betaFeedback

WorksPhysics and Astronomy

Electromagnetic Helicity in Twisted Cavity Resonators

Abstract

Through left- or right-handed twisting, we investigate the impact of mirror-asymmetry (chirality) of the conducting boundary conditions of an equilaterial triangular cross-section electromagnetic resonator. We observe the generation of eigenmodes with non-zero electromagnetic helicity as a result of the coupling of near degenerate TE11(p+1)_{11(p+1)} and TM11p_{11p} modes. This can be interpreted as an emergence of magneto-electric coupling, which in turn produces a measurable shift in resonant mode frequency as a function of twist angle. We show that this coupling mechanism is equivalent to introducing a non-zero chirality material parameter κeffκ_\text{eff} or axion field θeffθ_{\text{eff}} to the medium. Our findings demonstrate the potential for real-time, macroscopic manipulation of electromagnetic helicity.

Chiral (handed) electromagnetic radiation is distinguished by its left- or right-handed polarization, which arises from the intrinsic angular momentum of the electromagnetic field. The unique light-matter interactions exhibited by chiral radiation at the quantum level have garnered significant interest across various fields, including material science [1], nanophotonics [2, 3, 4], spectroscopy [3], and quantum information processing [5, 6]. These interactions offer numerous applications, from probing molecular chirality, with applications for detecting biomarkers related to neurodegenerative diseases [7], to advancing optical communication technologies [8, 9] and sensitivity to dark matter candidates [10].

Electromagnetic helicity, H\mathcal{H}, is derived by performing a projection measurement of a complex electromagnetic state vector’s spin onto its linear momentum [11, 12, 13, 14], in fitting with the first-principles definition of the term “helicity”. The sign of this projection value will be opposite for different handed helicities and the expectation value of this operator will effectively measure the local, time averaged helicity density. It has been shown that the result of this projection is intrinsically linked to the dual transformation of the electromagnetic state, effectively rotating the electric and magnetic properties of the medium. This results in the effective mixing of electric and magnetic fields and so electromagnetic helicity will necessarily produce an electro-magnetic coupling. Most chiroptical phenomena, which produces electromagnetic helicity, occur as surface states, at optical frequencies, or due to complex meta-structures [4, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24].

We show that electromagnetic cavity resonators with twisted electrically conducting boundary conditions generate helical modes in vacuo without the need for engineered materials. We investigate a cavity with an equilateral triangular cross-section and electrically conducting boundary conditions at the end faces. The mirror symmetry of this resonator can be broken by introducing an arbitrary twist angle ϕ\phi perpendicular to the central axis of the cavity, resulting in the chiral geometry shown in Figure 1a. A right-handed twist corresponds to ϕ>0\phi> 0, while a left-handed twist corresponds to ϕ<0\phi< 0. The breaking of spatial symmetry generates a magneto-

Geometry of the twisted triangular cavity resonator and schematic of the mechanism for generating helical modes

Figure 1. (a) The geometry of the twisted triangular cavity resonator. (b) A high-level diagram illustrating the mechanism for generating helical modes in a twisted triangular cavity resonator. By adjusting one of the two control parameters, either twist angle, ϕ\phi, or the chirality parameter, κ\kappa, across the cavity volume in an FEM simulation, a magneto-electric coupling is introduced within the cavity. This coupling results in a mixing angle, η\eta, between the electric field, E⃗\vec{E}, and the magnetic field, H⃗\vec{H}, of the resonant electromagnetic radiation, leading to the generation of chiral electromagnetic modes with non-zero helicity density, hh, and axion field, θ\theta, both of which can be derived from the FEM simulated results. The circled labels A, B and C indicate the relationships between parameters, given by equations (29), (22) and (25) respectively.

electric coupling that mixes the near degenerate transverse electric (TE) and transverse magnetic (TM) modes, generating eigenmodes with non-zero H\mathcal{H}.

Chirality can also be defined for materials, where the chirality of a material affects how chiral electromagnetic radiation interacts with it and is measured by the chirality material parameter κ\kappa. Material chirality is related to observable effects like optical rotation and circular dichroism [25]. We demonstrate the equivalency of twisting the resonator to introducing some effective, uniform, non-zero κ\kappa over the cavity volume, which we refer to as κeff\kappa_{\mathrm{eff}}. We demonstrate the generation of a κeff\kappa_{\mathrm{eff}} in free space that is two orders of magnitude larger than the bi-isotropoic effect exhibited by the material with the strongest known magneto-electric effect - Cr2O3\mathrm{Cr}_{2}\mathrm{O}_{3} [26].

Equivalent to κeff\kappa_{\mathrm{eff}}, the magneto-electric coupling can be thought to introduce an effective non-zero axion field θeff\theta_{\mathrm{eff}}, which is a pseudoscalar field hypothesised to solve the strong charge-parity (CP) problem [27, 28, 29, 30] in quantum chromodynamics (QCD) [31]. The strong CP-problem involves a CP-violating term theoretically allowed in the strong interaction not being observed in experiments. Introduced by the Peccei-Quinn mechanism, the axion field dynamically cancels out this CP-violating term, resolving the discrepancy. When quantized, the axion field gives rise to a particle called the axion, which is a strong candidate for dark matter. The generation of an axion field provides a method for a highly sensitive search for dark matter [10].

We demonstrate that regardless of the origin: twisting, chiral media, or axion field, the resulting mode mixing leads to a measurable shift in resonant frequency Δω/ω\Delta\omega/\omega as a function of twist angle, ϕ\phi. Further, we demonstrate the inter-relationship between Δω/ω\Delta\omega/\omega, H\mathcal{H}, the all-volume effective chirality parameter κeff\kappa_{\mathrm{eff}} and the effective axion field θeff\theta_{\mathrm{eff}} of the cavity mode. These relationships are summarised in Figure 1(b).

Ultimately, the improved understanding of these systems paves the way for optimal design and modelling of useful devices with the potential to generate a unique state of helical electromagnetic resonance within a macroscopic free-space volume. As has been hinted above, such devices may find use as molecular or solid-state chirality sensors, molecular sorters/filters, and sensors for fundamental physics such as dark-matter or gravity waves [32].

Total resonant helicity Hi\mathcal{H}_i for a given mode, ii, can be formally defined for a monochromatic electromagnetic field in a resonant mode as the integral of the helicity density, hih_i over the cavity volume, VV, expressed as:

hi(r⃗)=2Im⁡[e⃗i(r⃗)⋅h⃗i∗(r⃗)]=2Im⁡[E⃗i(r⃗)⋅H⃗i∗(r⃗)]VEH(1)h_i(\vec{r}) = 2\operatorname{Im}\left[\vec{e}_i(\vec{r}) \cdot\vec{h}_i^{*}(\vec{r})\right] = \frac{2\operatorname{Im}\left[\vec{E}_i(\vec{r}) \cdot\vec{H}_i^{*}(\vec{r})\right]}{V\mathcal{E}\mathcal{H}} \tag*{(1)}

where E\mathcal{E} and H\mathcal{H} are real constants, E⃗i(r⃗)=Ee⃗i(r⃗)\vec{E}_i(\vec{r}) = \mathcal{E}\vec{e}_i(\vec{r}) and H⃗i(r⃗)=Hh⃗i(r⃗)\vec{H}_i(\vec{r}) = \mathcal{H}\vec{h}_i(\vec{r}) are the electric and magnetic vector fields of the mode, respectively, and e⃗i(r⃗)\vec{e}_i(\vec{r}) and h⃗i(r⃗)\vec{h}_i(\vec{r}) are the normalised position dependent eigenvectors such that 1V∫e⃗i(r⃗)∗⋅e⃗i(r⃗)dV=1V∫h⃗i(r⃗)∗⋅h⃗i(r⃗)dV=1\frac{1}{V}\int\vec{e}_i(\vec{r})^{*}\cdot\vec{e}_i(\vec{r})dV = \frac{1}{V}\int\vec{h}_i(\vec{r})^{*}\cdot\vec{h}_i(\vec{r})dV = 1. Thus, the total mode helicity may be written as:

Hi=∫hidV.(2)\mathcal{H}_i = \int h_i dV. \tag*{(2)}

While Hi\mathcal{H}_i offers a quantitative measure of the global chirality of the electromagnetic radiation in the resonant mode, hih_i provides insight into the chirality of the radiation at specific points within the cavity. It is easy to show that E=1V∫∣E⃗i(r⃗)∣2dV\mathcal{E} = \sqrt{\frac{1}{V}\int\lvert\vec{E}_i(\vec{r})\rvert^2dV} and H=1V∫∣H⃗i(r⃗)∣2dV\mathcal{H} = \sqrt{\frac{1}{V}\int\lvert\vec{H}_i(\vec{r})\rvert^2dV}, making (1) and (2) consistent with other definitions of helicity [10].

