optical manipulation and quantum information processing[13, 14, 15, 16, 17]. The extension of the OAM concept to matter waves has enabled the realization of vortex electron beams carrying quantized OAM [18, 19, 20, 21, 22, 23, 24, 25, 26]. In contrast to photons, electrons are charged massive particles, and their OAM is directly coupled to external electromagnetic fields. This feature gives rise to distinct magnetic and dynamical properties, and has motivated a range of studies in electron microscopy, fundamental quantum physics and high energy nuclear and particle physics[27, 28, 29, 30]. One essential aspect of vortex particles is the characterization of their OAM degree of freedom and the associated polarization. In particular, the evolution of orbital angular momentum in external electromagnetic fields becomes a nontrivial problem for charged particle beams [31]. This is especially relevant in accelerator for high energy vortex electron beams, where the interplay between OAM, beam dynamics, and polarization effects remains to be systematically understood.

While spin polarization from synchrotron radiation is well understood, the polarization of OAM as a dynamical degree of freedom in such radiative processes is still elusive. Recent theoretical works suggest that electrons in a uniform magnetic field can occupy Landau states labeled not only by the principal quantum number nn and spin projection ζ\zeta, but also by a discrete OAM quantum number ℓ\ell [32, 33, 34]. This raises the fundamental question of whether synchrotron radiation, just as it polarizes spin, also induces a net polarization in the OAM of an electron beam. Addressing this question requires a unified quantum-electrodynamic treatment, which simultaneously accounts for both spin and OAM degrees of freedom. To our knowledge, this task has not been fully accomplished.

In this article, we present a comprehensive theoretical analysis of radiation and polarization phenomena for relativistic electrons in a uniform magnetic field, with particular emphasis on OAM effects. Starting from the Dirac wave function in a magnetic field, the explicit expressions for transition probabilities and radiation intensities involving both spin and OAM are derived. Using the Wentzel–Kramers–Brillouin (WKB) approximation for large quantum numbers, the analytical results are obtained which reveal how synchrotron radiation polarizes electron beams not only in spin but also in OAM. Our calculations show that for low-photon-energy emission, the OAM transition rates are asymmetric, favoring transitions that reduce the OAM quantum number, thereby driving the electron beam toward a state of net OAM polarization. The formulas for the relaxation times and the polarization dynamics of spin and OAM are provided.

Throughout the paper, we use Gaussian units unless otherwise noted. Quantum states before and after a radiative transition are denoted by unprimed and primed quantum numbers, respectively (e.g., n,ℓ,ζn,\ell,\zeta for the initial state and n′,ℓ′,ζ′n',\ell',\zeta' for the final state). The paper is organized as follows. In Sec. II, the Dirac wave function for an electron in a uniform magnetic field is presented, highlighting the quantum numbers associated with spin and OAM; then the general formalism for the transition amplitude and radiation probability is developed, including the treatment of polarization states.

The WKB approximation for the relevant radial matrix elements is introduced here, which leads to analytical expressions for the OAM-transition rates in Sec. III. The resulting polarization dynamics are analyzed in Sec. IV, where the rate equations for spin and OAM distributions are solved. Finally, the implications of our findings are discussed and summarized in Sec. V. Useful integrals and some details are collected in the Appendices. Our findings bridge the well-established theory of spin polarization with the emerging physics of vortex electron beams, offering new insights into the angular-momentum structure of synchrotron radiation and its potential for generating and controlling structured charged-particle beams.

A Review of Quantum Radiation Theory

Radiation of electron in magnetic field

The eigenfunctions of the Dirac equation for an electron in a uniform, static magnetic field are presented here. This solution, well-established in the literature [3], incorporates the spin degree of freedom through a conserved polarization operator aligned with the field. The magnetic field is taken to be along the positive zz-axis, H=(0,0,H)\mathbf{H}=(0,0,H), with the corresponding vector potential in the symmetric gauge A=(−yH/2,xH/2,0)\mathbf{A}=(-yH/2,xH/2,0). The Hamiltonian H\mathcal{H} and the commuting polarization operator μ3\mu_3 are

H=cα⋅P+ρ3m0c2,P=p−ecA,(1)\mathcal{H}=c\boldsymbol{\alpha}\cdot\mathbf{P}+\rho_3m_0c^2,\qquad\mathbf{P}=\mathbf{p}-\frac{e}{c}\mathbf{A}, \tag*{(1)}
μ3=σ3+ρ2(σ×P)3m0c,with k0μ3ψ=K0ζψ.(2)\mu_3=\sigma_3+\rho_2\frac{(\boldsymbol{\sigma}\times\mathbf{P})_3}{m_0c},\qquad\text{with } k_0\mu_3\psi=K_0\zeta\psi. \tag*{(2)}

Here, e=−∣e0∣e=-|e_0| is the electron charge, EE is the electron energy, m0m_0 is the electron rest mass, cc is the speed of light, k0=m0c/ℏk_0=m_0c/\hbar, K=E/(ℏc)K=E/(\hbar c), and K0=K2−k32K_0=\sqrt{K^2-k_3^2}. The symbols α\boldsymbol{\alpha} and ρi\rho_i denote the Dirac matrices, and σ\boldsymbol{\sigma} represents the four-dimensional Pauli matrices. The eigenvalue ζ=±1\zeta=\pm1 corresponds to the electron spin being parallel or antiparallel to the magnetic field, respectively. The wave function solution, Ψ(r,t)=ψ(r)e−iEt/ℏ\Psi(\mathbf{r},t)=\psi(\mathbf{r})e^{-iEt/\hbar}, is separable in cylindrical coordinates (r,ϕ,z)(r,\phi,z), and the separated functions satisfy Eq. (1) and Eq. (2),

ψ(r)=ψℓ,k3(r,ϕ,z)f.(3)\psi(\mathbf{r})=\psi_{\ell,k_3}(r,\phi,z)\mathbf{f}. \tag*{(3)}

The spatial part is characterized by the OAM quantum number ℓ\ell (which is associated with the zz-component OAM, a notation that is adopted here) and the longitudinal wave number k3k_3,

ψℓ,k3(r,ϕ,z)=1L12πeik3zei(ℓ−1/2)ϕ,(4)\psi_{\ell,k_3}(r,\phi,z)=\frac{1}{\sqrt{L}}\frac{1}{\sqrt{2\pi}}e^{ik_3z}e^{i(\ell-1/2)\phi}, \tag*{(4)}

where LL is a normalization length. The four-component spinor ff is given in Eq. (75), from which the radial dependence is encoded in the functions In,s(ρ)I_{n,s}(\rho), defined as

In,s(ρ)=(s!n!)1/2e−ρ/2ρ(n−s)/2Lsn−s(ρ).(5)I_{n,s}(\rho)=\left(\frac{s!}{n!}\right)^{1/2}e^{-\rho/2}\rho^{(n-s)/2}L_s^{n-s}(\rho). \tag*{(5)}

where γ=e0H/(2cℏ)\gamma=e_0H/(2c\hbar), ρ=γr2\rho=\gamma r^2, Lsa(ρ)L_s^{a}(\rho) is the generalized Laguerre function. The quantum numbers are related by

n=ℓ+s,(n,s≥0;−∞<ℓ≤n),(6)n=\ell+s,\qquad(n,s\ge0;\quad-\infty<\ell\le n), \tag*{(6)}
E2=(m0c2)2+(ℏc)2(k32+4γn).E^2=(m_0c^2)^2+(\hbar c)^2(k_3^2+4\gamma n).

The integer nn is the principal quantum number, ss is the radial quantum number, and ℓ\ell is related to the OAM quantum number, which corresponds to the zz-component of the electron’s total angular momentum (TAM),

Jzψ=ℏ(ℓ−12)ψ.(7)J_z\psi=\hbar\left(\ell-\frac{1}{2}\right)\psi. \tag*{(7)}

where JzJ_z is the zz-component operator of the TAM.

The emission of synchrotron radiation by a relativistic electron in a uniform magnetic field is considered while the interaction with the quantized radiation field is introduced via the time-dependent potential UU. Starting from the Hamiltonian,

iℏ∂Ψ∂t=HΨ,H=cα⋅P+ρ3m0c2+U,(8)i\hbar\frac{\partial\Psi}{\partial t}=\mathcal{H}\Psi,\qquad\mathcal{H}=c\boldsymbol{\alpha}\cdot\mathbf{P}+\rho_3m_0c^2+U, \tag*{(8)}

the interaction Hamiltonian in the second-quantized form is in Eq. (78).

Applying time-dependent perturbation theory for the emission of a single photon (k,λ)(\mathbf{k},\lambda) [8], the transition amplitude from an initial electron state ψb\psi_b to a final state ψa\psi_a is in Eq. (79), so the transition probability per unit time follows as

wb→a=e022πℏ∫d3kκδ(Eb−Eaℏc−κ)∑λ∣α‾⋅ϵ(λ)∗∣2.(9)w_{b\to a}=\frac{e_0^2}{2\pi\hbar}\int\frac{d^3k}{\kappa}\delta\left(\frac{E_b-E_a}{\hbar c}-\kappa\right)\sum_\lambda\left|\overline{\boldsymbol{\alpha}}\cdot\boldsymbol{\epsilon}^{(\lambda)*}\right|^2. \tag*{(9)}

where α‾\overline{\boldsymbol{\alpha}} is the spatial matrix element,

α‾=∫d3r ψa†(r)αe−ik⋅rψb(r).(10)\overline{\boldsymbol{\alpha}}=\int d^3r\,\psi_a^\dagger(\mathbf{r})\boldsymbol{\alpha}e^{-i\mathbf{k}\cdot\mathbf{r}}\psi_b(\mathbf{r}). \tag*{(10)}

Here, k\mathbf{k} is the photon wavevector with magnitude κ=∣k∣\kappa=|\mathbf{k}|, ϵ(λ)\boldsymbol{\epsilon}^{(\lambda)} (λ=2,3)(\lambda=2,3) are the polarization unit vectors, the axial symmetry of the magnetic field along the zz-axis can be exploited to simplify the calculation. The photon wavevector in spherical coordinates is k=κ(sin⁡θcos⁡ϕ′,sin⁡θsin⁡ϕ′,cos⁡θ)\mathbf{k}=\kappa(\sin\theta\cos\phi',\sin\theta\sin\phi',\cos\theta), where θ\theta is the polar angle relative to the magnetic field. As the radiation intensity is independent of the azimuthal angle ϕ′\phi', one can therefore set ϕ′=π/2\phi'=\pi/2 for convenience, yielding k=κ(0,sin⁡θ,cos⁡θ)\mathbf{k}=\kappa(0,\sin\theta,\cos\theta).

The linear polarization basis vectors ϵ(2),ϵ(3)\boldsymbol{\epsilon}^{(2)},\boldsymbol{\epsilon}^{(3)} are chosen to align with this coordinate choice: ϵ(2)=ex\boldsymbol{\epsilon}^{(2)}=\mathbf{e}_x,

ϵ(3)=cos⁡θ ey−sin⁡θ ez\boldsymbol{\epsilon}^{(3)}=\cos\theta\,\mathbf{e}_y-\sin\theta\,\mathbf{e}_z. With this basis, the polarization sum in Eq. (9) for linear polarization simplifies to

∑λ=2,3∣α‾⋅ϵ(λ)∗∣2=∣α‾1∣2+∣α‾2cos⁡θ−α‾3sin⁡θ∣2≡Φ2+Φ3.(11)\sum_{\lambda=2,3}\left|\overline{\boldsymbol{\alpha}}\cdot\boldsymbol{\epsilon}^{(\lambda)*}\right|^2=|\overline{\alpha}_1|^2+|\overline{\alpha}_2\cos\theta-\overline{\alpha}_3\sin\theta|^2\equiv\Phi_2+\Phi_3. \tag*{(11)}

Here, α‾i\overline{\alpha}_i (i=1,2,3)(i=1,2,3) are components of the matrix element vector α‾\overline{\boldsymbol{\alpha}}.

The matrix elements α‾i\overline{\alpha}_i are evaluated using the Dirac wavefunctions given before. The initial and final states are labeled by their quantum numbers: b=(n,s,ℓ,ζ,k3)b=(n,s,\ell,\zeta,k_3) and a=(n′,s′,ℓ′,ζ′,k3′)a=(n',s',\ell',\zeta',k'_3), where ζ,ζ′=±1\zeta,\zeta'=\pm1 denote the spin projection. The spatial integrals in Eq. (10) separate into longitudinal (zz), angular (ϕ\phi), and radial (rr) parts.