Origin of Single Modes with Non-Zero Helicity due to Twisting

A generic resonator geometry is represented in Figure 1(a) where vv is the vertex length of the triangular cross-section, ll is the length of the resonator and ϕ\phi the angle that the cross-section is rotated over the length of the resonator. The eigenfrequencies fif_i of these systems were solved using finite element modelling (FEM), for various aspect ratios (AR=v/lAR = v/l) as a function of ϕ\phi, with results shown in Figure 2(a). Previous studies have examined the modes in un-twisted triangular cross-section prisms of in-

Normalised eigenfrequencies of resonant modes in an equilateral triangular resonator

Figure 2. (a) The normalised eigenfrequencies (fiv)2(f_i v)^2 of the resonant modes in an equilateral triangular resonator as a function of ϕ\phi for four different ARAR values, with solution colour corresponding to Hi\mathcal{H}_i, and the green, black, red and blue dashed lines corresponding to the TE10p\mathrm{TE}_{10p}, TE11p\mathrm{TE}_{11p}, TE20p\mathrm{TE}_{20p}, & TM11p\mathrm{TM}_{11p}. (b) The case of AR=0.133AR = 0.133 (AR)2=0.0178(AR)^2 = 0.0178. Labels A, C and D lie along the ψ1+\psi_1^{+} mode, and label B lies on the ψ1−\psi_1^{-} mode.

finite length [33, 34], and have shown that the normalised eigenfrequencies of the form (fiv)2(f_{iv})^2 are linearly dependent on the square of the aspect ratio (AR)2(AR)^2. These ϕ=0\phi= 0 solutions are plotted as the dashed lines in Figure 2(a) as a function of (AR)2(AR)^2. The colour density of this figure represents Hi\mathcal{H}_i of the resonant modes, calculated by evaluating the integrals in (2) from the FEM solutions. The frequencies of the resonant modes in the non-twisted resonators with equilateral cross-section are given by:

fm,n,p=23c(mv)2+(nv)2+(mnv)2+(pl)2(3)f_{m,n,p} = \frac{2}{3}c\sqrt{\left(\frac{m}{v}\right)^2+\left(\frac{n}{v}\right)^2+\left(\frac{mn}{v}\right)^2+\left(\frac{p}{l}\right)^2} \tag*{(3)}

where cc is the speed of light, and mm, nn, and pp are integer mode numbers (here i≡m,n,pi \equiv m,n,p). The mode numbers mm and nn represent the number of variations in the standing wave pattern in the transverse plane of the cavity, while pp corresponds to the number of variations in the longitudinal direction. For a given set of mode numbers (m,n,p)(m,n,p), the matching transverse electric (TE) and transverse magnetic (TM) modes are thus degenerate in frequency. The selection rules for the TE modes are m≥n≥0m \ge n \ge0, m≠0m \ne0, and p>0p > 0, while for the TM modes: m≥n>0m \ge n > 0 and p≥0p \ge0. The straight lines generated by (3) on Figure 2 intersect the (fiv)2(f_{iv})^2 vs ϕ\phi plots at ϕ=0\phi= 0, as is expected. However, as ϕ\phi is increased the majority of the eigenmodes asymmetrically detune in opposite directions from their untwisted frequencies, with a corresponding increase in the magnitude of their Hi\mathcal{H}_i. For positive ϕ\phi we observe negative Hi\mathcal{H}_i modes increasing in frequency whilst positive Hi\mathcal{H}_i modes decrease, and vice-versa for negative ϕ\phi.

We observe that as ϕ\phi increases in magnitude, the near degenerate TM11p_{11p} and TE11(p+1)_{11(p+1)} modes mix to produce

two seperate modes with non-zero Hi\mathcal{H}_i, which approaches unity for relatively small angles. We denote modes produced via in-phase mixing of the form TM11p_{11p}+TE11(p+1)_{11(p+1)} as ψp+1+\psi_{p+1}^{+} and modes produced via out-of-phase mixing of the form TM11p_{11p}−TE11(p+1)_{11(p+1)} as ψp+1−\psi_{p+1}^{-}. Here given we only investigate the mixing of the m=n=1m=n=1 mode family, we have dropped these subscripts in our ψ±\psi^{\pm} nomenclature. Note that ψp+1+\psi_{p+1}^{+} modes will have negative Hi\mathcal{H}_i due to the anti-parallel fields of the untwisted modes, and vice-versa for ψp+1−\psi_{p+1}^{-}. The lowest order ψp+1+\psi_{p+1}^{+} and ψp+1−\psi_{p+1}^{-} modes are labelled A and B, respectively, in the normalised eigenfrequency spectra for (AR)2=0.0178(AR)^2=0.0178 in Figure 2(b).

The mixing occurs due to the introduction of mirror-asymmetry to the boundary conditions due to non-zero ϕ\phi. This breaking of spatial symmetry introduces a magneto-electric coupling that mixes the E⃗i\vec{E}_i and H⃗i\vec{H}_i fields of the TE and TM modes of the untwisted resonator, with examples of the FEM simulated fields shown in Figure 3(a) and (b). This mixing effect can be described by the mixing angle η\eta that transforms the fields in the resonator according to the dual transformation [11, 35, 36]:

Fields and normalised densities for untwisted and twisted cavity modes

Figure 3. The E⃗⊥\vec{E}_{\perp} (black) & jH⃗⊥j\vec{H}_{\perp} (red) fields and the normalised densities of the axial fields EzE_z, HzH_z & Im⁡[Ei⋅Hi∗]\operatorname{Im}[\mathbf{E}_i \cdot\mathbf{H}_i^{*}] for the (a) TM110_{110} mode, (b) TE111_{111} mode, and the twisted (ϕ=2π\phi= 2\pi) (c) ψ1−\psi_1^{-} mode, and (d) ψ1+\psi_1^{+} mode. Note, the mode number pp is easily counted by observing the number of maximums in density plot of ∣Im⁡[Ei⋅Hi∗]∣|\operatorname{Im}[\mathbf{E}_i \cdot\mathbf{H}_i^{*}]|.

(E⃗icμ0H⃗i)=(cos⁡(η)sin⁡(η)−sin⁡(η)cos⁡(η))(E⃗0cμ0H⃗0)(4)\begin{pmatrix} \vec{E}_i \\ c\mu_0\vec{H}_i \end{pmatrix} = \begin{pmatrix} \cos(\eta) & \sin(\eta) \\ -\sin(\eta) & \cos(\eta) \end{pmatrix} \begin{pmatrix} \vec{E}_0 \\ c\mu_0\vec{H}_0 \tag*{(4)} \end{pmatrix}

where E⃗0\vec{E}_0 and H⃗0\vec{H}_0 are the original vector fields in the untwisted resonator. The duality transformation is real and occurs in the time domain.

The ψp+1±\psi_{p+1}^{\pm} modes are shown in Figure 3(c) and (d), and form a new orthogonality basis. The normalised field maxima can only be counted when evaluating the field product Im⁡[E⃗i⋅H⃗i∗]\operatorname{Im}[\vec{E}_i \cdot\vec{H}_i^{*}] to determine the mode number pp. The sign of this field product directly maps to the new orthogonality basis (positive for ψp+1−\psi_{p+1}^{-}, negative for ψp+1+\psi_{p+1}^{+}).

Note that in Figure 2, there also exists modes that do not tune to the same extent as ϕ\phi is varied and have relatively low Hi\mathcal{H}_{i} values. These modes are of the TE10(p+1)\mathrm{TE}_{10(p+1)} family, which do not have the TM counterpart required to couple together, as dictated by the aforementioned selection rules (i.e. TM10p\mathrm{TM}_{10p} is forbidden) and hence do not mix to produce a larger value of Hi\mathcal{H}_{i}.

The ψ±\psi^{\pm} modes can be more precisely described by a fractional mixing of the states ∣TM11p⟩\lvert\mathrm{TM}_{11p}\rangle and ∣TE11(p+1)⟩\lvert\mathrm{TE}_{11(p+1)}\rangle given by

∣ψp+1±⟩=∣δ∣∣TM11p⟩±∣β∣∣TE11(p+1)⟩,(5)\lvert\psi_{p+1}^{\pm}\rangle= \lvert\delta\rvert\lvert\mathrm{TM}_{11p}\rangle\pm\lvert\beta\rvert\lvert\mathrm{TE}_{11(p+1)}\rangle, \tag*{(5)}

where δ\delta and β\beta are weighting factors that measure the field overlap between the twisted resonant modes and the untwisted resonant modes TM11p\mathrm{TM}_{11p} and TE11(p+1)\mathrm{TE}_{11(p+1)}. The exact forms of these expressions are given by

δ=∫E(r⃗)TM11p⋅E(r⃗)ψp+1± dV∫E(r⃗)TM11p⋅E(r⃗)TM11p dV,(6)\delta= \frac{\int\mathbf{E}(\vec{r})_{\mathrm{TM}_{11p}}\cdot\mathbf{E}(\vec{r})_{\psi_{p+1}^{\pm}}\,dV}{\int\mathbf{E}(\vec{r})_{\mathrm{TM}_{11p}}\cdot\mathbf{E}(\vec{r})_{\mathrm{TM}_{11p}}\,dV}, \tag*{(6)}

and

β=∫H(r⃗)TE11(p+1)⋅H(r⃗)ψp+1± dV∫H(r⃗)TE11(p+1)⋅H(r⃗)TE11(p+1) dV.(7)\beta= \frac{\int\mathbf{H}(\vec{r})_{\mathrm{TE}_{11(p+1)}}\cdot\mathbf{H}(\vec{r})_{\psi_{p+1}^{\pm}}\,dV}{\int\mathbf{H}(\vec{r})_{\mathrm{TE}_{11(p+1)}}\cdot\mathbf{H}(\vec{r})_{\mathrm{TE}_{11(p+1)}}\,dV}. \tag*{(7)}

where E(r⃗)ψp+1±\mathbf{E}(\vec{r})_{\psi_{p+1}^{\pm}} represents the electric field vector of the twisted ψp+1±\psi_{p+1}^{\pm} modes, and similar notation applies for the magnetic field vectors. The same notation is used for the untwisted modes.