The longitudinal integral yields momentum conservation along the field axis:

1L∫−L/2L/2dz e−iz(k3′+κcos⁡θ−k3)=δk3′,k3−κcos⁡θ.(12)\frac{1}{L}\int_{-L/2}^{L/2}dz\,e^{-iz(k'_3+\kappa\cos\theta-k_3)}=\delta_{k'_3,k_3-\kappa\cos\theta}. \tag*{(12)}

The angular and radial integrals involve the wavefunction components In,s(ρ)I_{n,s}(\rho) defined in Eq. (5). The matrix elements can be expressed in terms of the radial functions and the spinor coefficients of Eq. (76) (see Appendix A, Eqs. (93) and (94)).

Transitions in the high-electron-energy limit

Since the present analysis focuses on high-energy electrons, the case can be specialized to the high-energy, quasi-classical regime relevant for synchrotron radiation. The situation is considered where an electron in a positive-energy state (ϵ=+1\epsilon=+1) has vanishing initial longitudinal momentum (k3=0k_3=0) by an appropriate choice of the reference coordinate system. Furthermore, the conditions are used that the quantum numbers nn and n′n' are large, allowing the sum over the final radial quantum number s′s' to be performed via the sum rule:

∑s′=0∞Is′,s2(x)=1.(13)\sum_{s'=0}^{\infty}I_{s',s}^{2}(x)=1. \tag*{(13)}

In this frame, k3′=−κcos⁡θk'_3=-\kappa\cos\theta from Eq. (12). The transition probability per unit time, integrated over the photon energy, is given in Eq. (90). For large nn and n′n', it is convenient to replace the sum over the discrete change ν\nu by an integral: ∑ν→∫0ndν\sum_\nu\to\int_0^n d\nu. Transforming the integration variables from ν\nu to the dimensionless parameter yy via Eq. (84), Eq. (85), Eq. (86), the total transition probability of spin after considering OAM transitions takes the form

w(ζ,ζ′)=2716πce02e09Eξ0R2∫0∞y dy(1+ξ0y)4∫dΩ F.(14)w(\zeta,\zeta')=\frac{27}{16\pi}\frac{ce_0^2}{e_0^9E\xi_0R^2}\int_0^\infty\frac{y\,dy}{(1+\xi_0y)^4}\int d\Omega\,\mathcal{F}. \tag*{(14)}

and the corresponding total radiation intensity is

W(ζ,ζ′)=2716πce02ε09/2R2∫0∞y2 dy(1+ξ0y)5∫dΩ F.(15)W(\zeta,\zeta')=\frac{27}{16\pi}\frac{ce_0^2}{\varepsilon_0^{9/2}R^2}\int_0^\infty\frac{y^2\,dy}{(1+\xi_0y)^5}\int d\Omega\,\mathcal{F}. \tag*{(15)}

Here, R=n/γR=\sqrt{n/\gamma} is the effective orbital radius of the electron, ε0=k02/K2\varepsilon_0=k_0^2/K^2, ξ0=3γ/(K2ε03/2)\xi_0=3\gamma/(K^2\varepsilon_0^{3/2}), F=F(θ,y;ζ,ζ′)\mathcal{F}=\mathcal{F}(\theta,y;\zeta,\zeta') and G=G(θ,y;ζ,ζ′;s,s′)\mathcal{G}=\mathcal{G}(\theta,y;\zeta,\zeta';s,s') are derived from Φ\Phi and Eq. (13) after the variable change to yy. The formulas for OAM-specific transitions, w(ℓ,ℓ′)w(\ell,\ell') and W(ℓ,ℓ′)W(\ell,\ell'), are obtained by summing over final spin states and averaging over initial spins without the relation in Eq. (13), as we need to maintain ℓ\ell and ℓ′\ell' explicitly

w(ℓ,ℓ′)=2716πce02ε09/2Eξ0R2∫0∞y dy(1+ξ0y)4∫dΩ 12∑ζ,ζ′G.(16)w(\ell,\ell')=\frac{27}{16\pi}\frac{ce_0^2}{\varepsilon_0^{9/2}E\xi_0R^2}\int_0^\infty\frac{y\,dy}{(1+\xi_0y)^4}\int d\Omega\,\frac{1}{2}\sum_{\zeta,\zeta'}\mathcal{G}. \tag*{(16)}

and radiation intensity

W(ℓ,ℓ′)=2716πce02ε09/2R2∫0∞y2 dy(1+ξ0y)5∫dΩ 12∑ζ,ζ′G.(17)W(\ell,\ell')=\frac{27}{16\pi}\frac{ce_0^2}{\varepsilon_0^{9/2}R^2}\int_0^\infty\frac{y^2\,dy}{(1+\xi_0y)^5}\int d\Omega\,\frac{1}{2}\sum_{\zeta,\zeta'}\mathcal{G}. \tag*{(17)}

WKB approximation

The WKB method, which is usually used in Schrödinger equation, can now be applied to evaluate the functions In,n′(x)I_{n,n'}(x) and Is,s′(x)I_{s,s'}(x) defined in Eq. (5) for large quantum numbers n,n′,s,s′n,n',s,s'. These functions satisfy a second order differential equation of the form

d2dx2[x1/2y(x)]−fn,n′(x)x1/2y(x)=0,(18)\frac{d^2}{dx^2}\left[x^{1/2}y(x)\right]-f_{n,n'}(x)x^{1/2}y(x)=0, \tag*{(18)}

where y(x)=In,n′(x)y(x)=I_{n,n'}(x), and the function fn,n′(x)f_{n,n'}(x) is

fn,n′(x)=(n−n′)2−14x2−n+n′+12x+14.(19)f_{n,n'}(x)=\frac{(n-n')^2-1}{4x^2}-\frac{n+n'+1}{2x}+\frac{1}{4}. \tag*{(19)}

Equation (18) can be viewed as a Schrödinger-type equation with effective potential fn,n′(x)f_{n,n'}(x). The classical turning points x0x_0 and x0′x'_0 (x0<x0′x_0<x'_0) are the roots of fn,n′(x)=0f_{n,n'}(x)=0. For large quantum numbers, these are given by

x0≈(n−n′)2,x0′≈(n+n′)2.(20)x_0\approx(\sqrt{n}-\sqrt{n'})^2,\qquad x'_0\approx(\sqrt{n}+\sqrt{n'})^2. \tag*{(20)}

The argument x=κ2sin⁡2θ/(4γ)x=\kappa^2\sin^2\theta/(4\gamma) in our radiation problem, as defined in Eq. (89), is kinematically constrained. For the large quantum numbers n,n′n,n', it gets x≈x0sin⁡2θ≲x0x\approx x_0\sin^2\theta\lesssim x_0. Consequently, xx typically lies to the left of the first turning point x0x_0 and approaches it. In contrast, the functions Is,s′(x)I_{s,s'}(x) involving the radial quantum numbers do not share this constraint, so the argument can span the region between and beyond both its turning points. Therefore, a uniform asymptotic approximation valid across all xx is required. Following the standard WKB connection procedure in Eq. (96) [5], the solution In,n′(x)I_{n,n'}(x) is approximated differently in five regions

defined by the turning points, as illustrated in Figure 1, while Figure 2 demonstrates the agreement between the exact In,n′(x)I_{n,n'}(x) and its WKB approximation across different regions for representative quantum numbers.

Figure 1. Effective potential fn,n′(x)f_{n,n'}(x) (black solid line) from Eq. (19) for n=1000n=1000, n′=300n'=300. The roots x0x_0 and x0′x'_0 (red dots) are the classical turning points. The vertical dashed lines demarcate the five asymptotic regions (I–V) for the WKB solution given in Eq. (96).

Figure 2. Comparison of the exact function In,n′(x)I_{n,n'}(x) (blue solid line) with its uniform WKB approximation (red dashed line) from Eq. (96) for n=1000n=1000, n′=300n'=300. The vertical dotted lines indicate the boundaries between the asymptotic regions. The approximation accurately captures both the exponential decay (region I & V), the transitional Airy behavior (regions II & IV), and the oscillatory regime (region III).

The radiative polarization mechanism, i.e., Sokolov-Ternov effect in Appendix C, can drive the electron beam toward a high degree of spin antiparallel to the magnetic field, independent of the initial polarization. For GeV-energy electrons in typical storage ring magnetic fields of ∼1\sim1 T, the polarization time τspin\tau_{\mathrm{spin}} in Eq. (107) is on the order of hours, which makes this effect observable in experiments with prolonged beam circulation.

The formalism for OAM transitions mirrors the spin case but requires careful treatment of the radial quantum number sums. The key difference is that one cannot use the sum rule in Eq. (13) because the final OAM quantum number ℓ′\ell' must be retained, which constrains the relation between s′s' and ℓ′\ell' via n′=ℓ′+s′n'=\ell'+s'. The WKB approximation Eq. (96) must therefore be applied to both In,n′(x)I_{n,n'}(x) and Is,s′(x)I_{s,s'}(x) functions.

Radiation and Polarization in the Low-Photon-Energy Limit

Quasiclassical states for large quantum numbers

To facilitate the analysis of radiative transitions in the limit of large quantum numbers, the WKB approximation is employed here to evaluate the matrix elements In,n′(x)I_{n,n'}(x) and Is,s′(x)I_{s,s'}(x). The validity of this approximation requires the spacing between adjacent Landau levels to be much smaller than the electron energy, i.e., ωB≪E\omega_B \ll E [5]. For typical storage ring parameters (E∼1 GeVE \sim1\ \mathrm{GeV}, B∼1 TB \sim1\ \mathrm{T}), this condition is well satisfied. Moreover, since the energy of the emitted photon is directly tied to the change in the electron’s Landau level index, the regime of low-photon-energy naturally emerges as the relevant one for the WKB treatment, where the quantum number change ∣Δn∣|\Delta n| is small in this regime, which is consistent with the quasiclassical picture. The classical synchrotron spectrum peaks near the critical frequency ωc=32γ3c/R\omega_c = \frac{3}{2}\gamma^3 c/R [35], but quantum effects become significant when the photon energy approaches the electron energy. In the opposite limit of low-photon-energy (ℏω≪E\hbar\omega\ll E), the transition probabilities are dominated by processes with ∣Δn∣,∣Δℓ∣∼1|\Delta n|, |\Delta\ell| \sim1, and exhibit a systematic asymmetry that is central to the OAM polarization effect studied here. The combined considerations for the validity of WKB approximation and the spectral characteristics therefore lead us to focus on the low-photon-energy limit in the following analysis.

The regime of interest is where the radial quantum numbers ss and s′s' are large, while the OAM quantum numbers ℓ\ell and ℓ′\ell' are relatively small. Specifically, this is

∣ℓ∣n≪1,∣ℓ′∣n′≪1,∣ℓ−ℓ′∣n≪1,(21)\frac{|\ell|}{n}\ll1,\qquad\frac{|\ell'|}{n'}\ll1,\qquad\frac{|\ell-\ell'|}{n}\ll1, \tag*{(21)}

Under these conditions, the left turning point x0px_0^p for the function Is,s′(x)I_{s,s'}(x), defined as the root of corresponding potential fs,s′(x)=0f_{s,s'}(x)=0, can be expressed in terms of the principal quantum number turning point x0x_0 given in Eq. (20). Using the expansion s=n−ℓ≈n(1−ℓ/(2n))\sqrt{s}=\sqrt{n-\ell}\approx\sqrt{n}(1-\ell/(2n)), one obtains

x0p=(s−s′)2≈[(n−n′)−12(ℓn−ℓ′n′)]2.(22)\begin{aligned} x_0^p=(\sqrt{s}-\sqrt{s'})^2 \\ &\approx\left[(\sqrt{n}-\sqrt{n'})-\frac{1}{2}\left(\frac{\ell}{\sqrt{n}}-\frac{\ell'}{\sqrt{n'}}\right)\right]^2. \tag*{(22)} \end{aligned}

The difference n−n′\sqrt{n}-\sqrt{n'} is related to the spectral variable yy through the energy relation n/n′≈(K/K′)2=(1+ξ0y)2n/n'\approx(K/K')^2=(1+\xi_0y)^2 in Eq. (85). This yields

n−n′≈ξ0yn1+ξ0y.(23)\sqrt{n}-\sqrt{n'}\approx\frac{\xi_0y\sqrt{n}}{1+\xi_0y}. \tag*{(23)}

Consequently, the turning point ratio can be written as (ℓ≠ℓ′)(\ell\ne\ell')

x0px0≈[1+1+ξ0y ℓ′−ℓξ0y12n]2,(24)\frac{x_0^p}{x_0}\approx\left[1+\frac{1+\xi_0y\,\ell'-\ell}{\xi_0y}\frac{1}{2n}\right]^2, \tag*{(24)}

where the result has neglected terms of order ℓ/n\ell/n and used the low-photon-energy limit for the radiated photon, which implies yy is small. The kinematic constraint x=x0sin⁡2θx=x_0\sin^2\theta places the argument of Is,s′(x)I_{s,s'}(x) near its turning point x0px_0^p when the emitted photon energy is small. For the dominant contribution to OAM transitions, we focus on the region where yy is sufficiently small such that

y≤∣ℓ−ℓ′∣2nξ0.(25)y\leq\frac{|\ell-\ell'|}{2n\xi_0}. \tag*{(25)}

In this regime, the factor η=(x0/x0p)\eta=(x_0/x_0^p) satisfies η≥0\eta\geq0, and the quantity (1−x/x0p)(1-x/x_0^p) can be expressed in terms of (1−x/x0)(1-x/x_0) and η\eta. Using the relation (1−x/x0)≈(1+ξ0y)ε(1-x/x_0)\approx(1+\xi_0y)\varepsilon from Eq. (81), one can find

1−xx0p≈(1−xx0)1ε[1+η(ε−1)].(26)1-\frac{x}{x_0^p}\approx\left(1-\frac{x}{x_0}\right)\frac{1}{\varepsilon}[1+\eta(\varepsilon-1)]. \tag*{(26)}

In the WKB region II (near the first turning point), the function Is,s′(x)I_{s,s'}(x) is approximated by the modified Bessel function K1/3K_{1/3} as in Eq. (97):

Is,s′(x)≈1π3(1−xx0p)1/2K1/3(z′),(27)I_{s,s'}(x)\approx\frac{1}{\pi\sqrt{3}}\left(1-\frac{x}{x_0^p}\right)^{1/2}K_{1/3}(z'), \tag*{(27)}
z′=23((x0p)2ss′)1/4(1−xx0p)3/2.z'=\frac{2}{3}\left((x_0^p)^2ss'\right)^{1/4}\left(1-\frac{x}{x_0^p}\right)^{3/2}.

Using the relations, the argument z′z' can be written in terms of the variable zz from Eq. (81):

z′≈∣1+1+ξ0y ℓ′−ℓξ0y12n∣3/2[1+η(ε−1)]3/2z.(28)z'\approx\left|1+\frac{1+\xi_0y\,\ell'-\ell}{\xi_0y}\frac{1}{2n}\right|^{3/2}[1+\eta(\varepsilon-1)]^{3/2}z. \tag*{(28)}

For yy satisfying Eq. (25), the first factor is dominated by the OAM difference term:

z′≈c0∣y+1+ξ0y ℓ′−ℓξ012n∣[1+η(ε−1)]3/2,(29)z'\approx c_0\left|y+\frac{1+\xi_0y\,\ell'-\ell}{\xi_0}\frac{1}{2n}\right|[1+\eta(\varepsilon-1)]^{3/2}, \tag*{(29)}

where c0=(1/2)ε0−3/2c_0=(1/2)\varepsilon_0^{-3/2}. To verify that this approximation lies within the WKB region II, we check the condition x0p−δ1<x<x0p+x_0^p-\delta_1<x<x_0^p+ δ2\delta_2, where δ1,δ2\delta_1,\delta_2 are the transitional widths defined in Eq. (96). For large nn, this condition reduces to

η−5/6[(ξ0y)−1/6η−1/6−(nξ0y)−1/3]<(ξ0y)−1/6.(30)\eta^{-5/6}\left[(\xi_0 y)^{-1/6}\eta^{-1/6}-(n\xi_0 y)^{-1/3}\right]<(\xi_0 y)^{-1/6}. \tag*{(30)}

Combined with the upper bound from Eq. (25), this inequality is satisfied for typical parameters n≫1n\gg1 and ∣ℓ−ℓ′∣∼1|\ell-\ell'|\sim1, confirming the consistency of using the region-II approximation.

Asymptotic behavior in the low-photon-energy limit

The dominant contribution to OAM transitions in our case comes from the emission of soft photons, i.e., y→0y\to0. In this limit, the modified Bessel function Kμ(z′)K_\mu(z') can be approximated by its small-argument expansion,

Kμ(z′)≈Γ(μ)2(2z′)μ,(31)K_\mu(z')\approx\frac{\Gamma(\mu)}{2}\left(\frac{2}{z'}\right)^\mu, \tag*{(31)}

as illustrated in Figure 3. Here, Γ(μ)\Gamma(\mu) is the Gamma function. Substituting Eq. (29) into this expansion, we obtain explicit expressions for the two cases ℓ′<ℓ\ell'<\ell (OAM decrease) and ℓ′>ℓ\ell'>\ell (OAM increase):

Figure 3. Modified Bessel functions K1/3(z)K_{1/3}(z) and K2/3(z)K_{2/3}(z) (solid lines) compared with their small-argument expansions from Eq. (31) (dashed lines). The approximation is accurate for z≤0.1z\leq0.1.

Is,s′(x)≈Γ(1/3)2π3{(2−c0y+ℓ′−ℓ3)1/3,ℓ′<ℓ,(2c0y+ℓ′−ℓ3)1/3,ℓ′>ℓ.(32)I_{s,s'}(x)\approx\frac{\Gamma(1/3)}{2\pi\sqrt{3}} \begin{cases} \left(\dfrac{2}{-c_0y+\dfrac{\ell'-\ell}{3}}\right)^{1/3}, & \ell'<\ell,\\ \left(\dfrac{2}{c_0y+\dfrac{\ell'-\ell}{3}}\right)^{1/3}, & \ell'>\ell. \end{cases} \tag*{(32)}

The asymmetry between the two cases stems from the sign of OAM difference in the denominator of the argument. As shown in Figure 4, for the dominant contribution corresponding to the smallest nonzero change ∣ℓ−ℓ′∣=1|\ell-\ell'|=1, Eq. (32) simplifies further to

Figure 4. Comparison of the function Is,s′(x)I_{s,s'}(x) for different values of b=∣ℓ′−ℓ∣b=|\ell'-\ell|. The function is normalized such that its value is unity for b=1b=1 and y=0y=0. The dashed lines indicate the upper bounds of yy in Eq. (32) for the respective bb values. The yy is in units of 10−1010^{-10}.

Is,s′(x)≈Γ(1/3)2π3(2∣∓c0y+13∣)1/3,(33)I_{s,s'}(x)\approx\frac{\Gamma(1/3)}{2\pi\sqrt{3}}\left(\frac{2}{\left|\mp c_0y+\dfrac{1}{3}\right|}\right)^{1/3}, \tag*{(33)}

where the upper (lower) sign corresponds to ℓ′=ℓ−1\ell'=\ell-1 (ℓ′=ℓ+1\ell'=\ell+1). This expression captures the essential singularity as y→0y\to0, which governs the behavior of the OAM transition rates.

As the approximate form of Is,s′(x)I_{s,s'}(x) is established, one is now equipped to evaluate the transition probabilities and radiation intensities for OAM changes.

In the limit of low photon energy (y→0y\to0) and high electron energy (ϵ0≪1\epsilon_0\ll1), the spinor coefficients from Eq. (82) simplify to Eq. (83). Substituting these into the matrix element expressions Eqs. (87) and (88) yields our results.

The modified Bessel functions in the WKB approximation for the principal quantum number functions can be

expanded for small zz (i.e., small yy) using

K1/3(z)≈Γ(1/3)2(2z)1/3=A1y−1/3(ϵ0ϵ)1/2,(34)K_{1/3}(z)\approx\frac{\Gamma(1/3)}{2}\left(\frac{2}{z}\right)^{1/3}=A_1y^{-1/3}\left(\frac{\epsilon_0}{\epsilon}\right)^{1/2}, \tag*{(34)}
K2/3(z)≈Γ(2/3)2(2z)2/3=A2y−2/3(ϵ0ϵ),(35)K_{2/3}(z)\approx\frac{\Gamma(2/3)}{2}\left(\frac{2}{z}\right)^{2/3}=A_2y^{-2/3}\left(\frac{\epsilon_0}{\epsilon}\right), \tag*{(35)}

where A1=Γ(1/3)41/3/2A_1=\Gamma(1/3)4^{1/3}/2 and A2=Γ(2/3)42/3/2A_2=\Gamma(2/3)4^{2/3}/2. The argument zz is given by Eq. (81). In the small-yy limit, K2/3(z)≫K1/3(z)K_{2/3}(z)\gg K_{1/3}(z), so the combinations of II-functions in Eq. (80) lead to K2/3(z)K_{2/3}(z) dominating.

The transition rate for OAM changes requires the spin-averaged angular integral of Φ=Φ1+Φ2\Phi=\Phi_1+\Phi_2, where Φ1=∣αˉ1∣2\Phi_1=|\bar{\alpha}_1|^2 and Φ2=∣αˉ2cos⁡θ−αˉ3sin⁡θ∣2\Phi_2=|\bar{\alpha}_2\cos\theta-\bar{\alpha}_3\sin\theta|^2. After substituting the small-yy approximations and performing the angular integration over θ\theta and ϕ\phi, the dominant contribution arises from Φ1\Phi_1. The spin-averaged result is

12∑ζ,ζ′∫dΩ Φ1≈43πA22ϵ02y−4/3Is,s′2(x),(36)\frac{1}{2}\sum_{\zeta,\zeta'}\int d\Omega\,\Phi_1\approx\frac{4}{3\pi}A_2^2\epsilon_0^2y^{-4/3}I_{s,s'}^2(x), \tag*{(36)}

where dΩ=sin⁡θdθdϕd\Omega=\sin\theta d\theta d\phi. The integral Is,s′2(x)I_{s,s'}^2(x) is evaluated using the small-argument approximation from Eq. (33) for ∣ℓ′−ℓ∣=1|\ell'-\ell|=1 and is finally multiplied by the correction factor 1/31/3 because the variation is near 1/31/3 which is not following the small variation condition strictly. This gives

Is,s′2(x)≈13π2(12)2/3A12(1∣∓c0y+13∣)2/313,(37)I_{s,s'}^2(x)\approx\frac{1}{3\pi^2}\left(\frac{1}{2}\right)^{2/3}A_1^2\left(\frac{1}{\left|\mp c_0y+\dfrac{1}{3}\right|}\right)^{2/3}\frac{1}{3}, \tag*{(37)}

Combining Eqs. (36) and (37), we define the spectral function F(y)F(y) that appears in the integrand of the transition rate formula Eq. (16):

F(y)≡427π3(12)2/3A12A22ε02y−1/3(1∣c0y±1/3∣)2/3.(38)F(y) \equiv\frac{4}{27\pi^3}\left(\frac{1}{2}\right)^{2/3} A_1^2 A_2^2 \varepsilon_0^2 y^{-1/3}\left(\frac{1}{\lvert c_0y \pm1/3\rvert}\right)^{2/3}. \tag*{(38)}

Figure 5 compares this analytical approximation with the exact numerical evaluation of F(y)F(y), confirming its accuracy in the small-yy region.

Figure 5. Comparison of the approximate spectral function F(y)F(y) (approximation) and its integral with the exact counterparts for the case ℓ′=ℓ+1\ell'=\ell+1 (Δℓ=+1\Delta\ell=+1), with n=1016n=10^{16}, H=104H=10^4 Gauss. The approximation is accurate for y≤10−8y\leq10^{-8}, which suffices for y0≈10−10y_0\approx10^{-10}.