Figure 4 depicts how the weighting factors of the first three ψp+1±\psi_{p+1}^{\pm} modes change as a function of ϕ\phi and the associated impact on their Hi\mathcal{H}_{i}. As expected, at ϕ=0\phi=0, β\beta dominates for the ψ1+\psi_{1}^{+} mode as it exists as a TE111\mathrm{TE}_{111} mode, and δ\delta dominates for the ψ1−\psi_{1}^{-} mode as it exists as a TM110\mathrm{TM}_{110} mode. As ϕ\phi is increased, the dominant weighting factor reduces in magnitude while its counterpart increases demonstrating how the dominant mode mixes with its partner mode, with a corresponding increase in the magnitude of the mode’s Hi\mathcal{H}_{i}. For the ψ1+\psi_{1}^{+} mode, β\beta is positive meaning that the EzE_{z} field of the TE111\mathrm{TE}_{111} mode is being mixed with the HzH_{z} of the TM110\mathrm{TM}_{110} mode, which in their untwisted forms are anti-parallel, therefore the vector product of the two results in a negative Hi\mathcal{H}_{i}. Whereas, for the ψ1−\psi_{1}^{-} mode, δ\delta and β\beta have opposite signs meaning that the EzE_{z} and HzH_{z} fields being mixed are parallel in their untwisted forms, therefore the vector product of the two results in a positive Hi\mathcal{H}_{i}. The same phenomenon occurs for the higher-order modes (e.g. ψ2−\psi_{2}^{-}), where the next pair of near degenerate modes, TM111\mathrm{TM}_{111} and TE112\mathrm{TE}_{112}, mix.

Weighting factors and helicity for the first three twisted modes

Figure 4. The weighting factors δ\delta (black) and β\beta (pink) of (a) ∣ψ1+⟩\lvert\psi_{1}^{+}\rangle and (b) ∣ψ1−⟩\lvert\psi_{1}^{-}\rangle, and the corresponding Hi\mathcal{H}_{i} (blue) as a function of ϕ\phi for the equilateral triangular cross-section resonator. Similarly, the δ\delta (red) and β\beta (green) weighting factors for the (c) ψ2−\psi_{2}^{-} mode.

The mode mixing gets stronger as ϕ\phi increases and reaches a maximum at the point at which δ\delta and β\beta are the closest in magnitude, corresponding to the strongest coupling strength, indicated by the red line in Figure 4. Note that in Figure 4(a), the dips in weighting factor and Hi\mathcal{H}_{i} of the ψ1+\psi_{1}^{+} mode are due to mode interactions with

ψp+1−\psi_{p+1}^{-} modes at certain twist angles, which can also be seen at points C and D in Figure 2(b). As a result of these interactions, the simulated data points deviate from the fitted lines for numerous ϕ\phi values and it is difficult to say with certainty at which ϕ\phi value a maximum Hi\mathcal{H}_{i} is achieved.

Experimental Validation

To validate the resonant frequency splitting caused by twisting in the triangular resonator, three copper cavity prototypes were 3D-printed using selective laser melting. The cavity dimensions were set to v=20v=20 mm, l=150l=150 mm, and therefore (AR)2=0.0177(AR)^2=0.0177, with one resonator having a right-handed twist (ϕ=10π/9)(\phi=10\pi/9), another a left-handed twist (ϕ=−10π/9)(\phi=-10\pi/9), and a third untwisted (ϕ=0)(\phi=0). These dimensions were chosen to ensure the prototypes were easily printable.

Transmission (S21)(S_{21}) measurements were obtained through vector network analysis across two magnetic field antenna placed at the cavity end faces, with results shown in Figure 5. As predicted the ψp+1\psi_{p+1} modes in the twisted resonators are shifted down in frequency compared to the original TE110\mathrm{TE}_{110} and TM111\mathrm{TM}_{111} modes in the untwisted resonator. By comparison, the lower order TE10\mathrm{TE}_{10} mode, which, as mentioned previously, does not undergo any mixing effect due to twisting, has the same resonant frequency in both the untwisted and twisted resonators. In addition, we observe that the frequency shift due to twisting is equal in magnitude for both the left and right handed resonators, confirming that the tuning effect is symmetric about ϕ=0\phi=0, as predicted in Figure 2(b), where the twist angles of the experimental cavities are shown by the grid lines. The observed frequency shift of the lowest frequency ψ\psi mode between the ϕ=10π/9\phi=10\pi/9 and ϕ=0\phi=0 resonators is measured to be −335-335 MHz, whilst the frequency shift predicted for AR=0.0177AR=0.0177, v=20v=20 mm from Figure 2(b) is ∼−379\sim-379 MHz – in reasonable agreement given the printed dimensions may not exactly match those that were simulated.

Experimental S21 transmission measurements for twisted and untwisted resonators

Figure 5. The experimental S21S_{21} transmission measurements taken from 3D printed copper resonators with dimensions v=20v=20 mm and l=150l=150 mm at twist angles of −10π/9-10\pi/9 (black), 10π/910\pi/9 (blue) and 00 (red), corresponding to the vertical cut lines in Figure 2(b). The gridlines indicate the corresponding locations of the resonant transmission peaks of the experimental data, specifically the modes TE10\mathrm{TE}_{10} for all twist angles, TM110\mathrm{TM}_{110} & TE111\mathrm{TE}_{111} for the untwisted resonator and ψ±\psi^{\pm}, for the twisted resonators are indicated by dashed lines.

Generation of Helicity due to Material Parameter κ\kappa

Electromagnetic helicity commonly appears when photons interact with chiral molecules or materials [37, 38, 39]. The chiral strength of these materials is quantified by the parameter κ\kappa, which modifies Maxwell’s equations by introducing an imaginary coupling term between the electric and magnetic fields. This magnetoelectric coupling results from κ\kappa breaking spatial symmetry, leading to a mixing of the perpendicular electric E\mathbf{E} and magnetic H\mathbf{H} fields in the material and therefore a non-zero Im⁡[E⋅H∗]\operatorname{Im}[\mathbf{E}\cdot\mathbf{H}^{*}] product. From the form of (1), this will result in a non-zero, hh, akin to the effect of twisting the resonator. The effect of κ\kappa is in contrast to a real, in-phase coupling between electric and magnetic fields, which is represented by the non-reciprocity parameter χ\chi, which rather than generating helical radiation, generates nonreciprocity. Specifically, χ\chi influences the phase of the propagating electric field without altering its polarisation, resulting in a time-dependent non-orthogonality with the magnetic field. On the other hand, κ\kappa affects the polarisation of the propagating field without changing its phase, leading to a spatial non-orthogonality with the magnetic field [40]. A positive κ\kappa indicates left-handed polarisation rotation in the direction of propagation, whilst a negative κ\kappa indicates right-handed rotation. A material with non-zero κ\kappa and ξ=0\xi=0 is referred to as a Pasteur or chiral medium, whilst a material with non-zero χ\chi but κ=0\kappa=0 is referred to as a Tellegen material. The general case with both parameters non-zero is referred to as “bi-isotropic”, which will have the following constitutive equations inside the medium:

D=ϵE+ξH,\mathbf{D}=\epsilon\mathbf{E}+\xi\mathbf{H},
B=ζE+μH,(8)\mathbf{B}=\zeta\mathbf{E}+\mu\mathbf{H}, \tag*{(8)}

where ϵ\epsilon is the dielectric permittivity of the medium, μ\mu is the magnetic permeability of the medium, D\mathbf{D} is the electric displacement field and H\mathbf{H} is the inductive magnetic field. The coupling between the electric and magnetic fields is characterised by the parameters:

ξ=(χ−iκ)μ0ϵ0,(9)\xi=(\chi-\mathrm{i}\kappa)\sqrt{\mu_0\epsilon_0}, \tag*{(9)}
ζ=(χ+iκ)μ0ϵ0,(10)\zeta=(\chi+\mathrm{i}\kappa)\sqrt{\mu_0\epsilon_0}, \tag*{(10)}

with ϵ0\epsilon_0 and μ0\mu_0 are the permittivity and permeability of free space, respectively. These constitutive equations are presented in the frequency domain.

Perturbation Theory

It should be noted that in the case of a mirror-symmetric resonator, a frequency shift will be produced by the introduction of a chiral material if the perturbed E\mathbf{E} and H\mathbf{H} fields in (16) are significantly altered from their original states. This will only be true when the companion field generated by the electro-magnetic coupling introduced by the media; i.e. iκμ0ϵ0Hz\mathrm{i}\kappa\sqrt{\mu_0\epsilon_0}H_z originating from an applied resonant EzE_z, is itself resonant. This is the case for the TM11p\mathrm{TM}_{11p} and TE11(p+1)\mathrm{TE}_{11(p+1)} modes in the cavities being discussed due to the equilateral cross section and hence their near-degenerate frequencies.