Transition rates and relaxation times

The transition rate for an OAM change Δℓ=ℓ′−ℓ=±1\Delta\ell=\ell'-\ell=\pm1, which gives the dominant contribution for the function, is obtained by integrating F(y)F(y) over yy up to the cutoff y0=∣Δℓ∣/(2nξ0)y_0=\lvert\Delta\ell\rvert/(2n\xi_0) from Eq. (25):

wΔℓ=±1=Q∫0y0dy F(y),(39)w_{\Delta\ell=\pm1}=Q\int_0^{y_0}dy\,F(y), \tag*{(39)}

where Q=2716ce02ε09/2Eξ0R2Q=\frac{27}{16}\frac{ce_0^2}{\varepsilon_0^{9/2}E\xi_0R^2} is the prefactor from Eq. (16). Performing the integral analytically yields

w+≡wΔℓ=+1=Q427π3A12A22ε03β+,(40)w_+\equiv w_{\Delta\ell=+1}=Q\frac{4}{27\pi^3}A_1^2A_2^2\varepsilon_0^3\beta_+, \tag*{(40)}
w−≡wΔℓ=−1=Q427π3A12A22ε03β−,(41)w_-\equiv w_{\Delta\ell=-1}=Q\frac{4}{27\pi^3}A_1^2A_2^2\varepsilon_0^3\beta_-, \tag*{(41)}

with the integration constants β+≈1.2310\beta_+\approx1.2310 and β−≈3.6276\beta_-\approx3.6276. The ratio of the rates is

w+w−≈0.3393,(42)\frac{w_+}{w_-}\approx0.3393, \tag*{(42)}

which indicates that transitions with decreasing OAM (Δℓ=−1\Delta\ell=-1) are approximately three times more probable than those that increase it (Δℓ=+1\Delta\ell=+1). Similar to Eq. (106), this asymmetry drives the OAM polarization of the electron beam.

For a numerical estimate, we consider typical synchrotron radiation parameters such that electron energy E≈1 GeVE\approx1\,\mathrm{GeV}, magnetic field B=1 TB=1\,\mathrm{T} (corresponding to H=104 GaussH=10^4\,\mathrm{Gauss}), and principal quantum number n≈1016n\approx10^{16}. The spin-conserving transition rate (no spin flip) is of order wno−flip∼109 s−1w_{\mathrm{no-flip}}\sim10^9\,\mathrm{s}^{-1}, while the spin-flip rates are much smaller, wflip∼10−5 s−1w_{\mathrm{flip}}\sim10^{-5}\,\mathrm{s}^{-1} to 10−4 s−110^{-4}\,\mathrm{s}^{-1} in Appendix C1. The OAM transition rates in our case when emitted photon energy is small for ∣Δℓ∣=1\lvert\Delta\ell\rvert=1 are intermediate, with w+≈1.67×104 s−1w_+\approx1.67\times10^4\,\mathrm{s}^{-1} and w−≈4.92×104 s−1w_-\approx4.92\times10^4\,\mathrm{s}^{-1}.

The relaxation dynamics of the spin and OAM distributions can be described by rate equations. For spin, the polarization evolves toward an equilibrium value with a characteristic time τspin\tau_{\mathrm{spin}} given by Eq. (107). Figure 6 shows the time evolution of the spin populations, where one can see a stationary-state population of 96.19% for N−1N_{-1}, which is antiparallel to the magnetic field. For OAM, the evolution depends on the initial ℓ0\ell_0 value, as will be detailed later.

Figure 6. Time evolution of the spin population fractions N1(t)N_1(t) (spin parallel to the magnetic field) and N−1(t)N_{-1}(t) (spin antiparallel), starting from an unpolarized beam. The dashed line indicates the final distribution. The system reaches an equilibrium polarization of approximately 92.38%, corresponding to a stationary-state population of 96.19%.

The analytical results presented here provide a quantitative description of how synchrotron radiation polarizes relativistic electron beams in OAM at low-photon-energy limit. The derived transition rates and relaxation times form the basis for predicting the polarization dynamics in high-energy storage rings and astrophysical environments.

OAM polarization dynamics

With the asymmetric transition rates for OAM-changing processes derived, the collective polarization dynamics of an ensemble of electrons under synchrotron radiation is now analyzed. This asymmetry, analogous to the spin-flip asymmetry in the Sokolov-Ternov effect [1, 3], suggests that repeated photon emissions will drive the electron beam toward an equilibrium stationary state with a net OAM polarization. However, this polarization behavior is very different from that of spin.

To describe this evolution quantitatively, one must first specify the space of OAM states accessible to the system. In contrast to atoms, where the TAM quantum number ll imposes a rigid bound ∣m∣≤l|m|\leq l on the projection mm, the Landau states in a uniform magnetic field allow, in principle, arbitrary integer values of the OAM quantum number ℓ\ell [36]. This reflects the fact that TAM and OAM are not conserved and that ℓ\ell encodes both the angular momentum projection and the guiding center position [37]. Consequently, the OAM space is formally infinite dimensional.

In practice, however, two considerations justify truncating this space to a finite range. First, as established in Sec. III B, radiative transitions dominantly couple only states with ∣Δℓ∣=1|\Delta\ell|=1. Thus, for a given initial distribution, the accessible OAM states are only those reachable via repeated ±1\pm1 steps. Second, for large ∣ℓ∣|\ell|, the corresponding classical orbits have guiding centers far from the axis [28]. Such states either have negligible overlap with the initial beam profile or are kinematically suppressed in the low-photon-energy regime where our analytical rates apply (see Eqs. (21) and (25)).

To better describe this polarization behavior, without loss of generality, one can therefore consider an ensemble of electrons whose OAM distribution is initially confined to a finite range [−ℓ0,ℓ0][-\ell_0,\ell_0], where ℓ0\ell_0 is the maximum allowed OAM projection in the beam, and define the relevant polarization parameters. This range is preserved under the dynamics because transitions only occur between adjacent ℓ\ell states. Let nm(t)n_m(t) denote the population of electrons in states with OAM quantum number ℓ=m\ell=m, where mm ranges from −ℓ0-\ell_0 to +ℓ0+\ell_0 in integer steps. The time evolution of these populations is governed by a set of coupled rate equations that account for the incoming and outgoing fluxes for each OAM state. For interior states (−ℓ0<m<ℓ0-\ell_0<m<\ell_0), the population changes due to incoming transitions from m+1m+1 and m−1m-1, and outgoing transitions from mm to these states. For the boundary states m=ℓ0m=\ell_0 and m=−ℓ0m=-\ell_0, only one neighboring state exists. This physical picture leads to the following system of equations [38]:

Figure 7. Time evolution of the OAM distribution fraction for different initial ℓ0\ell_0. The plots show the population fractions of states with ℓ=m\ell=m for m=−ℓ0,−ℓ0+1,…,ℓ0−1,ℓ0m=-\ell_0,-\ell_0+1,\ldots,\ell_0-1,\ell_0; the dashed line indicates the final distribution. The asymmetry between the transition rates w+w_+ and w−w_- leads to a net polarization toward lower ℓ\ell values. The fifth subplot in each set (a, b, c, d) shows the final distribution when the initial particle distribution is unchanged but the lowest accessible state is (−ℓ0−2)(-\ell_0-2).

dnℓ0dt=nℓ0−1w+−nℓ0w−,dnmdt=nm+1w−+nm−1w+−nm(w++w−),(−ℓ0<m<ℓ0),dn−ℓ0dt=n−ℓ0+1w−−n−ℓ0w+,(43)\begin{aligned} \frac{d n_{\ell_0}}{d t} &= n_{\ell_0-1}w_+ - n_{\ell_0}w_-,\\ \frac{d n_m}{d t} &= n_{m+1}w_- + n_{m-1}w_+ - n_m(w_+ + w_-), \qquad(-\ell_0 < m < \ell_0),\\ \frac{d n_{-\ell_0}}{d t} &= n_{-\ell_0+1}w_- - n_{-\ell_0}w_+, \tag*{(43)} \end{aligned}

Figure 8. POAM\mathcal{P}_{\mathrm{OAM}} and the fraction of the minimal mm state n−ℓ0n_{-\ell_0} vary with ℓ0\ell_0. For large ℓ0\ell_0, their values approach a constant.

together with the conservation of total particle number,

∑m=−ℓ0ℓ0nm(t)=n=constant.(44)\sum_{m=-\ell_0}^{\ell_0} n_m(t)=n=\text{constant}. \tag*{(44)}

The initial condition for an unpolarized OAM distribution is

nm(0)=n2ℓ0+1,m=−ℓ0,−ℓ0+1,…,ℓ0.(45)n_m(0)=\frac{n}{2\ell_0+1},\qquad m=-\ell_0,-\ell_0+1,\ldots,\ell_0. \tag*{(45)}

Equations (43)–(45) constitute a master equation for the OAM distribution, with w+w_{+} and w−w_{-} serving as the transition rates per electron for increasing and decreasing the OAM projection, respectively. The truncation at ±ℓ0\pm\ell_0 is not an algebraic necessity but a requirement for better physical observation of this polarization behavior, which reflects the finite initial OAM bandwidth of the beam and the fact that transitions outside this band are suppressed by the dynamics. As will be shown, the resulting stationary distribution decays exponentially toward lower ℓ\ell and, for sufficiently large ℓ0\ell_0, the behavior of polarization becomes insensitive to the precise value of ℓ0\ell_0.

The stationary-state solution of Eq. (43) is found to be

nm(st)=n1−μ1−μ2ℓ0+1μℓ0−m,μ≠1,(46)n_m^{(\mathrm{st})}=n\frac{1-\mu}{1-\mu^{2\ell_0+1}}\mu^{\ell_0-m},\qquad\mu\ne1, \tag*{(46)}
nm(st)=n2ℓ0+1,μ=1,(47)n_m^{(\mathrm{st})}=\frac{n}{2\ell_0+1},\qquad\mu=1, \tag*{(47)}

where μ≡w−/w+≈2.947\mu\equiv w_{-}/w_{+}\approx2.947 from Eq. (42). Since μ>1\mu>1, Eq. (46) describes an exponential distribution favoring states with lower mm, i.e. the beam becomes polarized toward the minimal OAM projection state m=−ℓ0m=-\ell_0. This is illustrated in Fig. 7 for various initial ℓ0\ell_0 values. For large ℓ0\ell_0, the stationary-state population of the minimal mm is 66.07%, and this value remains unchanged even when the lowest accessible state is shifted to (−ℓ0−2)(-\ell_0-2) while the initial distribution is unchanged. This indicates that, in this case, the polarization behavior toward lower mm tends to stabilize. The figure also shows that the equilibration time (in units of each τ\tau) increases with ℓ0\ell_0.

The relaxation toward the stationary state is governed by the smallest non-zero eigenvalue −λ1-\lambda_1 of the transition matrix derived from Eq. (43). The corresponding relaxation time is

τOAM=1λ1,(48)\tau_{\mathrm{OAM}}=\frac{1}{\lambda_1}, \tag*{(48)}

which depends on ℓ0\ell_0 and the ratio μ\mu. For GeV-energy electrons in typical storage ring magnetic fields of ∼1\sim1 T, the total polarization time, which is many times τOAM\tau_{\mathrm{OAM}}, is on the order of seconds to minutes and is significantly shorter than τspin\tau_{\mathrm{spin}}. A detailed derivation of λ1\lambda_1 and its asymptotic forms is provided in Sec. IV. The polarization dynamics of OAM thus exhibit a directional bias, with transitions to lower ℓ\ell states being more probable, leading to a net accumulation of electrons in states with near the minimal OAM projection. The OAM polarization is defined here analogously to the spin case, as in Eq. (C9),

POAM=∑mm⋅nm(∞)/nℓmin⁡=⟨ℓ⟩ℓ0.(49)\mathcal{P}_{\mathrm{OAM}}=\frac{\sum_m m\cdot n_m(\infty)/n}{\ell_{\min}}=\frac{\langle\ell\rangle}{\ell_0}. \tag*{(49)}

Where ⟨ℓ⟩\langle\ell\rangle is the average ℓ\ell of the electrons, ℓmin⁡\ell_{\min} denotes the minimum possible value of ℓ\ell. Fig. 8 shows that when ℓ0\ell_0 is large, POAM\mathcal{P}_{\mathrm{OAM}} approaches unity. For such ℓ0\ell_0, the three states nearest the minimal OAM projection (TNM) account for approximately 96.09% of the total population. In contrast to spin polarization, where the spin-flip asymmetry concentrates electrons almost entirely in a single spin state antiparallel to the field, the OAM polarization spreads over several low-ℓ\ell states. This difference arises because OAM transitions (∣Δℓ∣=1|\Delta\ell| = 1) involve multiple accessible states and a larger rate ratio w+/w−≈0.3393w_{+}/w_{-} \approx0.3393 (compared to 0.0396 for spin, Eq. (106)). Consequently, OAM polarization manifests as an enhanced population near the minimal projection rather than a collapse into a single quantum state. The relaxation time τOAM\tau_{\mathrm{OAM}} for OAM polarization varies with the ratio w+/w−w_{+}/w_{-}, as illustrated in Figure 9 for ℓ0=2\ell_{0} = 2. The unit of τOAM\tau_{\mathrm{OAM}} is 1/w−1/w_{-}.

Figure 9. Variation of the OAM relaxation time τOAM\tau_{\mathrm{OAM}} with the transition rate ratio w+/w−w_{+}/w_{-} for an initial state with ℓ0=2\ell_0=2. The relaxation time increases first and then decreases as the ratio varies.

OAM Polarization And Relaxation TIME

Stationary-state solution

One can consider the case where an electron has the maximum initial OAM projection ℓ0\ell_{0} and the range for m=−ℓ0,−ℓ0+1,…,ℓ0−1,ℓ0m = -\ell_{0}, -\ell_{0} + 1, \ldots, \ell_{0} - 1, \ell_{0}. Let nm(t)n_{m}(t) denote the population in state mm, with transition rates for w+w_{+} (m→m+1m \to m + 1, increasing OAM projection) and w−w_{-} (m→m−1m \to m - 1, decreasing OAM projection).

The evolution equations and the conservation of total particle number are given in Eqs. (43)–(45). At the stationary state, all time derivatives vanish. Denoting stationary state populations by nmstn_{m}^{\mathrm{st}},

0=nℓ0−1stw+−nℓ0stw−,(50)0 = n_{\ell_{0}-1}^{\mathrm{st}}w_{+} - n_{\ell_{0}}^{\mathrm{st}}w_{-}, \tag*{(50)}
0=n−ℓ0+1stw−−n−ℓ0stw+,(51)0 = n_{-\ell_{0}+1}^{\mathrm{st}}w_{-} - n_{-\ell_{0}}^{\mathrm{st}}w_{+}, \tag*{(51)}
0=nm−1stw++nm+1stw−−nmst(w++w−)(−ℓ0<m<ℓ0).(52)0 = n_{m-1}^{\mathrm{st}}w_{+} + n_{m+1}^{\mathrm{st}}w_{-} - n_{m}^{\mathrm{st}}(w_{+} + w_{-}) \qquad(-\ell_{0} < m < \ell_{0}). \tag*{(52)}

Define the asymmetry parameter:

μ≡w−w+.(53)\mu\equiv\frac{w_{-}}{w_{+}}. \tag*{(53)}

Assume a solution of the form nmst=Cχℓ0−mn_{m}^{\mathrm{st}} = C\chi^{\ell_{0}-m} in the simplest case, where CC is a normalization constant and χ\chi is a parameter to be determined. Substituting into Eq. (52) yields:

w+χ2−(w++w−)χ+w−=0.(54)w_{+}\chi^{2} - (w_{+} + w_{-})\chi+ w_{-} = 0. \tag*{(54)}

Dividing by w+w_{+} and using Eq. (53):

χ2−(1+μ)χ+μ=0.(55)\chi^{2} - (1 + \mu)\chi+ \mu= 0. \tag*{(55)}

The solutions are:

χ=(1+μ)±∣1−μ∣2⇒χ1=1,χ2=μ.(56)\chi= \frac{(1 + \mu) \pm|1 - \mu|}{2} \quad\Rightarrow\quad\chi_{1} = 1,\quad\chi_{2} = \mu. \tag*{(56)}

The general solution is a linear combination:

nmst=A+Bμℓ0−m.(57)n_{m}^{\mathrm{st}} = A + B\mu^{\ell_{0}-m}. \tag*{(57)}

Applying boundary conditions (50) and (51) gives:

A(1−μ)=0,(58)A(1-\mu) = 0, \tag*{(58)}

If μ≠1\mu\ne1, from Eq. (58), A=0A = 0. Equation (58) is then automatically satisfied. The solution is:

nmst=Bμℓ0−m.(59)n_{m}^{\mathrm{st}} = B\mu^{\ell_{0}-m}. \tag*{(59)}

Normalization via Eq. (44) determines BB:

B=n1−μ1−μ2ℓ0+1,(60)B = n\frac{1-\mu}{1-\mu^{2\ell_{0}+1}}, \tag*{(60)}

yielding the stationary-state distribution:

nmst=n1−μ1−μ2ℓ0+1μℓ0−m,μ≠1.(61)n_{m}^{\mathrm{st}} = n\frac{1-\mu}{1-\mu^{2\ell_{0}+1}}\mu^{\ell_{0}-m}, \qquad\mu\ne1. \tag*{(61)}

If μ=1\mu= 1 (symmetric case), equations (58) gives an constant. The general solution reduces to nmst=A+Bn_{m}^{\mathrm{st}} = A + B, and normalization gives:

nmst=n2ℓ0+1,μ=1.(62)n_{m}^{\mathrm{st}} = \frac{n}{2\ell_{0}+1}, \qquad\mu= 1. \tag*{(62)}

The relaxation time from eigenvalue analysis

The relaxation dynamics of the OAM polarization are governed by the linear system of differential equations and particle number conservation. In matrix form, these equations become:

dndt=Mn,(63)\frac{d\mathbf{n}}{dt} = M\mathbf{n}, \tag*{(63)}

where n(t)=(nℓ0,nℓ0−1,…,n−ℓ0)T\mathbf{n}(t) = (n_{\ell_{0}}, n_{\ell_{0}-1}, \ldots, n_{-\ell_{0}})^{\mathrm{T}} and MM is the (2ℓ0+1)×(2ℓ0+1)(2\ell_{0}+1) \times(2\ell_{0}+1) tridiagonal matrix:

M=(−w−w+0⋯0w−−(w++w−)w+⋯00w−−(w++w−)⋯000⋯w−−w+).(64)M = \begin{pmatrix} -w_{-} & w_{+} & 0 & \cdots& 0 \\ w_{-} & -(w_{+}+w_{-}) & w_{+} & \cdots& 0 \\ 0 & w_{-} & -(w_{+}+w_{-}) & \cdots& 0 \\ 0 & 0 & \cdots& w_{-} & -w_{+} \end{pmatrix}. \tag*{(64)}

The formal solution of Eq. (63) is given by the matrix exponential:

n(t)=eMtn(0).(65)\mathbf{n}(t) = e^{Mt}\mathbf{n}(0). \tag*{(65)}

If MM is diagonalizable, there exists an invertible matrix PP and a diagonal matrix Λ=diag⁡(−λ0,−λ1,…,−λ2ℓ0)\Lambda= \operatorname{diag}(-\lambda_{0}, -\lambda_{1}, \ldots, -\lambda_{2\ell_{0}}) such that M=PΛP−1M = P\Lambda P^{-1}, and the solution becomes:

n(t)=∑i=02ℓ0cie−λitvi,(66)\mathbf{n}(t) = \sum_{i=0}^{2\ell_{0}} c_{i}e^{-\lambda_{i}t}\mathbf{v}_{i}, \tag*{(66)}

where −λi-\lambda_{i} are the eigenvalues of MM, vi\mathbf{v}_{i} are the corresponding eigenvectors, and cic_{i} are expansion coefficients determined by the initial condition. Conservation of total particle number implies that the rows of MM sum to zero, i.e., ∑jMij=0\sum_{j}M_{ij}=0 for all ii. Consequently, MM has a zero eigenvalue, which we denote −λ0=0-\lambda_0 = 0, with the corresponding eigenvector proportional to the stationary-state distribution nstn^{\mathrm{st}} satisfying Mnst=0M n^{\mathrm{st}} = 0.

All other eigenvalues have negative real parts, ensuring exponential decay of any deviation from the stationary state. Writing the solution as:

n(t)=nst+∑i=12ℓ0cie−λitvi,(67)n(t) = n^{\mathrm{st}} + \sum_{i=1}^{2\ell_0} c_i e^{-\lambda_i t} v_i, \tag*{(67)}

the long-time behavior is dominated by the slowest decaying mode, i.e., the eigenvalue with the largest real part among the negative eigenvalues (closest to zero). Denoting this eigenvalue as −λ1-\lambda_1, the relaxation time τ\tau is defined as:

τ≡1λ1.(68)\tau\equiv\frac{1}{\lambda_1}. \tag*{(68)}

To determine λ1\lambda_1, we solve the eigenvalue problem Mv=−λvM v = -\lambda v. Writing the components of vv as vmv_m, for ℓ0=1\ell_0 = 1, the system has three OAM states m=1,0,−1m = 1, 0, -1. The rate matrix MM is

M=(−w−w+0w−−(w++w−)w+0w−−w+).(69)M = \begin{pmatrix} -w_- & w_+ & 0 \\ w_- & -(w_+ + w_-) & w_+ \\ 0 & w_- & -w_+ \end{pmatrix}. \tag*{(69)}

The characteristic polynomial det⁡(M+λI)=0\det(M + \lambda I) = 0 yields

λ[λ2−2(w++w−)λ+(w+2+w+w−+w−2)]=0,(70)\lambda\left[\lambda^2 - 2(w_+ + w_-)\lambda+ (w_+^2 + w_+w_- + w_-^2)\right] = 0, \tag*{(70)}

Thus one eigenvalue is λ0=0\lambda_0 = 0; the other two satisfy

λ2−2(w++w−)λ+(w+2+w+w−+w−2)=0,(71)\lambda^2 - 2(w_+ + w_-)\lambda+ (w_+^2 + w_+w_- + w_-^2) = 0, \tag*{(71)}

which gives

λ1,2=(w++w−)±w+w−,(72)\lambda_{1,2} = (w_+ + w_-) \pm\sqrt{w_+w_-}, \tag*{(72)}

The smallest positive root (which gives the slowest decay) is

λ=(w++w−)−w+w−,(73)\lambda= (w_+ + w_-) - \sqrt{w_+w_-}, \tag*{(73)}

Hence the OAM relaxation time is

τOAM=1λ=1(w++w−)−w+w−.(74)\tau_{\mathrm{OAM}} = \frac{1}{\lambda} = \frac{1}{(w_+ + w_-) - \sqrt{w_+w_-}}. \tag*{(74)}

In the general asymmetric case (w+≠w−w_+ \ne w_-), the parameters must be specified so that the eigenvalues can be computed numerically. The relaxation time is still given by Eq. (68), where −λ1-\lambda_1 is the eigenvalue with the largest real part among the negative eigenvalues (closest to zero). This follows directly from the expansion (67), as the mode with eigenvalue −λ1-\lambda_1 decays slowest and thus dominates the long-time approach to stationary state, which can also be proved for the spin case in Eqs. (108) and (C5), and the same results are obtained for Eqs. (112) to (115).

Discussion and Summary

We have presented a comprehensive theoretical analysis of radiation and polarization phenomena for relativistic electrons in a uniform magnetic field, with particular emphasis on the role of OAM. Starting from the Dirac wave function, we derived expressions for transition probabilities and radiation intensities involving both spin and OAM degrees of freedom. In the high-electron-energy and low-photon-energy limits, the WKB approximation yields analytical results that reveal how synchrotron radiation polarizes electron beams in OAM.