From perturbation theory, for a cavity resonator with bi-isotropic inclusions, there is a known relationship between the perturbed value κ1\kappa_1 and a shift in resonant frequency Δω=ω1−ω0\Delta\omega= \omega_1-\omega_0, where ω1\omega_1 is the new resonant frequency and ω0\omega_0 is the unperturbed cavity frequency. Many studies suggest this relationship as a means of measuring the value of κ\kappa [40, 41, 42]. This perturbation has previously only been considered for the inclusion of a small bi-isotropic sphere into a resonator resulting in small deviations around the isotropic properties κ0=0\kappa_0 = 0 and ξ0=0\xi_0 = 0 [40]. Here, we derive the relationship between Δω\Delta\omega and κ1\kappa_1 using perturbation around an arbitrary κ0\kappa_0 in terms of the mode helicity.

Consider a resonator filled with some bi-anistropic material that has some small change in the chiral material parameter Δκ=κ1−κ0\Delta\kappa= \kappa_1-\kappa_0. The unperturbed electric and magnetic fields E0\mathbf{E}_0, H0\mathbf{H}_0 and the perturbed fields E1\mathbf{E}_1, H1\mathbf{H}_1 inside the resonator all satisfy Maxwell’s equations (with no source terms), i.e.

∇×E=−jωBand∇×E∗=jωB∗,∇×H=jωDand∇×H∗=−jωD∗.(11)\nabla\times\mathbf{E}=-j\omega\mathbf{B}\quad\text{and}\quad\nabla\times\mathbf{E}^{*}=j\omega\mathbf{B}^{*}, \\ \nabla\times\mathbf{H}=j\omega\mathbf{D}\quad\text{and}\quad\nabla\times\mathbf{H}^{*}=-j\omega\mathbf{D}^{*}. \tag*{(11)}

The analysis begins with considering the two electromagnetic processes (E0,H0,D0,B0)(\mathbf{E}_0,\mathbf{H}_0,\mathbf{D}_0,\mathbf{B}_0) and (E1,H1,D1,B1)(\mathbf{E}_1,\mathbf{H}_1,\mathbf{D}_1,\mathbf{B}_1) at different oscillation frequencies ω0\omega_0 and ω1\omega_1, and considering the following combinations of vector fields,

∇⋅(E0∗×H1+E1×H0∗)=H1⋅∇×E0∗−E0∗⋅∇×H1+H0∗⋅∇×E1−E1⋅∇×H0∗(12)\nabla\cdot\left(\mathbf{E}_0^{*}\times\mathbf{H}_1+\mathbf{E}_1\times\mathbf{H}_0^{*}\right) =\mathbf{H}_1\cdot\nabla\times\mathbf{E}_0^{*} -\mathbf{E}_0^{*}\cdot\nabla\times\mathbf{H}_1 +\mathbf{H}_0^{*}\cdot\nabla\times\mathbf{E}_1 -\mathbf{E}_1\cdot\nabla\times\mathbf{H}_0^{*} \tag*{(12)}

Combining (8) and (11) with (12) we obtain,

∇⋅(E0∗×H1+E1×H0∗)=jω0μH0∗⋅H1+jω0ζ0E0∗⋅H1−jω1ϵE1⋅E0∗−jω1ξ1H1⋅E0∗−jω1μH1⋅H0∗−jω1ζ1E1⋅H0∗+jω0ϵE0∗⋅E1+jω0ξ0H0∗⋅E1(13)\begin{aligned} \nabla\cdot\left(\mathbf{E}_0^{*}\times\mathbf{H}_1+\mathbf{E}_1\times\mathbf{H}_0^{*}\right) ={}&j\omega_0\mu\mathbf{H}_0^{*}\cdot\mathbf{H}_1+j\omega_0\zeta_0\mathbf{E}_0^{*}\cdot\mathbf{H}_1\\ &-j\omega_1\epsilon\mathbf{E}_1\cdot\mathbf{E}_0^{*}-j\omega_1\xi_1\mathbf{H}_1\cdot\mathbf{E}_0^{*}-j\omega_1\mu\mathbf{H}_1\cdot\mathbf{H}_0^{*}\\ &-j\omega_1\zeta_1\mathbf{E}_1\cdot\mathbf{H}_0^{*}+j\omega_0\epsilon\mathbf{E}_0^{*}\cdot\mathbf{E}_1+j\omega_0\xi_0\mathbf{H}_0^{*}\cdot\mathbf{E}_1 \tag*{(13)} \end{aligned}

where ζ0\zeta_0 and ξ0\xi_0 are the unperturbed bi-isotropic parameters of the system (i.e. with κ0\kappa_0) and ζ1\zeta_1 and ξ1\xi_1 the perturbed parameters (i.e. with κ1\kappa_1).

Next we set the non-reciprocity parameters, χ0,χ1=0\chi_0,\chi_1=0, in order to considering a Pasteur material with non zero κ\kappa, then we apply perturbation theory where we assume H1=H0\mathbf{H}_1=\mathbf{H}_0 and E1=E0\mathbf{E}_1=\mathbf{E}_0. Given the complex Poynting vector, S0=1/2(E0×H0∗)\mathbf{S}_0=1/2(\mathbf{E}_0\times\mathbf{H}_0^{*}), and integrating over the volume, (13) becomes.

4∭Re⁡(∇⋅S0) dV=∭[−j(ω1−ω0)(μ∣H0∣2+ϵ∣E0∣2)+jϵ0μ0(ω1κ1−ω0κ0)2Im⁡[E0⋅H0∗]] dV(14)4\iiint\operatorname{Re}(\nabla\cdot\mathbf{S}_0)\,dV =\iiint\left[-j(\omega_1-\omega_0)(\mu|\mathbf{H}_0|^2+\epsilon|\mathbf{E}_0|^2) +j\sqrt{\epsilon_0\mu_0}(\omega_1\kappa_1-\omega_0\kappa_0)2\operatorname{Im}[\mathbf{E}_0\cdot\mathbf{H}_0^{*}]\right]\,dV \tag*{(14)}

In the ideal lossless scenario, the left hand side of (14) is zero, in this case, to first order in perturbations we may show,

δω=ω0δκ+δωκ02μrϵrH0(15)\delta\omega=\frac{\omega_0\delta\kappa+\delta\omega\kappa_0}{2\mu_r\epsilon_r}\mathcal{H}_0 \tag*{(15)}

and with some rearranging

δωω0=δκκ0(κ0H02μrϵr/(1−κ0H02μrϵr))(16)\frac{\delta\omega}{\omega_0}=\frac{\delta\kappa}{\kappa_0}\left(\frac{\kappa_0\mathcal{H}_0}{2\mu_r\epsilon_r}\middle/\left(1-\frac{\kappa_0\mathcal{H}_0}{2\mu_r\epsilon_r}\right)\right) \tag*{(16)}

where the helicity of the unperturbed mode, H0\mathcal{H}_0, is given by (2) and ϵr\epsilon_r and μr\mu_r are the relative permittivity and permeability of the materials respectively. In the case κ0≪1\kappa_0\ll1 then (16) becomes,

(δωω0)κ0≪1=δκH02μrϵr.(17)\left(\frac{\delta\omega}{\omega_0}\right)_{\kappa_0\ll1}=\frac{\delta\kappa\mathcal{H}_0}{2\mu_r\epsilon_r}. \tag*{(17)}

The ψ1−\psi^-_1 mode has H<0\mathcal{H}<0 for ϕ>0\phi>0, and since this corresponds to the handedness as radiation with κ<0\kappa<0, (17) predicts that the ψ1−\psi^-_1 mode will shift to a lower frequency, i.e., δωω0\frac{\delta\omega}{\omega_0}. This is consistent with the FEM-simulated eigenfrequencies shown in Figure 2(b).

Rearranging (16) we may also represent the helicity in term of the frequency perturbation as,

H0=2μrϵrω0δωδκ1+κ0ω0δωδκ(18)\mathcal{H}_0=\frac{\frac{2\mu_r\epsilon_r}{\omega_0}\frac{\delta\omega}{\delta\kappa}}{1+\frac{\kappa_0}{\omega_0}\frac{\delta\omega}{\delta\kappa}} \tag*{(18)}

Thus the helicity may be calculated either from the volume integral of the fields as indicated in (2) or through an incremental frequency rule as indicated in (18). This is very similar to the way electromagnetic filling factors of resonant modes may be numerically calculated, that is via volume integrals or an incremental frequency rule as indicated in [43].

From the form of (17), we expect a linear relationship between κ≃κeff\kappa\simeq\kappa_{\mathrm{eff}} and Δω/ω\Delta\omega/\omega if we assume Hi\mathcal{H}_i is a constant. We can verify this relationship by solving an electromagnetic FEM of an untwisted triangular resonator with modified Maxwell’s equations according to (8) and sweeping the chirality parameter, κeff\kappa_{\mathrm{eff}}. The resulting eigenfrequencies and Hi\mathcal{H}_i calculated from the eigenvectors are shown in Figure 6(a) for a v=20v=20 mm, l=150l=150 mm and ϕ=0\phi=0 triangular cross-section cavity. We observe that for a given eigenmode, once the Hi\mathcal{H}_i saturates to a constant (black lines), we indeed observe a linear relationship between frequency and κeff\kappa_{\mathrm{eff}} proportional to the Hi\mathcal{H}_i value. Importantly, (17) demonstrates the relationship between frequency shift and electromagnetic helicity, where the rate of change of frequency with respect to material chirality (δω/δκ)(\delta\omega/\delta\kappa) is determined by mode helicity at a given κ0\kappa_0.