Our key findings illustrate that, in the low-photon-energy limit, synchrotron radiation induces an asymmetry in OAM transitions similar to that in spin-flip transitions, leading to spontaneous OAM polarization just as in the spin case. Unlike spin, however, the OAM quantum number in a magnetic field has in principle no lower bound, yet for physically allowed large negative −ℓ0-\ell_0 the polarization behavior stabilizes, and the final distribution differs from that of spin. Meanwhile, the relaxation time τOAM\tau_{\mathrm{OAM}} is significantly shorter than τspin\tau_{\mathrm{spin}}. The stationary state polarization Pspin\mathcal{P}_{\mathrm{spin}} reaches about 92.38%, corresponding to N−1N_{-1} of 96.19%, with a characteristic relaxation time given by Eq. (107). In the case of small photon energy, the OAM transition rates between adjacent ℓ\ell states are asymmetric, with w+/w−≈0.3393w_+/w_- \approx0.3393. This asymmetry drives the electron distribution toward states with minimal OAM projection, leading to an OAM polarization POAM\mathcal{P}_{\mathrm{OAM}} of unity and TNN of 96.09% for large ℓ0\ell_0, which indicates that the electrons ultimately become polarized near the minimal OAM projection states. The derived analytical formulas, particularly Eqs. (40) and (41) for OAM transition rates, provide a quantitative basis for predicting polarization dynamics in high-energy electron beams.

These results extend the well-known Sokolov-Ternov effect to the OAM degree of freedom. The polarization of spin and OAM suggests new possibilities for generating and controlling vortex electron beams with defined angular momentum properties. Such beams could find applications in advanced accelerator science and high-energy physics. Our work provides the foundational radiation theory necessary to understand how such vortex states evolve and polarize under synchrotron emission. Building on the numerous existing studies on the generation of vortex electron beams [39, 40, 41, 42, 43], our work provides a theoretical foundation for the future production of polarized vortex electron beams. The polarization behavior under other conditions will be investigated in our future study, along with the experimental signatures of OAM-polarized beams in storage rings and the interplay between spin and OAM polarization in tailored magnetic field configurations.

acknowledgments

The work was supported by the National Key R&D Program of China No. 2024YFE0109802, the National Natural Science Foundation of China (Grants No. 12175320 and No. 12375084), and Guangdong Basic and Applied Basic Research Foundation (Grant No. 2022A1515010280).

Useful Definitions and Integrals

The four-component spinor f\mathbf{f} in Eqs. (3) is

f=(2γ)1/2(C1In−1,s(ρ)e−iϕ/2iC2In,s(ρ)eiϕ/2C3In−1,s(ρ)e−iϕ/2iC4In,s(ρ)eiϕ/2)(75)\mathbf{f}=(2\gamma)^{1/2}\begin{pmatrix} C_1 I_{n-1,s}(\rho)e^{-i\phi/2}\\ iC_2 I_{n,s}(\rho)e^{i\phi/2}\\ C_3 I_{n-1,s}(\rho)e^{-i\phi/2}\\ iC_4 I_{n,s}(\rho)e^{i\phi/2} \end{pmatrix} \tag*{(75)}

where γ=ϵ0H/(2cℏ)\gamma=\epsilon_0H/(2c\hbar) and ρ=γr2\rho=\gamma r^2. The coefficients CiC_i are determined by the spin projection ζ\zeta and the kinetic parameters

(C1C2C3C4)=122(B3(A3+A4)B4(A4−A3)B3(A3−A4)B4(A3+A4))(76)\begin{pmatrix} C_1\\ C_2\\ C_3\\ C_4 \end{pmatrix}=\frac{1}{2\sqrt{2}}\begin{pmatrix} B_3(A_3+A_4)\\ B_4(A_4-A_3)\\ B_3(A_3-A_4)\\ B_4(A_3+A_4) \tag*{(76)} \end{pmatrix}

with

A3=(1+ϵk3K)1/2,A4=ϵζ(1−ϵk3K)1/2,A_3=\left(1+\frac{\epsilon k_3}{K}\right)^{1/2},\qquad A_4=\epsilon\zeta\left(1-\frac{\epsilon k_3}{K}\right)^{1/2},
B3=(1+ζk0K0)1/2,B4=ζ(1−ζk0K0)1/2.(77)B_3=\left(1+\frac{\zeta k_0}{K_0}\right)^{1/2},\qquad B_4=\zeta\left(1-\frac{\zeta k_0}{K_0}\right)^{1/2}. \tag*{(77)}

Here, ϵ=±1\epsilon=\pm1 labels the positive or negative-energy solutions, k0=m0c/ℏk_0=m_0c/\hbar, K=E/(ℏc)K=E/(\hbar c), and K0=K2−k32K_0=\sqrt{K^2-k_3^2}.

And the interaction Hamiltonian in Eqs. (8) is

U=U(+)+U(−),U=U^{(+)}+U^{(-)},
U(−)=ϵ0L3/2∑k,λ2πℏcκ α⋅ϵk,λ(λ)ak,λe−ik⋅r+iκt,U^{(-)}=\frac{\epsilon_0}{L^{3/2}}\sum_{\mathbf{k},\lambda}\sqrt{\frac{2\pi\hbar c}{\kappa}}\,\boldsymbol{\alpha}\cdot\boldsymbol{\epsilon}^{(\lambda)}_{\mathbf{k},\lambda}a_{\mathbf{k},\lambda}e^{-i\mathbf{k}\cdot\mathbf{r}+i\kappa t},
U(+)=ϵ0L3/2∑k,λ2πℏcκ α⋅ϵk,λ(λ)∗ak,λ†eiκt−ik⋅r.(78)U^{(+)}=\frac{\epsilon_0}{L^{3/2}}\sum_{\mathbf{k},\lambda}\sqrt{\frac{2\pi\hbar c}{\kappa}}\,\boldsymbol{\alpha}\cdot\boldsymbol{\epsilon}^{(\lambda)*}_{\mathbf{k},\lambda}a^\dagger_{\mathbf{k},\lambda}e^{i\kappa t-i\mathbf{k}\cdot\mathbf{r}}. \tag*{(78)}

Here ak,λa_{\mathbf{k},\lambda}, ak,λ†a^\dagger_{\mathbf{k},\lambda} are the photon annihilation and creation operators, respectively. The quantization volume is L3L^3. The transition amplitude from an initial electron state ψb\psi_b to a final state ψa\psi_a is [8]

Ca(t)=−ie0ℏL3/22πℏcκ∫0tdt′e−i(Eb−Ea−ℏκ)t′/ℏα‾⋅ϵ(λ)∗,(79)C_a(t)=-\frac{ie_0}{\hbar L^{3/2}}\sqrt{\frac{2\pi\hbar c}{\kappa}}\int_0^t dt'e^{-i(E_b-E_a-\hbar\kappa)t'/\hbar}\overline{\boldsymbol{\alpha}}\cdot\boldsymbol{\epsilon}^{(\lambda)*}, \tag*{(79)}

For large n,n′n,n', the approximation of II-functions follows

\begin{pmatrix} I_{nn'}\\ I_{n,n'-1}\\ I_{n-1,n'}\\ I_{n-1,n'-1} \end{pmatrix}=2N(\varepsilon)^{1/2}\left[K_{1/3}+\begin{pmatrix} 0\\ -(1+\xi_0y)\\ 1\\ -\xi_0y \end{pmatrix}\varepsilon^{1/2}K_{2/3}\right] \tag*{(80)}\end{pmatrix}

where the normalization factor and arguments are

N=(1+ξ0y)1/22π3,z=y2(εε0)3/2,N=\frac{(1+\xi_0y)^{1/2}}{2\pi\sqrt{3}},\qquad z=\frac{y}{2}\left(\frac{\varepsilon}{\varepsilon_0}\right)^{3/2},
ε=1−β2sin⁡2θ,ε0=1−β2.(81)\varepsilon=1-\beta^2\sin^2\theta,\qquad\varepsilon_0=1-\beta^2. \tag*{(81)}

Here β=v/c\beta=v/c is the electron speed, ξ0\xi_0 is defined in Eq. (86), and yy is the spectral variable from Eq. (84).

The spinor coefficients AiA_i, BiB_i in the high-electron-energy limit (E≫m0c2E\gg m_0c^2, ε01/2≪1\varepsilon_0^{1/2}\ll1) are expanded to first order in ε01/2\varepsilon_0^{1/2}

(A3A4)=(1ζ),(B3B4)=(1ζ)(1±12ζε01/2),\begin{pmatrix} A_3\\ A_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta\end{pmatrix},\qquad \begin{pmatrix} B_3\\ B_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta\end{pmatrix}\left(1\pm\frac{1}{2}\zeta\varepsilon_0^{1/2}\right),
(A3′A4′)=(1ζ′)(1+12ξ0ycos⁡θ),\begin{pmatrix} A'_3\\ A'_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta' \end{pmatrix}\left(1+\frac{1}{2}\xi_0y\cos\theta\right),
\begin{pmatrix} B'_3\\ B'_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta' \end{pmatrix}\left(1\pm\frac{1}{2}\zeta'\varepsilon_0^{1/2}(1+\xi_0y)\right). \tag*{(82)}\end{pmatrix}

In the high-electron-energy and low-photon-energy limits, one can get from Eq. (82)

(A3A4)=(1ζ),(B3B4)=(1ζ),\begin{pmatrix} A_3\\ A_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta\end{pmatrix},\qquad \begin{pmatrix} B_3\\ B_4 \end{pmatrix}=\begin{pmatrix} 1\\ \zeta\end{pmatrix},
\begin{pmatrix} A'_3\\ A'_4 \end{pmatrix}\approx\begin{pmatrix} 1\\ \zeta' \end{pmatrix}\left(1+\frac{1}{2}\xi_0y\cos\theta\right),\qquad \begin{pmatrix} B'_3\\ B'_4 \end{pmatrix}\approx\begin{pmatrix} 1\\ \zeta' \end{pmatrix}. \tag*{(83)}\end{pmatrix}

the dimensionless parameter yy is defined as

y=Kκ03/23γ(1−κ/(K−k0)),with ε0=k02K2=(m0c2E)2,(84)y=\frac{K\kappa_0^{3/2}}{3\gamma(1-\kappa/(K-k_0))},\qquad\text{with }\varepsilon_0=\frac{k_0^2}{K^2}=\left(\frac{m_0c^2}{E}\right)^2, \tag*{(84)}

it simplifies the expressions. In the high-electron-energy limit (E≫m0c2E\gg m_0c^2, ε01/2≪1\varepsilon_0^{1/2}\ll1), the relations can be obtained

κ=Kξ0y1+ξ0y,K′=K1+ξ0y(85)\kappa=\frac{K\xi_0y}{1+\xi_0y},\qquad K'=\frac{K}{1+\xi_0y} \tag*{(85)}
dk=ξ0K(1+ξ0y)2dy,ξ0≡3γKε03/2=32HH0Em0c2,(86)dk=\frac{\xi_0K}{(1+\xi_0y)^2}dy,\qquad\xi_0\equiv\frac{3\gamma}{K\varepsilon_0^{3/2}}=\frac{3}{2}\frac{H}{H_0}\frac{E}{m_0c^2}, \tag*{(86)}

where H0=m02c3/(e0ℏ)H_0=m_0^2c^3/(e_0\hbar) is the Schwinger critical field. Substituting Eqs. (80), (A8) into the expressions for Φ2\Phi_2 and Φ3\Phi_3 [Eq. (11)], and inserting it into the general transition rate formula Eq. (14) while using the integral properties, one can get the result in Eqs. (106) and obtain

\begin{pmatrix}-\bar{\alpha}_1\\ \bar{\alpha}_2\end{pmatrix} =\frac{1}{4}\left[(A'_3A_4+A'_4A_3)(B'_3B_4 I_{n,n'-1}(x)+B'_4B_3 I_{n-1,n'}(x))\right]I_{s,s'}(x), \tag*{(87)}\end{pmatrix}
αˉ3=14[(A3′A3−A4′A4)(B3′B3In−1,n′−1(x)+B4′B4In,n′(x))]Is,s′(x).(88)\bar{\alpha}_3=\frac{1}{4}\left[(A'_3A_3-A'_4A_4)(B'_3B_3 I_{n-1,n'-1}(x)+B'_4B_4 I_{n,n'}(x))\right]I_{s,s'}(x). \tag*{(88)}

The arguments Ai,BiA_i,B_i and Ai′,Bi′A'_i,B'_i refer to the initial and final states, respectively, and are given by Eqs. (A3). The variable xx is defined as

x=κ2sin⁡2θ4γ,(89)x=\frac{\kappa^2\sin^2\theta}{4\gamma}, \tag*{(89)}

The related transition probability per unit time is

w(n,s,ζ→n′,s′,ζ′)=e022πℏ∫0∞dκ κ δ(κ+K′(κ)−K)×∫dΩ Φ(κ,θ;ζ′,s;s′,ζ′),(90)w(n,s,\zeta\to n',s',\zeta')=\frac{e_0^2}{2\pi\hbar}\int_0^\infty d\kappa\,\kappa\,\delta(\kappa+K'(\kappa)-K) \times\int d\Omega\,\Phi(\kappa,\theta;\zeta',s;s',\zeta'), \tag*{(90)}

where K=E/(ℏc)K=E/(\hbar c), K′(κ)=κ02+4γn′+κ2cos⁡2θK'(\kappa)=\sqrt{\kappa_0^2+4\gamma n'+\kappa^2\cos^2\theta}, dΩ=sin⁡θ dθdϕd\Omega=\sin\theta\,d\theta d\phi, and Φ=Φ2+Φ3\Phi=\Phi_2+\Phi_3 from Eq. (11).

The Dirac delta function enforces energy conservation, E=E′+ℏcκE=E'+\hbar c\kappa. Defining κ0\kappa_0 as the positive root of κ+K′(κ)−K=0\kappa+K'(\kappa)-K=0, it gets

κ0=K−K2−4γνsin⁡2θsin⁡2θ,(91)\kappa_0=\frac{K-\sqrt{K^2-4\gamma\nu\sin^2\theta}}{\sin^2\theta}, \tag*{(91)}

and the derivative

ddκ(κ+K′(κ)−K)∣κ=κ0=1+κ0cos⁡2θK′(κ0).(92)\left.\frac{d}{d\kappa}(\kappa+K'(\kappa)-K)\right|_{\kappa=\kappa_0} =1+\frac{\kappa_0\cos^2\theta}{K'(\kappa_0)}. \tag*{(92)}

Where ν=n−n′\nu=n-n'. Using the delta function identity, the κ\kappa-integral in Eq. (A16) is performed analytically.

The following integrals involving modified Bessel functions appear in the evaluation of the angular and spectral integrals in the main text:

12π∫02πdϕ∫0∞drexp⁡{−iκrsin⁡θsin⁡ϕ+i(ℓ−ℓ′)ϕ}=Jℓ−ℓ′(κrsin⁡θ)(93)\frac{1}{2\pi}\int_0^{2\pi}d\phi\int_0^\infty dr\exp\{-i\kappa r\sin\theta\sin\phi+i(\ell-\ell')\phi\} =J_{\ell-\ell'}(\kappa r\sin\theta) \tag*{(93)}
∫0∞Jℓ−ℓ′(2(xρ)1/2)In′,s′(ρ)In,s(ρ)dρ=In,n′(x)Is,s′(x)(94)\int_0^\infty J_{\ell-\ell'}\left(2(x\rho)^{1/2}\right)I_{n',s'}(\rho)I_{n,s}(\rho)d\rho=I_{n,n'}(x)I_{s,s'}(x) \tag*{(94)}

Where ρ=γr2\rho=\gamma r^2, x=(κ2sin⁡2θ)/(4γ)x=(\kappa^2\sin^2\theta)/(4\gamma), n′=ℓ′+s′n'=\ell'+s', n=ℓ+sn=\ell+s, and Jℓ−ℓ′J_{\ell-\ell'} is the Bessel function. Integral formula for yy

∫0y01y1/3(1ay+13)2/3dy=12a2/3ln⁡(1+u0+u02(1−u0)2)+3a2/3(π6−arctan⁡(2u0+13)),u0=(ay0ay0+1/3)1/3.(95)\int_0^{y_0}\frac{1}{y^{1/3}}\left(\frac{1}{ay+\frac{1}{3}}\right)^{2/3}dy =\frac{1}{2a^{2/3}}\ln\left(\frac{1+u_0+u_0^2}{(1-u_0)^2}\right) +\frac{\sqrt{3}}{a^{2/3}}\left(\frac{\pi}{6}-\arctan\left(\frac{2u_0+1}{\sqrt{3}}\right)\right),\quad u_0=\left(\frac{ay_0}{ay_0+1/3}\right)^{1/3}. \tag*{(95)}

The uniform asymptotic formula of WKB approximation is [5]

In,n′(x)≈{12π(x0−x)1/4(x0′−x)1/4exp⁡[−S1(x)],I: x≤x0−δ1,x01/3(nn′)1/121x1/2Ai⁡((nn′)1/6x02/3(x0−x)),II: x0−δ1<x<x0+δ2,2π1(x−x0)1/4(x0′−x)1/4sin⁡[S2(x)+π4],III: x0+δ2≤x≤x0′−δ3,(−1)p+1x0′1/3(nn′)1/121x1/2Ai⁡((nn′)1/6x02/3(x−x0′)),IV: x0′−δ3<x<x0′+δ4,(−1)p+112π(x−x0)1/4(x−x0′)1/4exp⁡[−S3(x)],V: x≥x0′+δ4.(96)I_{n,n'}(x)\approx \begin{cases} \frac{1}{\sqrt{2\pi}(x_0-x)^{1/4}(x'_0-x)^{1/4}}\exp[-S_1(x)],&\mathrm{I}:~x\le x_0-\delta_1,\\ \frac{x_0^{1/3}}{(nn')^{1/12}}\frac{1}{x^{1/2}}\operatorname{Ai}\left(\frac{(nn')^{1/6}}{x_0^{2/3}}(x_0-x)\right),&\mathrm{II}:~x_0-\delta_1<x<x_0+\delta_2,\\ \sqrt{\frac{2}{\pi}}\frac{1}{(x-x_0)^{1/4}(x'_0-x)^{1/4}}\sin\left[S_2(x)+\frac{\pi}{4}\right],&\mathrm{III}:~x_0+\delta_2\le x\le x'_0-\delta_3,\\ (-1)^{p+1}\frac{x_0'^{1/3}}{(nn')^{1/12}}\frac{1}{x^{1/2}}\operatorname{Ai}\left(\frac{(nn')^{1/6}}{x_0^{2/3}}(x-x'_0)\right),&\mathrm{IV}:~x'_0-\delta_3<x<x'_0+\delta_4,\\ (-1)^{p+1}\frac{1}{\sqrt{2\pi}(x-x_0)^{1/4}(x-x'_0)^{1/4}}\exp[-S_3(x)],&\mathrm{V}:~x\ge x'_0+\delta_4. \end{cases} \tag*{(96)}

where Ai⁡(x)\operatorname{Ai}(x) is an Airy function, S1,S2,S3S_1,S_2,S_3 are the positive WKB action integrals in the classically allowed regions. The transitional widths δi\delta_i scale as δ1,δ2≳x0/[(nn′)1/6(n−n′)1/3]\delta_1,\delta_2\gtrsim x_0/[(nn')^{1/6}(\sqrt{n}-\sqrt{n'})^{1/3}] and $\delta_3,\delta_4\gtrsim x0′/[(nn′)1/6(n+n′)1/3]x'_0/[(nn')^{1/6}(\sqrt{n}+\sqrt{n'})^{1/3}].

Since x≲x0x \lesssim x_0 in our problem, region II is the most relevant. In this region, the Airy function can be expressed in terms of modified Bessel functions, the approximation can be obtained

In,n′(x)≈1π3(1−xx0)1/2K1/3(z),(97)I_{n,n'}(x) \approx\frac{1}{\pi\sqrt{3}}\left(1-\frac{x}{x_0}\right)^{1/2}K_{1/3}(z), \tag*{(97)}
z=23(x02nn′)1/4(1−xx0)3/2.z=\frac{2}{3}(x_0^2nn')^{1/4}\left(1-\frac{x}{x_0}\right)^{3/2}.

The uniform WKB approximation for the radial functions In,n′(x)I_{n,n'}(x) is constructed by matching asymptotic solutions across the turning points x0x_0 and x0′x'_0. Here pp is an integer determined by the Bohr–Sommerfeld quantization condition:

∫x0x0′−fn,n′(x′) dx′=(p−12)π.(98)\int_{x_0}^{x'_0}\sqrt{-f_{n,n'}(x')} \, dx'=\left(p-\frac{1}{2}\right)\pi. \tag*{(98)}

The integral in Eq. (98) can be evaluated exactly, but we only require the limit of large nn and n′n', where

fn,n′(x)≈(x−x0)(x−x0′)4x2.(99)f_{n,n'}(x)\approx\frac{(x-x_0)(x-x'_0)}{4x^2}. \tag*{(99)}

Performing the integration in Eq. (98) gives

∫x0x0′−fn,n′(x′) dx′=π(min⁡{n,n′}+12).(100)\int_{x_0}^{x'_0}\sqrt{-f_{n,n'}(x')} \, dx'=\pi\left(\min\{n,n'\}+\frac{1}{2}\right). \tag*{(100)}

Therefore,

p=min⁡{n,n′}+1.(101)p=\min\{n,n'\}+1. \tag*{(101)}

The relevant action integrals are

S1(x)=∫xx0f(x′) dx′,x<x0,(102)S_1(x)=\int_x^{x_0}\sqrt{f(x')} \, dx', \qquad x<x_0, \tag*{(102)}
S2(x)=∫x0x−f(x′) dx′,x0<x<x0′,(103)S_2(x)=\int_{x_0}^{x}\sqrt{-f(x')} \, dx', \qquad x_0<x<x'_0, \tag*{(103)}
S3(x)=∫x0′xf(x′) dx′,x>x0′.(104)S_3(x)=\int_{x'_0}^{x}\sqrt{f(x')} \, dx', \qquad x>x'_0. \tag*{(104)}

Explicit expressions for these integrals in terms of elementary functions can be obtained but are omitted here for brevity. The related formulas yield the piecewise approximation Eq. (96) used in the main text.

Spin Transitions and Polarization Dynamics

Spin transitions

The WKB approximation can be applied here to derive analytic expressions for the spin-flip transition rates. For large nn, n′n', the combinations of II-functions appearing in the matrix elements Eqs. (87) and (A14) are given in Eq. (80), which can be expressed linearly in terms of modified Bessel functions K1/3K_{1/3} and K2/3K_{2/3} [8]. Combining all the relations, one can arrive at the compact result for the spin-dependent transition rate:

w(ζ,ζ′)=536e02ℏccREm0c2[1+ζζ′2(1−16345ζ0+2518ζ02−ζ′5ζ0(1−2039ζ))+1−ζζ′2ζ026(1−ζ′8315)].(105)w(\zeta,\zeta')=\frac{5\sqrt{3}}{6}\frac{e_0^2}{\hbar c}\frac{c}{R}\frac{E}{m_0c^2}\left[\frac{1+\zeta\zeta'}{2}\left(1-\frac{16\sqrt{3}}{45}\zeta_0+\frac{25}{18}\zeta_0^2-\frac{\zeta'}{5}\zeta_0\left(1-\frac{20\sqrt{3}}{9}\zeta\right)\right)+\frac{1-\zeta\zeta'}{2}\frac{\zeta_0^2}{6}\left(1-\zeta'\frac{8\sqrt{3}}{15}\right)\right]. \tag*{(105)}

The rate for a spin-flip transition (ζ′=−ζ\zeta'=-\zeta) follows as

wflip(ζ)=12τ(1+8315ζ),wflip(−1→1)wflip(1→−1)≈0.0396,(106)w_{\mathrm{flip}}(\zeta)=\frac{1}{2\tau}\left(1+\frac{8\sqrt{3}}{15}\zeta\right), \qquad\frac{w_{\mathrm{flip}}(-1\to1)}{w_{\mathrm{flip}}(1\to-1)}\approx0.0396, \tag*{(106)}

where the characteristic relaxation time τ\tau is

τ=8ℏ253m0ce02(m0c2E)2(H0H)3.(107)\tau=\frac{8\hbar^2}{5\sqrt{3}m_0ce_0^2}\left(\frac{m_0c^2}{E}\right)^2\left(\frac{H_0}{H}\right)^3. \tag*{(107)}

Equation (C2) reveals the asymmetry between spin-flip rates for ζ=+1\zeta=+1 and ζ=−1\zeta=-1, which is the origin of the Sokolov-Ternov effect leading to spontaneous spin polarization of the electron beam.