Eigenfrequency solutions for the $v=20$ mm and $l=150$ mm cavity as a function of $\kappa_{\mathrm{eff}}$ for twist angles of (a) 0 and (b) $\frac{2\pi}{3}$

Figure 6. Eigenfrequency solutions for the v=20v = 20 mm and l=150l = 150 mm cavity as a function of the material parameter κeff\kappa_{\mathrm{eff}} for twist angles of (a) 0 and (b) 2π3\frac{2\pi}{3} where the colour is Hi\mathcal{H}_{i}. The red grid lines mark the κeff\kappa_{\mathrm{eff}} values at which there is zero Hi\mathcal{H}_{i}.

Furthermore, it is interesting to investigate the above situation in a twisted cavity, as shown in Figure 6(b) for ϕ=2π/3\phi=2\pi/3. We note that the symmetry around κeff=0\kappa_{\mathrm{eff}}=0 is broken in this case with a new symmetry axis (shown in red) existing at offset κeff\kappa_{\mathrm{eff}} values, which are different for the ψ1+/−\psi^{+/-}_1 modes and the TE10\mathrm{TE}_{10} modes. This result points to a mode-family-dependent equivalency between twisting and κ\kappa where the two effects have the potential to cancel one another at certain values. Importantly, the result of

Fig. 6(b) is that two materials with the same magnitude κeff\kappa_{\mathrm{eff}} but different signs (i.e. a left- and right-handed form of the same substance) will produce distinguishable spectra in a twisted cavity, whilst they will produce identical spectra in an untwisted cavity.

Equivalence between κeff\kappa_{\mathrm{eff}} and ϕ\phi

Given that Hi\mathcal{H}_{i} values resulting from twisting the triangular cross-section resonators are purely real (i.e. E⃗⋅H⃗∗\vec{\mathbf{E}} \cdot\vec{\mathbf{H}}^{*} is purely imaginary), we can safely assume that the process of twisting can be alternatively described by the introduction of some κ\kappa value for the propagation media whilst setting χ=0\chi= 0. We can demonstrate the explicit equivalence between a ϕ\phi twist and filling the cavity with some isotropic chiral media by firstly simulating these two distinct cases. From the solved eigenvectors we can plot Hi\mathcal{H}_{i} as a function of κeff\kappa_{\mathrm{eff}} and twist angle ϕ\phi as shown in Figure 7(a). We see that these two results agree when an appropriate, phenomenological scaling is applied to the xx-axis.

Plots of helicity as a function of twist angle and effective chirality

Figure 7. For the equilateral triangular cross-section resonator with dimensions of v=20v = 20 mm and l=150l = 150 mm: (a) Hi\mathcal{H}_{i} of the ψ1±\psi_{1}^{\pm} mode as a function of introduced ϕ\phi (red) and κeff\kappa_{\mathrm{eff}} (blue) to the system. (b) FEM simulated Hi\mathcal{H}_{i} values as both ϕ\phi and κeff\kappa_{\mathrm{eff}} are varied demonstrating null points where the helicity induced by one effect is cancelled by the other.

Furthermore, the true equivalence of the two mechanisms can be demonstrated through nullifying the Hi\mathcal{H}_{i} generated by twisting with an opposite acting κeff\kappa_{\mathrm{eff}}. This effectively represents how a uniform chiral media surrounded by conducting boundary conditions can be twisted to produce linearly polarised, un-coupled TE and TM modes and corresponds to the situation represented by the red κeff0\kappa_{\mathrm{eff0}} line in Figure 6(b) for ϕ=2π3\phi= \frac{2\pi}{3}. To accurately locate these null-points we must evaluate the integral of ∣h∣|h| over the volume, which is written as:

Hi=∫∣hp∣ dV.(19)\mathcal{H}_{i} = \int|h_{p}| \, dV. \tag*{(19)}

The reason for this is that artificial null points can arise due to mode interferences whereby Hi\mathcal{H}_{i} over one portion of the cavity volume will cancel Hi\mathcal{H}_{i} in the remaining volume and it is not strictly 0 at every point.

These “de-hybridisation” points are tracked in Figure 7(b). As expected, the κeff\kappa_{\mathrm{eff}} required to nullify Hi\mathcal{H}_{i} introduced by twisting is linearly proportional to ϕ\phi, where a left handed κeff\kappa_{\mathrm{eff}} cancels out a right-handed Hi\mathcal{H}_{i} and vice versa. Given that a right-handed Hi\mathcal{H}_{i} corresponds to a negative κeff\kappa_{\mathrm{eff}} but a positive ϕ\phi, a positive κeff\kappa_{\mathrm{eff}} is required to cancel a positive ϕ\phi. For the specific cavity dimensions used in the FEM simulation (v=20v = 20 mm, l=150l = 150 mm), ϕ≈55.8κeff\phi\approx55.8\kappa_{\mathrm{eff}} for the ψ1±\psi_{1}^{\pm} mode. Whilst the relationship between κ\kappa and δω\delta\omega has the analytical form of (17), the conversion between ϕ\phi and κ\kappa will be mode dependent and cavity geometry dependent, hence only phenomenological equivalences can be derived.

As indicated by (17), we expect a linear relationship between κ≃κeff\kappa\simeq\kappa_{\mathrm{eff}} if Hi\mathcal{H}_{i} is constant. It can be seen from Figure 7(a) that for the chosen cavity dimensions Hi\mathcal{H}_{i} of the ψ1±\psi_{1}^{\pm} modes saturates out for relatively small κeff\kappa_{\mathrm{eff}} values at ∣Hi∣=0.935|\mathcal{H}_{i}| = 0.935. In this region where Hi\mathcal{H}_{i} can be assumed constant, we would therefore expect the fractional frequency shift to be linearly dependent on κ\kappa with a coefficient of Hi/2\mathcal{H}_i/2. Indeed, when taking the eigenfrequency solutions of the FEM in which κeff\kappa_{\mathrm{eff}} is swept, we can derive the empirical relationship Δω/ω=0.453κeff\Delta\omega/\omega= 0.453\kappa_{\mathrm{eff}} for the ψ1−\psi_1^{-}, which is shown in Figure 8.

FEM simulated shift in eigenfrequencies for the $\psi_1^{\pm}$ resonant modes

Figure 8. The FEM simulated shift in eigenfrequencies for the ψ1±\psi_1^{\pm} resonant modes in the equilateral triangular resonator (blue) of dimensions w=20w = 20 mm and ℓ=150\ell= 150 mm as a function of κeff\kappa_{\mathrm{eff}}. For comparison, the eigenfrequencies for the square (red) and circle (black) cross-sections, which are inscribed within the triangle’s circumcircle, are also included.

Since the mode-mixing effect that causes a frequency shift is now proven to be equivalent to the introduction of a twist angle ϕ\phi to the metallic boundary conditions, we can explore how the cavity cross section plays a role. For a circular cross-section resonator, as a twist is introduced, there will be no impact on the boundary conditions and hence we anticipate no first-order increase in H\mathcal{H} and Δω/ω=0\Delta\omega/\omega= 0. It therefore stands to reason that for regular polygons of the dihedral group with the same circumcircle, the lowest symmetry order cross-section (i.e. the triangle) will generate the maximal possible frequency shift per change in ϕ\phi or κeff\kappa_{\mathrm{eff}}. As a further proof of equivalency between ϕ\phi and κeff\kappa_{\mathrm{eff}}, we investigate how changing κeff\kappa_{\mathrm{eff}} as a material parameter over the volume of different geometry cavities affects the resonant frequency of the ψ1±\psi_1^{\pm} modes, as shown in Figure 8. Whilst we don’t observe an absence of Δω/ω\Delta\omega/\omega for the circular cross-section cavity as we would expect for varying ϕ\phi, we do indeed observe a minimal amount compared to the maximal for the triangular case. This residual quadratic tuning is a result of higher-order effects that are not taken into consideration in the perturbation theory presented above (which is only to first-order). It is interesting to note that this tuning mechanism appears very similar to the frequency behaviour of the TE01\mathrm{TE}_{01} mode family (see Figure 2 and Figure 6), which would also not be expected to tune given there’s no degenerate TM counterpart them, in that it is quadratic and smaller in magnitude compared to the other modes.

Using the FEM simulations of Figure 2 we can estimate that a ∼20\sim20 GHz cavity with a twist angle of ϕ=2π\phi= 2\pi will result in Δω=−500\Delta\omega= -500 MHz for the ψ1−\psi_1^{-} mode. Therefore Δω/ω=−0.025\Delta\omega/\omega= -0.025 and such a cavity would have an equivalent chirality material parameter of κeff=−0.055\kappa_{\mathrm{eff}} = -0.055. For comparison, Cr2O3\mathrm{Cr}_2\mathrm{O}_3, the crystalline material that exhibits the strongest observed magneto-electric effect, only has a non-reciprocity parameter of χ=3.11⋅10−4\chi= 3.11 \cdot10^{-4}[26] (see appendix VI A). Note that χ\chi and κ\kappa are quadrature terms in the magneto-electric vector and hence it is reasonable to compare their magnitudes. The materials with the largest κ\kappa values (ranging from 3.9⋅10−53.9 \cdot10^{-5} to 0.340.34), at microwave frequencies, are man-made materials (such as metal helices in epoxy) and are very lossy [40], whereas we generate a modest κ\kappa value limited only by conductivity losses.

Generation of Helicity due to Axion Field

Another method of generating helical (chiral) radiation is to introduce the axion field θ\theta to Maxwell’s equations. The constitutive relation for electromagnetism will then be given in the time domain by [35]:

D⃗=ϵE⃗−θαπcμH⃗,\vec{\mathbf{D}} = \epsilon\vec{\mathbf{E}} - \frac{\theta\alpha}{\pi}c\mu\vec{\mathbf{H}},
cB⃗=θμαπE⃗+μH⃗,(20)c\vec{\mathbf{B}} = \frac{\theta\mu\alpha}{\pi}\vec{\mathbf{E}} + \mu\vec{\mathbf{H}}, \tag*{(20)}

where α=e2ℏc\alpha= \frac{e^2}{\hbar c} is the fine structure constant. The axion is calculated to couple to photons through the axion-electromagnetic chiral anomaly, whether in condensed matter physics or in QCD. The chiral anomaly is described by the following interaction terms, added to the Lagrangian of the photonic degrees of freedom of the standard model [44]:

Laγγ=gaγγαμ0E⃗⋅H⃗.(21)\mathcal{L}_{a\gamma\gamma} = g_{a\gamma\gamma}\alpha\mu_0\vec{\mathbf{E}}\cdot\vec{\mathbf{H}}. \tag*{(21)}

Here, the axion modification to the equations of motion are given by the product of the pseudoscalar, α\alpha, and the photon-axion coupling term gaγγg_{a\gamma\gamma} multiplied by E⋅H\mathbf{E}\cdot\mathbf{H}, which may look familiar as the numerator of Hi\mathcal{H}_i (2), multiplied by jj.

The square of the magnitude of Hi\mathcal{H}_i, ∣Hi∣2\lvert\mathcal{H}_i\rvert^2, is equivalent to the form factor that couples low-mass axions to photons as an amplitude modulation of a single mode through a parametric interaction. This axion-photon coupling due to the twisting of the resonator is explained in more detail in [10].

The introduction of the axion field θ\theta as a modification to Maxwell’s equations couples the electric and magnetic fields i.e. magneto-electric coupling, mixing them together. This can be thought of as a “rotation” of the electric and magnetic fields into each other, where the rotation angle is the mixing angle η\eta. The axion field θ\theta can be related to η\eta by [35, 36]:

η=tan⁡−1(αθμcπ).(22)\eta= \tan^{-1}\left(\frac{\alpha\theta\mu c}{\pi}\right). \tag*{(22)}

This mixing effect is precisely the effect we describe in Section I due to the twisting of the resonator, or in section II by introducing some chirality material parameter κeff\kappa_{\mathrm{eff}}.

We can establish a relationship between the angle θ\theta and E⃗′⋅H⃗′\vec{E}' \cdot\vec{H}', where the prime indicates the axion modified fields, by starting with the inverse of the previously discussed duality transformation (4):

(E⃗0cμ0H⃗0)=(−cμ0sin⁡(η)H⃗′+cos⁡(η)E⃗′cμ0cos⁡(η)H⃗′+sin⁡(η)E⃗′)(23)\begin{pmatrix} \vec{E}_0 \\ c\mu_0\vec{H}_0 \end{pmatrix} = \begin{pmatrix} -c\mu_0\sin(\eta)\vec{H}' + \cos(\eta)\vec{E}' \\ c\mu_0\cos(\eta)\vec{H}' + \sin(\eta)\vec{E}' \tag*{(23)} \end{pmatrix}

Given that the electric and magnetic fields are orthogonal in the untwisted (or κeff=0\kappa_{\mathrm{eff}} = 0 or θ=0\theta= 0) case, their dot product must be zero.

0=E⃗0⋅cμ0H⃗0=2cμ0E⃗′⋅H⃗′cos⁡2η+sin⁡2η(E⃗′⋅E⃗′−c2μ02H⃗′⋅H⃗′)(24)0 = \vec{E}_0 \cdot c\mu_0\vec{H}_0 = 2c\mu_0\vec{E}' \cdot\vec{H}'\cos2\eta+ \sin2\eta\left(\vec{E}' \cdot\vec{E}' - c^2\mu_0^2\vec{H}' \cdot\vec{H}'\right) \tag*{(24)}

By rearranging to solve for η\eta, we obtain:

η=12tan⁡−1(2cμ0E⃗′⋅H⃗′c2μ02H⃗′⋅H⃗′−E⃗′⋅E⃗′)(25)\eta= \frac{1}{2}\tan^{-1}\left(\frac{2c\mu_0\vec{E}' \cdot\vec{H}'}{c^2\mu_0^2\vec{H}' \cdot\vec{H}' - \vec{E}' \cdot\vec{E}'}\right) \tag*{(25)}

Relating this to the mixing angle (22), we derive the relationship between E⃗0′⋅H⃗0′\vec{E}_0' \cdot\vec{H}_0' and θ\theta:

θ=παμ0ctan⁡(12tan⁡−1(2cμ0E⃗′⋅H⃗′c2μ02H⃗′⋅H⃗′−E⃗′⋅E⃗′))(26)\theta= \frac{\pi}{\alpha\mu_0c}\tan\left(\frac{1}{2}\tan^{-1}\left(\frac{2c\mu_0\vec{E}' \cdot\vec{H}'}{c^2\mu_0^2\vec{H}' \cdot\vec{H}' - \vec{E}' \cdot\vec{E}'}\right)\right) \tag*{(26)}

By the form of (1), E⃗′⋅H⃗′\vec{E}' \cdot\vec{H}' in this expression is proportional to hh. This expression can be evaluated at each point within the resonator to qualitatively describe the axion field. For a quantitative measure of axion field strength, the integral of (26) can be taken over the volume of the resonator, giving the effective axion field strength θeff\theta_{\mathrm{eff}}.

Using the small angle approximation (tan⁡(x)≈arctan⁡(x)≈1\tan(x) \approx\arctan(x) \approx1), we can relate equations (25) and (26) through the following formula:

θ≈παμ0cη(27)\theta\approx\frac{\pi}{\alpha\mu_0c}\eta \tag*{(27)}

The effective axion field, θeff\theta_{\mathrm{eff}}, and ηeff\eta_{\mathrm{eff}} can be evaluated for the ψ1±\psi_1^{\pm} eigenmodes from the ϕ=2π/3\phi= 2\pi/3 twisted cavity simulation, along with the mode’s Hi\mathcal{H}_i, with results shown in Figure 9(a). In this figure, θeff\theta_{\mathrm{eff}} and ηeff\eta_{\mathrm{eff}} are indeed proportional to each other for a given Hi\mathcal{H}_i.

Plots of effective axion and chirality parameters

Figure 9. (a) The derived results (θeff\theta_{\mathrm{eff}} and ηeff\eta_{\mathrm{eff}}) are shown in relation to H\mathcal{H} for the ψ1±\psi_1^{\pm} resonant modes derived from the FEM-simulated twisting of an equilaterial triangular resonator with dimensions v=20v = 20 mm and l=150l = 150 mm. These derived results (θeff\theta_{\mathrm{eff}} and ηeff\eta_{\mathrm{eff}}) are shown in (b) as a function of twist angle ϕ\phi by the blue and black lines, respectively, along with the derived θeff\theta_{\mathrm{eff}} and ηeff\eta_{\mathrm{eff}} due to the introduction of the chirality material parameter κeff\kappa_{\mathrm{eff}} over the volume of the resonator, as shown by the green and red lines, respectively.

Therefore, we can state that for the twisted electromagnetic resonators, the act of twisting is equivalent to introducing some non-zero axion field θeff\theta_{\mathrm{eff}} to the radiation.

Discussion

We can analytically derive the equivalency of the chirality material parameter κ\kappa and an axion field θ\theta at any given point in the resonator by comparing the coefficients in the constitutive relations (8) and (20). This relationship is given by:

κ=−αμ0cπθ(28)\kappa= -\frac{\alpha\mu_0c}{\pi}\theta \tag*{(28)}

where the previously established relationship between η\eta and θ\theta (22) allows us to further derive that κ\kappa is related to η\eta by:

κ=−tan⁡η.(29)\kappa= -\tan\eta. \tag*{(29)}

Additionally, (25) connects mixing angle η\eta to the helicity density hh at any given point in the volume of the resonator. Equivalent forms of these relationships for the global effective parameters κeff\kappa_{\mathrm{eff}}, θeff\theta_{\mathrm{eff}}, and H\mathcal{H} can be attained by integrating their local forms over the volume of the resonator. Thus they can be written as:

ηeff=∫η dV,κeff=∫κ dV,(30)\begin{aligned} \eta_{\mathrm{eff}} &= \int\eta\,\mathrm{d}V,\\ \kappa_{\mathrm{eff}} &= \int\kappa\,\mathrm{d}V, \tag*{(30)} \end{aligned}

and we can see that by taking volume integrals of both sides of (28) will define θeff\theta_{\mathrm{eff}}. The theoretical links between κ\kappa, θ\theta, η\eta, and hh are consistent with the presented simulation results.

Our results show that twisting the boundary conditions of the cavity resonator introduces the same coupling mechanism as a non-zero, isotropic chirality material κ\kappa (κeff\kappa_{\mathrm{eff}}) or θeff\theta_{\mathrm{eff}}, which mixes the E⃗\vec{E} and H⃗\vec{H} fields, leading to the generation of non-zero electromagnetic helicity H\mathcal{H} in vacuum. Figure 9(b) illustrates the relationship between these coupling mechanisms for the ψ±\psi^{\pm} eigenmodes obtained from the ϕ=2π/3\phi= 2\pi/3 twisted cavity simulation. Introducing either a twist to the boundary conditions, ϕ\phi, or a non-zero effective chirality to the cavity volume, κeff\kappa_{\mathrm{eff}}, results in the generation of an effective mixing angle, ηeff\eta_{\mathrm{eff}}, between the electric field, E⃗p\vec{E}_{p}, and the magnetic field, H⃗p\vec{H}_{p}, of the resonant electromagnetic modes. This, in turn, leads to the emergence of an effective axion field, θeff\theta_{\mathrm{eff}}. Both mechanisms converge to similar values of θeff\theta_{\mathrm{eff}} and ηeff\eta_{\mathrm{eff}} for sufficiently large κeff\kappa_{\mathrm{eff}} and ϕ\phi (κeff>0.075\kappa_{\mathrm{eff}} > 0.075 and ϕ>5π4\phi> \frac{5\pi}{4}).

Each of these mechanisms introduces magneto-electric coupling that mixes the electric and magnetic fields by a mixing angle η\eta, resulting in a chiral electromagnetic field. The chirality is characterised by the helicity H\mathcal{H},

and the measurable effect of varying any of these three parameters is observed as a frequency shift in previously degenerate, orthogonal mode pairs. This frequency shift has been shown experimentally to agree very well with predictions.

Our approach provides a method to generate helical electromagnetic radiation in a monochromatic resonant mode in vacuo, which has potential applications in the measurement of material chirality. Moreover, if the twist angle of the system can be dynamically controlled, it is possible to actively manipulate and even reverse the electromagnetic helicity H\mathcal{H}, opening avenues for applications in material diagnostics, encryption, and communication.

Acknowledgments

This work was funded by the Australian Research Council Centre of Excellence for Engineered Quantum Systems, CE170100009 and Centre of Excellence for Dark Matter Particle Physics, CE200100008. E.C.P is partially funded through the Defence Science Centre Research Higher Degree Student Grant.

This work used the NCRIS and Government of South Australia enabled Australian National Fabrication Facility - South Australian Node (ANFF-SA).

Availability of Data

All data and FEM model parameters are available upon request to the authors.

Appendix

Chirality of Cr2_2O3_3

Whilst the chirality parameter κ\kappa is purely real, there is a purely imaginary version of this parameter called the four-dimensional relativistic invariant pseudoscalar α~\tilde{\alpha}. This parameter has been measured in Cr2_2O3_3 to be approximately 1.035 psm−1^{-1} at 285 K [26]. We can derive the relationship between the two chirality parameters to be

κ=cα~,(31)\kappa= c\tilde{\alpha}, \tag*{(31)}

by comparing the following two known equations for B in the context of magnetoelectric media [40, 45]

B=jκcE+μH,(32)\mathbf{B} = \frac{j\kappa}{c}\mathbf{E}+\mu\mathbf{H}, \tag*{(32)}

and

B=M+μ0μiH+αE.(33)\mathbf{B} = \mathbf{M}+\mu_0\mu_i\mathbf{H}+\alpha\mathbf{E}. \tag*{(33)}

Using (31) we can calculate the chirality parameter κeff\kappa_{\mathrm{eff}} to be 3.11 · 10−4^{-4} for Cr2_2O3_3 at 285 K.

Square Cross-section Resonator

For the resonator with the symmetric D4_4 pentagon (i.e. square), the degenerate modes that mix together with the introduction of a twist ϕ\phi are TE20p\mathrm{TE}_{20p} & TM21p\mathrm{TM}_{21p} (see 10(b)). The frequency tuning of the eigenmodes as a function of ϕ\phi can be seen in Figure 10(a) where, just like for the triangular case, H\mathcal{H} is generated with the introduction of a ϕ\phi, but doesn’t saturate at as high a value. This is because the square case is a closer approximation to a cylindrical resonator and hence twisting has less of an effect on the mode mixing. The generation of the new mode orthogonality basis is shown in Figure 10(b).

Eigenfrequencies and electromagnetic fields in twisted cavity resonators

Figure 10. The eigenfrequencies of the resonant modes in the square cross-section resonator with a side length of 25 mm as a function of ϕ\phi where colour corresponds to H\mathcal{H}. (b) The E⃗⊥\vec{E}_{\perp} (black), jH⃗⊥j\vec{H}_{\perp} (red) fields, and corresponding energy density of the axial fields EzE_z & HzH_z, for the (a) TM110\mathrm{TM}_{110} & (b) TE111\mathrm{TE}_{111} modes for the untwisted triangular cross-section resonator, and the ψ+\psi^{+} & ψ−\psi^{-} modes for the ϕ=2π\phi= 2\pi twisted resonator.

Symmetry of the Resonator Cross-Section

To optimise the mode mixing effect and hence κ\kappa generation, the polygon of the resonator cross-section must be symmetric, (i.e. equilaterial triangle, square, etc.), which is known as the dihedral group of regular polygons. This is because this symmetry tunes the TE and TM modes closest in frequency, and for a coupled mode system, hybridisation of the two individual systems is greatest when their frequencies are made degenerate. This is demonstrated in Figure 11(a) & (Figure 11b) where one of the vertex lengths of the triangular cross-section and the width of the square cross-section is multiplied by some factor R\mathcal{R}. For both cases, H\mathcal{H} is maximised when R=1\mathcal{R}=1, corresponding to the symmetric case of the respective cross-section.

Effect of cross-section scaling on helicity and frequency in triangular and square twisted cavity resonators

Figure 11. (a) The effect on H\mathcal{H} and fif_i as the vertex lengths of the triangular cross-section is multiplied by some factor R\mathcal{R} for different ϕ\phi values. For the untwisted resonator, the TE111\mathrm{TE}_{111} and TM110\mathrm{TM}_{110} fif_i are black & black-dashed respectively, and for ϕ=5π6\phi= \frac{5\pi}{6} and 2π2\pi, the ψ+\psi^{+} and ψ−\psi^{-} fif_i are blue and red respectively. For ϕ=2π\phi= 2\pi, H\mathcal{H} of the ψ+\psi^{+} & ψ−\psi^{-} states are gold and navy-blue respectively. (b) The effect on H\mathcal{H} and fif_i as the vertex lengths of the width of the square cross-section is multipled by some factor R\mathcal{R} for different ϕ\phi values. For the untwisted square cross-section resonator, the TE201\mathrm{TE}_{201} & TM210\mathrm{TM}_{210} modes fif_i are black. For ϕ=17π9\phi= \frac{17\pi}{9}, the ψ±\psi^{\pm} fif_i are blue & red respectively, and H\mathcal{H} is light red & purple respectively.

References

  1. [1]Zhaofeng Li, Mehmet Mutlu, and Ekmel Ozbay. Chiral metamaterials: from optical activity and negative refractive index to asymmetric transmission. Journal of Optics, 15(2):023001, February 2013.
  2. [2]Yiqiao Tang and Adam E. Cohen. Optical chirality and its interaction with matter. Phys. Rev. Lett., 104:163901, Apr 2010.pubmed.ncbi.nlm.nih.gov/20482049
  3. [3]Tonghan Zhao, Dejing Meng, Zhijian Hu, Wenjing Sun, Yinglu Ji, Jianlei Han, Xue Jin, Xiaochun Wu, and Pengfei Duan. Enhanced chiroptic properties of nanocomposites of achiral plasmonic nanoparticles decorated with chiral dye-loaded micelles. Nature Communications, 14:81, January 2023.DOI
  4. [4]Behrooz Semnani, Jeremy Flannery, Rubayet Al Maruf, and Michal Bajcsy. Spin-preserving chiral photonic crystal mirror. Light: Science & Applications, 9(1):23, December 2020.
  5. [5]Alberto Bottarelli, Massimo Frigerio, and Matteo G. A. Paris. Quantum routing of information using chiral quantum walks. AVS Quantum Science, 5(2):025001, June 2023.DOI
  6. [6]Yu You, Yiqi Hu, Gongwei Lin, Yihong Qi, Yueping Niu, and Shangqing Gong. Quantum nonreciprocity based on electromagnetically induced transparency in chiral quantum-optical systems. Phys. Rev. A, 103:063706, Jun 2021.
  7. [7]Bingnan Wang, Jiangfeng Zhou, Thomas Koschny, Thomas Koschny, Maria Kafesaki, Costas M. Soukoulis, and Costas M. Soukoulis. Chiral metamaterials: simulations and experiments. Journal of Optics, 11:114003, 2009.DOI
  8. [8]W. Z. Jia and L. F. Wei. Probing molecular chirality by coherent optical absorption spectra. Phys. Rev. A, 84:053849, Nov 2011.DOI
  9. [9]Lewis E. MacKenzie and Patrycja Stachelek. The twists and turns of chiral chemistry. Nature Chemistry, 13(6):521–522, January 2021.DOI
  10. [10]J. F. Bourhill, E. C. I. Paterson, M. Goryachev, and M. E. Tobar. Searching for ultralight axions with twisted cavity resonators of anyon rotational symmetry with bulk modes of nonzero helicity. Phys. Rev. D, 108:052014, Sep 2023.DOI
  11. [11]F. Alpeggiani, K. Y. Bliokh, F. Nori, and L. Kuipers. Electromagnetic helicity in complex media. Phys. Rev. Lett., 120:243605, Jun 2018.DOI
  12. [12]Konstantin Y. Bliokh, Yuri S. Kivshar, and Franco Nori. Magnetoelectric effects in local light-matter interactions. Phys. Rev. Lett., 113:033601, Jul 2014.pubmed.ncbi.nlm.nih.gov/25083644
  13. [13]Josep Martínez-Romeu, Iago Diez, Sebastian Golat, Francisco J. Rodríguez-Fortuño, and Alejandro Martínez. Chiral forces in longitudinally invariant dielectric photonic waveguides. Photon. Res., 12(3):431–443, Mar 2024.
  14. [14]Konstantin Y Bliokh, Aleksandr Y Bekshaev, and Franco Nori. Dual electromagnetism: helicity, spin, momentum, and angular momentum (2013 new j. phys. 15 033026). New Journal of Physics, 15:033026, 2015.
  15. [15]Mingkai Liu, David A. Powell, Ilya V. Shadrivov, Mikhail Lapine, and Yuri S. Kivshar. Spontaneous chiral symmetry breaking in metamaterials. Nature Communications, 5(1):4441, 2014.DOI
  16. [16]Alexander B. Khanikaev, S. Hossein Mousavi, Wang-Kong Tse, Mehdi Kargarian, Allan H. MacDonald, and Gennady Shvets. Photonic topological insulators. Nature Materials, 12(3):233–239, 2013.DOI
  17. [17]Maxim Goryachev and Michael E. Tobar. Reconfigurable microwave photonic topological insulator. Phys. Rev. Applied, 6:064006, Dec 2016.DOI
  18. [18]Konstantin Y. Bliokh, Daniel Leykam, Max Lein, and Franco Nori. Topological non-Hermitian origin of surface Maxwell waves. Nature Communications, 10:580, February 2019.DOI
  19. [19]Kirill Voronin, Alexey S. Taradin, Maxim V. Gorkunov, and Denis G. Baranov. Single-handedness chiral optical cavities. ACS Photonics, 9(8):2652–2659, 2022.arxiv.org/abs/2104.14230
  20. [20]Eric Plum and Nikolay I. Zheludev. Chiral mirrors. Applied Physics Letters, 106(22):221901, 06 2015.
  21. [21]Sergey A. Dyakov, Natalia S. Salakhova, Alexey V. Ignatov, Ilia M. Fradkin, and Nikolay A. Gippius. Chiral light in twisted Fabry-Pérot cavities. In David L. Andrews, Angus J. Bain, and Antonio Ambrosio, editors, Nanophotonics X, volume 12991 of Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, page 129911A, June 2024.
  22. [22]Joshua Feis, Dominik Beutel, Julian Köpfler, Xavier Garcia-Santiago, Carsten Rockstuhl, Martin Wegener, and Ivan Fernandez-Corbaton. Helicity-preserving optical cavity modes for enhanced sensing of chiral molecules. Phys. Rev. Lett., 124:033201, Jan 2020.DOI
  23. [23]Denis G. Baranov, Christian Schäfer, and Maxim V. Gorkunov. Toward molecular chiral polaritons. ACS Photonics, 10(8):2440–2455, 2023.DOI
  24. [24]Konstantin Baranov, Dmitry Vagin, and Maxim A. Gorlach. Effective chiral response of anisotropic multilayered metamaterials. Phys. Rev. B, 110:235150, Dec 2024.arxiv.org/abs/2409.18571
  25. [25]Christian Multunas, Andrew Grieder, Juncing Xu, Yuan Ping, and Ravishankar Sundararaman. Circular dichroism of crystals from first principles. Phys. Rev. Mater., 7:123801, Dec 2023.DOI
  26. [26]Friedrich W. Hehl, Yuri N. Obukhov, Jean-Pierre Rivera, and Hans Schmid. Relativistic nature of a magnetoelectric modulus of cr₂o₃ crystals: A four-dimensional pseudoscalar and its measurement. Phys. Rev. A, 77:022106, Feb 2008.DOI
  27. [27]R. D. Peccei and Helen R. Quinn. Cp conservation in the presence of pseudoparticles. Phys. Rev. Lett., 38:1440–1443, Jun 1977.
  28. [28]R. D. Peccei and Helen R. Quinn. Constraints imposed by CP conservation in the presence of pseudoparticles.
  29. [29]Steven Weinberg. A new light boson? Phys. Rev. Lett., 40:223–226, Jan 1978.DOI
  30. [30]F. Wilczek. Problem of strong p and t invariance in the presence of instantons. Phys. Rev. Lett., 40:279–282, Jan 1978.DOI
  31. [31]R. D. Peccei. The Strong CP problem and axions. Lect. Notes Phys., 741:3–17, 2008.DOI
  32. [32]Michael E. Tobar, Catriona A. Thomson, William M. Campbell, Aaron Quiskamp, Jeremy F. Bourhill, Benjamin T. McAllister, Eugene N. Ivanov, and Maxim Goryachev. Comparing instrument spectral sensitivity of dissimilar electromagnetic haloscopes to axion dark matter and high frequency gravitational waves. Symmetry, 14(10), 2022.DOI
  33. [33]Ana Lopez Moran, Juan Córcoles, Jorge Ruiz-Cruz, José Montejo-Garai, and Jesus Rebollar. Electromagnetic scattering at the waveguide step between equilateral triangular waveguides. Advances in Mathematical Physics, 2016:1–16, 01 2016.
  34. [34]Ana Lopez Moran, Jorge Ruiz-Cruz, Juan Córcoles, José Montejo-Garai, and Jesus Rebollar. Analytical expressions of the q-factor for the complete resonant mode spectrum of the equilateral triangular waveguide cavity. Electronics Letters, 55, 08 2019.
  35. [35]Axion Elektrodynamik, Topologiska Isolatorer, Andreas Asker, and Andreas Asker. Axion electrodynamics and measurable effects in topological insulators axion. 2018.
  36. [36]Josep Planelles. Axion Electrodynamics in Topological Insulators for beginners. arXiv e-prints, page arXiv:2111.07290, November 2021.arxiv.org/abs/2111.07290
  37. [37]Yiqiao Tang and Adam E. Cohen. Optical chirality and its interaction with matter. Phys. Rev. Lett., 104:163901, Apr 2010.pubmed.ncbi.nlm.nih.gov/20482049
  38. [38]Yiqiao Tang and Adam E. Cohen. Enhanced enantioselectivity in excitation of chiral molecules by superchiral light. Science, 332(6027):333–336, 2011.DOI
  39. [39]E. Hendry, T. Carpy, J. Johnston, M. Popland, R. V. Mikhaylovskiy, A. J. Lapthorn, S. M. Kelly, L. D. Barron, N. Gadegaard, and M. Kadodwala. Ultrasensitive detection and characterization of biomolecules using superchiral fields. Nature Nanotechnology, 5(11):783–787, 2010.DOI
  40. [40]Ismo Lindell, Ari Sihvola, Sergei Tretyakov, and Ari J. Viitanen. Electromagnetic waves in chiral and bi-isotropic media. Artech House, United Kingdom, 1994.
  41. [41]S. A. Tretyakov and A. J. Viitanen. Waveguide and resonator perturbation techniques measuring chirality and nonreciprocity parameters of biisotropic materials. IEEE Transactions on Microwave Theory Techniques, 43(1):222–225, January 1995.DOI
  42. [42]Jin Au Kong. Electromagnetic Wave Theory. Wiley, New York, 1990.
  43. [43]J. Krupka, K. Derzakowski, A. Abramowicz, M. E. Tobar, and R. G. Geyer. Use of whispering-gallery modes for complex permittivity determinations of ultra-low-loss dielectric materials. IEEE Transactions on Microwave Theory Techniques, 47(6):752–759, June 1999.DOI
  44. [44]Pierre Sivikiwie. Pierre Sivikiwie. Resonance axion search methods. Rev. Mod. Phys., 93:015004, Feb 2021.
  45. [45]Manfred Fiebig. TOPICAL REVIEW: Revival of the magnetoelectric effect. Journal of Physics D Applied Physics, 38(8):R123–R152, April 2005.

Paper details

vlnc:W_d4sr33nz
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