Spin polarization dynamics

The time evolution of the spin populations is governed by the coupled rate equations

dn↓dt=n↑wflip,ζ=1−n↓wflip,ζ=−1,(108)\frac{dn_{\downarrow}}{dt}=n_{\uparrow}w_{\mathrm{flip},\zeta=1}-n_{\downarrow}w_{\mathrm{flip},\zeta=-1}, \tag*{(108)}
dn↑dt=n↓wflip,ζ=−1−n↑wflip,ζ=1,(109)\frac{dn_{\uparrow}}{dt}=n_{\downarrow}w_{\mathrm{flip},\zeta=-1}-n_{\uparrow}w_{\mathrm{flip},\zeta=1}, \tag*{(109)}

where n↓n_{\downarrow} and n↑n_{\uparrow} denote the number of electrons with spin anti-parallel (ζ=−1\zeta=-1) and parallel (ζ=+1\zeta=+1) to the magnetic field, respectively. The total electron number is conserved:

n↓+n↑=n=constant.(110)n_{\downarrow}+n_{\uparrow}=n=\text{constant}. \tag*{(110)}

For an initially unpolarized beam at t=0t=0, it gets

n↓(0)=n↑(0)=n/2.(111)n_{\downarrow}(0)=n_{\uparrow}(0)=n/2. \tag*{(111)}

Using the spin-flip transition rates from Eq. (106), the solution of Eqs. (108)–(111) is

\begin{pmatrix} n_{\downarrow}(t) \\ n_{\uparrow}(t) \end{pmatrix} =\frac{n}{30}\begin{pmatrix} 15+8\sqrt{3}\left[1-e^{-t/\tau}\right] \\ 15-8\sqrt{3}\left[1-e^{-t/\tau}\right] \end{pmatrix}, \tag*{(112)}\end{pmatrix}

where the relaxation time τ\tau is given by Eq. (107). The asymptotic polarization degree for t≫τt\gg\tau is defined as

Pspin=n↓(∞)−n↑(∞)n↓(∞)+n↑(∞)=⟨ζ⟩ζmin⁡≈0.9238.(113)P_{\mathrm{spin}}=\frac{n_{\downarrow}(\infty)-n_{\uparrow}(\infty)}{n_{\downarrow}(\infty)+n_{\uparrow}(\infty)}=\frac{\langle\zeta\rangle}{\zeta_{\min}}\approx0.9238. \tag*{(113)}

Here ⟨s⟩\langle s\rangle is the average spin of the electrons, and smin⁡=−1/2s_{\min}=-1/2 is its minimum value. The result corresponds to approximately 92.38% of the net electrons having their spin antiparallel to the magnetic field. This stationary state polarization is illustrated in Figure 6.

Equivalently, the evolution of the average spin projection ⟨ζ⟩\langle\zeta\rangle can be derived from the transition rates. Using Eq. (106), it can be obtained

d⟨ζ⟩dt=∑ζ′=−ζ(ζ′−ζ)w(ζ,ζ′)=−1τ(⟨ζ⟩+8315)(114)\frac{d\langle\zeta\rangle}{dt}=\sum_{\zeta'=-\zeta}(\zeta'-\zeta)w(\zeta,\zeta')=-\frac{1}{\tau}\left(\langle\zeta\rangle+\frac{8\sqrt{3}}{15}\right) \tag*{(114)}

which integrates to

⟨ζ(t)⟩=−8315[1−e−t/τ].(115)\langle\zeta(t)\rangle=-\frac{8\sqrt{3}}{15}\left[1-e^{-t/\tau}\right]. \tag*{(115)}

Equation (C11) reduces to Eq. (113) in the equilibrium-state limit.

References

References

  1. [1]V. N. Baier, Radiative polarization of electrons in storage rings, Sov. Phys. Usp. 14, 695 (1972).
  2. [2]J. Schwinger, On the classical radiation of accelerated electrons, Phys. Rev. 75, 1912 (1949).
  3. [3]A. A. Sokolov and I. M. Ternov, On polarization and spin effects in the theory of synchrotron radiation, Sov. Phys. Dokl. 8, 1203 (1964).
  4. [4]Bryan W. Montague, Polarized beams in high energy storage rings, Phys. Rep 113, 1 (1984).
  5. [5]M. Buzzegoli, K. Tuchin, N. Vijaya­kumar, Quasi-classical approximation of electromagnetic radiation by fermions embedded in a rigidly rotating medium in a strong magnetic field, Phys. Rev. C 111, 054907 (2025).arxiv.org/abs/2503.06649
  6. [6]S. R. Mane, Yu. M. Shatunov, and K. Yokoya, Spin-polarized charged particle beams in high-energy accelerators, Rep. Prog. Phys. 68, (1997).
  7. [7]E. Karimi, L. Marrucci, V. Grillo, and E. Santamato, Spin-to-orbital angular momentum conversion and spin-polarization filtering in electron beams, Phys. Rev. Lett. 108, 044801 (2012).DOI
  8. [8]A. A. Sokolov and I. M. Ternov, Radiation from Relativistic Electrons, (American Institute of Physics, New York, 1986).
  9. [9]R. Rossmanith, High Energy Polarized Electron Beams, UNT Digital Library (1987).
  10. [10]S. R. Mane, Radiative Spin Polarization in High Energy Storage Rings, arXiv:2512.13958 (2025).arxiv.org/abs/2512.13958
  11. [11]Y. S. Derbenev and A. M. Kondratenko, Polarization kinetics of electrons in storage rings, Sov. Phys. JETP 37, 968 (1973).
  12. [12]B. W. Montague, Polarized beams in high energy storage rings, Phys. Rep. 113, 1 (1984).
  13. [13]T. Maruyama, T. Hayakawa, T. Kajino and M.-K. Cheoun, Generation of photon vortex by synchrotron radiation from electrons in Landau states under astrophysical magnetic fields, Phys. Lett. B 826, 136779 (2022).
  14. [14]radiation experiments in relativistic quantum approach, Phys. Rev. Research 5, 043289 (2023).
  15. [15]L. Allen, M. W. Beijersbergen, R. J. C. Spreeuw, and J. P. Woerdman, Orbital angular momentum of light and the transformation of Laguerre-Gaussian laser modes, Phys. Rev. A 45, 8185 (1992).
  16. [16]M. J. Padgett, Orbital angular momentum 25 years on, Opt. Express 25, 11265 (2017).DOI
  17. [17]S. M. Lloyd, M. Babiker, and J. Yuan, Interaction of electron vortices and optical vortices with matter and processes of orbital angular momentum exchange, Phys. Rev. A 86, 033824 (2012).DOI
  18. [18]K. Y. Bliokh, Y. P. Bliokh, S. Savel’ev, and F. Nori, Semiclassical dynamics of electron wave packet states with phase vortices, Phys. Rev. Lett. 99, 190404 (2007).DOI
  19. [19]Liping Zou, Pengming Zhang, and A. J. Silenko, General quantum-mechanical solution for twisted electrons in a uniform magnetic field, Phys. Rev. A 103, L010201 (2021).DOI
  20. [20]M. Uchida and A. Tonomura, Generation of electron beams carrying orbital angular momentum, Nature 464, 737 (2010).DOI
  21. [21]J. Verbeeck, H. Tian, P. Schattschneider, Production and application of electron vortex beams, Nature 467, 301 (2010).DOI
  22. [22]A. Yu. Murtazin, G. K. Sizykh, D. V. Grosman, U. G. Rybak, A. A. Shchepkin, and D. V. Karlovets, Photon emission by vortex particles accelerated in a linac, Phys. Rev. D 113, 036024 (2026).
  23. [23]M. Buzzegoli, J. D. K. Roth, K. Tuchin, and N. Vijayakumar, Photon radiation by relatively slowly rotating fermions in magnetic field, Phys. Rev. D 108, 096014 (2023).DOI
  24. [24]Qi Meng, Xuan Liu, Wei Ma, Zhen Yang, Liang Lu, A. J. Silenko, Pengming Zhang, and Liping Zou, Generalized Gouy rotation of electron vortex beams in uniform magnetic fields, Phys. Rev. Res 7, 023306 (2025).arxiv.org/abs/2407.02788
  25. [25]Qi Meng, Ziqiang Huang, Xuan Liu, Wei Ma, Zhen Yang, Liang Lu, A. J. Silenko, Pengming Zhang, and Liping Zou, Relativistic quantum mechanics of charged vortex particles accelerated in a uniform electric field, Phys. Rev. Res 7, 043213 (2025).arxiv.org/abs/2506.18492
  26. [26]K. Y. Bliokh, I. P. Ivanov, G. Guzzinati, et al., Theory and applications of free-electron vortex states, Phys. Rep. 690, 1 (2017).DOI
  27. [27]R. Juchtmans, A. Béché, A. Abakumov, M. Batuk, and J. Verbeeck, Using electron vortex beams to determine chirality of crystals in transmission electron microscopy, Phys. Rev. B 91, 094112 (2015).DOI
  28. [28]B. J. McMorran, A. Agrawal, I. M. Anderson, A. A. Herzing, H. J. Lezec, J. J. McClelland, and J. Unguris, Electron vortex beams with high quanta of orbital angular momentum, Science 331, 192 (2011).DOI
  29. [29]K. Y. Bliokh, I. P. Ivanov, G. Guzzinati, and et al., Theory and applications of free-electron vortex states, Phys. Rep. 690, 1 (2017).DOI
  30. [30]I. P. Ivanov, Promises and challenges of high-energy vortex states collisions, Prog. Part. Nucl. Phys. 127, 103987 (2022).DOI
  31. [31]D. Karlovets, D. Grosman, and I. Pavlov, Angular momentum dynamics of vortex particles in accelerators, Phys. Rev. Lett 136, 085002 (2026).DOI
  32. [32]K. Floettmann and D. Karlovets, Quantum mechanical formulation of the Busch theorem, Phys. Rev. A 102, 043517 (2020).
  33. [33]D. Karlovets, Relativistic vortex electrons: paraxial versus nonparaxial regimes, Phys. Rev. A 98, 012137 (2018).DOI
  34. [34]D. Karlovets, Vortex particles in axially symmetric fields and applications of the quantum Busch theorem, New J. Phys. 23, 033048 (2021).DOI
  35. [35]J. D. Jackson, Classical Electrodynamics, 3rd ed., (Wiley, New York, 1999).DOI
  36. [36]L. D. Landau and E. M. Lifshitz, The Classical Theory of Fields, (Butterworth-Heinemann, Oxford, 1975).
  37. [37]M. Katoh, M. Fujimoto, H. Kawaguchi, K. Tsuchiya, K. Ohmi, T. Kaneyasu, Y. Taira, M. Hosaka, A. Mochihashi, et al., Angular momentum of twisted radiation from an electron in spiral motion, Phys. Rev. Lett. 118, 094801 (2017).
  38. [38]A. J. Silenko, Pengming Zhang, and Liping Zou, Relativistic quantum dynamics of twisted electron beams in arbitrary electric and magnetic fields, Phys. Rev. Lett. 121, 043202 (2018).DOI
  39. [39]L. Clark, A. Béché, G. Guzzinati, et al., Exploiting lens aberrations to create electron-vortex beams, Phys. Rev. Lett. 111, 064801 (2013).DOI
  40. [40]Liping Zou, Pengming Zhang, Alexander J. Silenko, Production of twisted particles in magnetic fields, J. Phys. B 57, 045401 (2024).arxiv.org/abs/2207.14105
  41. [41]P. Schattschneider, M. Stöger-Pollach, and J. Verbeeck, Novel vortex generator and mode converter for electron beams, Phys. Rev. Lett. 109, 084801 (2012).pubmed.ncbi.nlm.nih.gov/23002749
  42. [42]P. Schattschneider, T. Schachinger, M. Stöger-Pollach, S. Löffler, A. Steiger-Thirsfeld, K. Y. Bliokh and F. Nori, Imaging the dynamics of free-electron Landau states, Nat. Commun. 5, 4586 (2014).DOI
  43. [43]E. Mafakheri, A. H. Tavabi, P.-H. Lu, R. Balboni, F. Venturi, C. Menozzi, G. Cazzadi, S. Frabboni, A. Sit, R. E. Dunin-Borkowski, E. Karimi, and V. Grillo, Realization of electron vortices with large orbital angular momentum using miniature holograms fabricated by electron beam lithography, Appl. Phys. Lett. 110, 093113 (2017).DOI

Paper details

vlnc:W_q8fcpd2s
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents