betaFeedback

WorksPhysics and Astronomy

Helicity-squared scaling of post-eikonal phases for electromagnetic and gravitational waves in Ricci-flat spacetimes

Abstract

Spin-optical descriptions of electromagnetic and gravitational waves isolate finite-frequency corrections that are odd in the helicity. We study the complementary, helicity-even part of the first post-eikonal correction for scalar, electromagnetic, and linearized gravitational waves on Ricci-flat backgrounds. In a parallel-propagated null tetrad, the direct Weyl coupling of the gravitational field projects onto the circular transverse-traceless states as the multiple 2Cmˉmmˉm2C_{\bar m m\bar m m} of the identity. The Ricci identity for the spin coefficient μμ then rewrites this projection as a total derivative plus optical terms. The helicity-even phase of a spin-ss wave, measured relative to the scalar wave, then obeys a helicity-squared law, Δ2=4Δ1Δ_2=4Δ_1, up to endpoint terms. The Weyl coupling is what completes the h2h^2 scaling in the gravitational case. For weak-deflection scattering by a Schwarzschild black hole we evaluate the relevant integrals along the ray and obtain Δs=−3πs2/(8Mωρ3)Δ_s=-3πs^2/(8Mωρ^3), with ρ=b/Mρ=b/M. The same coefficient follows independently, and without reference to any frame or gauge, from the Regge–Wheeler phase shifts combined with the asymptotics of the rotation functions dssld^l_{ss}. Direct numerical integration of the radial equations confirms it. At first order in MM the helicity-even phase difference vanishes, as required by the Born amplitudes. The helicity-odd spin-optical sector and the helicity-even post-eikonal sector are therefore organized by the first and second powers of the helicity, respectively.

Introduction

The high-frequency propagation of waves in curved spacetime has been studied since the geometrical-optics expansion of Isaacson, and the subject now includes systematic higher-order treatments of scalar, electromagnetic, and gravitational waves [1, 2, 3, 4, 5]. Spin optics and Berry geometry gave finite-frequency corrections a particularly clear form [6]. For circularly polarized electromagnetic and gravitational waves, polarization transport generates helicity-dependent corrections to the ray dynamics, among them the gravitational spin Hall effect [7, 8, 9, 10, 11, 12, 13, 14, 15]. These methods have been applied to Schwarzschild and Kerr, and the graviton Berry curvature has recently been obtained from quantum field theory [16, 17, 18].

The leading spin-optical correction is odd under helicity reversal, h→−hh \to-h. It is not, however, the whole first post-eikonal correction. At the same order in the inverse frequency the transport equation for the amplitude contains three kinds of terms: the Berry term, linear in the polarization connection and odd in hh; terms quadratic in the transverse derivatives of the polarization frame, which are even in hh; and, for gravitational waves only, a direct algebraic coupling to the background Weyl tensor,

(Wb)μν=2Cμανβbαβ.(1)(Wb)_{\mu\nu}=2C_{\mu\alpha\nu\beta}b^{\alpha\beta}. \tag*{(1)}

On a Ricci-flat background Maxwell’s equations have no term of this kind. Spin-optical treatments retain the first group of terms, and the helicity-even terms are usually left aside, either because they do not affect the ray trajectory at this order or because, taken separately, they depend on the choice of frame. Earlier work on polarized waves has accordingly focused on helicity-odd effects or on the general structure of the corrected field [4, 5, 19, 20], and beyond-eikonal corrections in Schwarzschild have been computed for scalar waves [21].

Two questions are left open. The first is how the helicity-even phase depends on the spin of the field. The second is whether the Weyl coupling gives gravitational waves a structure with no electromagnetic analogue at this order. Both bear on a structural statement that is well established for the helicity-odd sector. There the effective ray equations of gravitational waves have the same form as those derived from Maxwell’s equations [11], and fields of different spin differ only by a constant factor, the helicity, in the spin tensor [14]; methods developed for electromagnetic waves are carried over to gravitational waves by accounting for the difference in helicity [12]. Whether a comparable statement holds for the helicity-even sector, where the Weyl coupling is the only algebraic contact between the spin-2 field and the background curvature, has not been examined. To our knowledge, the spin dependence of the helicity-even post-eikonal phase has not been determined.

In this paper we show that the helicity-even sector obeys a helicity-squared law, and that the Weyl coupling is precisely what makes this law hold for gravity. Projected onto the circular transverse-traceless (TT) states, the Weyl operator is 2Cmˉmˉmm2C_{\bar{m}\bar{m}m m} times the identity, so it neither converts nor splits the two gravitational helicities. In a parallel-propagated null tetrad, the Ricci identity for the spin coefficient μ\mu rewrites CmˉmˉmmC_{\bar{m}\bar{m}m m} as a total derivative plus products of optical scalars. The frame-derivative terms contain a connection part weighted by s2s^{2} and an optical part weighted by ss. The two weights coincide for s=1s=1, and for s=2s=2 the Weyl coupling supplies exactly their difference. The helicity-even phase of a spin-ss wave, measured relative to the scalar wave on the same congruence, is then s2s^{2} times a common bulk integral plus endpoint terms. An analysis of the field strengths at the detector shows that the endpoint terms cancel, apart from local terms, so the law holds for the gauge-invariant phases. The helicity-odd sector is linear in hh, and the helicity-even sector is quadratic. In practical terms, the helicity-even phase of a gravitational wave follows from the electromagnetic one through the factor h2=4h^{2}=4, without a separate evaluation of the Weyl term.

We then evaluate the helicity-even sector for weak-deflection scattering by a Schwarzschild black hole, in two independent ways. One calculation integrates the bulk quantities along the scattered ray. The other uses the Regge–Wheeler phase shifts and the asymptotics of the rotation functions dsssd^{s}_{ss}, and involves no choice of frame or gauge. Both give

Δs=−3π8s2Mωρ3+O(1Mωρ4),ρ=bM,(2)\Delta_{s}=-\frac{3\pi}{8}\frac{s^{2}}{M\omega\rho^{3}}+\mathcal{O}\left(\frac{1}{M\omega\rho^{4}}\right),\qquad\rho=\frac{b}{M}, \tag*{(2)}

where bb is the impact parameter and ω\omega the frequency. The first order in MM cancels, as required by the exact Born amplitudes [22], and direct numerical integration of the radial equations confirms the second-order coefficient.

The main contributions of this work are therefore the following: (i) the identity that relates the algebraic Weyl projection to the optical scalars of the congruence, and the helicity-squared law that follows from it; (ii) the endpoint analysis that extends the law to the gauge-invariant field strengths; and (iii) an explicit coefficient, obtained by two independent methods, which also serves as an analytic benchmark for numerical treatments of wave propagation in black-hole spacetimes. The effect itself is far below present observational sensitivity (Sec. VII); the interest of the result lies in the structure it reveals.

The paper is organized as follows. § 0.2 collects the post-eikonal equations for s=0,1,2s=0,1,2 and the algebraic Weyl projection. Section III projects the first post-eikonal source in a parallel null tetrad. Section IV derives the helicity-squared law and treats the endpoint terms. § 0.5 evaluates the bulk integrals in Schwarzschild, and Sec. VI gives the partial-wave calculation. Section VII relates the result to spin optics and to wave-optics studies of lensing by rotating black holes, and Sec. VIII summarizes. Two appendixes contain the weak-deflection evaluation of the Weyl integral and the angular phase of dssℓd^{\ell}_{ss}.

We use signature (−,+,+,+)(-,+,+,+) and the curvature convention [∇c,∇b]Va=RadcbVd[\nabla_{c},\nabla_{b}]V^{a}=R^{a}{}_{dcb}V^{d}, for which the linearized Einstein equation takes the form ((II.1)).

Post-eikonal equations for s=0,1,2s=0,1,2

Consider a Ricci-flat background, Rμν=0R_{\mu\nu}=0. A massless scalar obeys □ϕ=0\Box\phi=0. In Lorenz gauge the Maxwell potential obeys □Aμ−RμνAν=0\Box A_{\mu}-R_{\mu}{}^{\nu}A_{\nu}=0, which reduces to □Aμ=0\Box A_{\mu}=0. For a traceless Lorenz-gauge metric perturbation, linearized gravity gives

□hμν+2Cμανβhαβ=0(3)\Box h_{\mu\nu}+2C_{\mu\alpha\nu\beta}h^{\alpha\beta}=0 \tag*{(3)}

[1, 3, 5]. The three cases may be written together as □u+Wsu=0\Box u+\mathcal{W}_{s}u=0, with W0=W1=0\mathcal{W}_{0}=\mathcal{W}_{1}=0 and W2=W\mathcal{W}_{2}=\mathcal{W}.

We use the WKB ansatz

u=eiS/ϵ(b0+ϵb1+O(ϵ2)),kμ=∇μS.(4)u=e^{iS/\epsilon}\left(b_{0}+\epsilon b_{1}+\mathcal{O}(\epsilon^{2})\right),\qquad k_{\mu}=\nabla_{\mu}S. \tag*{(4)}

At leading order kμkμ=0k^{\mu}k_{\mu}=0, so the rays are affinely parametrized null geodesics, kν∇νkμ=0k^{\nu}\nabla_{\nu}k^{\mu}=0. With the transport operator

D≡2kρ∇ρ+∇ρkρ,(5)\mathcal{D}\equiv2k^{\rho}\nabla_{\rho}+\nabla_{\rho}k^{\rho}, \tag*{(5)}

the next two orders give

Db0=0,iDb1+□b0+Wsb0=0.(6)\mathcal{D}b_{0}=0,\qquad i\mathcal{D}b_{1}+\Box b_{0}+\mathcal{W}_{s}b_{0}=0. \tag*{(6)}

The gauge conditions must be imposed order by order. At leading order they make b0b_0 transverse (and traceless for s=2s=2). At the next order they fix the longitudinal part of b1b_1, for example ikμb1μν=−∇μb0μνik^\mu b_{1\mu\nu}=-\nabla^\mu b_{0\mu\nu} for gravity. Let mμm^\mu be a complex transverse null vector with m⋅mˉ=1m\cdot\bar m=1, parallel transported along the rays. The circular states of helicity h=±sh=\pm s are

e(+s)=m⊗s,e(−s)=mˉ⊗s,(7)e^{(+s)}=m^{\otimes s},\qquad e^{(-s)}=\bar m^{\otimes s}, \tag*{(7)}

and the leading amplitude is b0=Be(h)b_0=\mathcal{B}e^{(h)}. Here B\mathcal{B} is real and obeys the flux law 2k⋅∇B+(∇⋅k)B=02k\cdot\nabla\mathcal{B}+(\nabla\cdot k)\mathcal{B}=0.

Algebraic Weyl projection

For s=2s=2, the matrix elements of W\mathcal{W} between the circular TT states are

eˉ(±)μνWμναβeαβ(∓)=0,eˉ(+)μνWμναβeαβ(+)=eˉ(−)μνWμναβeαβ(−)=2Cmˉmmˉm.(8)\bar e^{\mu\nu}_{(\pm)}\mathcal{W}_{\mu\nu}{}^{\alpha\beta}e^{(\mp)}_{\alpha\beta}=0,\qquad\bar e^{\mu\nu}_{(+)}\mathcal{W}_{\mu\nu}{}^{\alpha\beta}e^{(+)}_{\alpha\beta}=\bar e^{\mu\nu}_{(-)}\mathcal{W}_{\mu\nu}{}^{\alpha\beta}e^{(-)}_{\alpha\beta}=2C_{\bar m m\bar m m}. \tag*{(8)}

The off-diagonal elements vanish because the Weyl tensor is antisymmetric in each index pair. The diagonal elements coincide by pair symmetry, and CmˉmmˉmC_{\bar m m\bar m m} is real. Therefore

PTTWPTT=2Cmˉmmˉm1TT.(9)P_{\mathrm{TT}}\mathcal{W}P_{\mathrm{TT}}=2C_{\bar m m\bar m m}\mathbf{1}_{\mathrm{TT}}. \tag*{(9)}

The direct Weyl coupling produces neither +2↔−2+2\leftrightarrow-2 conversion nor a splitting of the two helicities: it is helicity even. The scalar CmˉmmˉmC_{\bar m m\bar m m} is invariant under m→eiψmm\to e^{i\psi}m. It is not invariant under a null rotation m→m+ζkm\to m+\zeta k, which preserves all inner products of mm but mixes in components of the Weyl tensor along kk. Such a null rotation corresponds, at leading order, to a residual gauge transformation of b0b_0. For this reason the integral of CmˉmmˉmC_{\bar m m\bar m m} along a ray is not by itself an observable. The next sections show how it combines with the derivative terms.

Projected transport in a parallel null tetrad

Frame gradients

Complete (k,m,mˉ)(k,m,\bar m) to a null tetrad (k,n,m,mˉ)(k,n,m,\bar m), with k⋅n=−1k\cdot n=-1, parallel transported along the rays. Since k=∇Sk=\nabla S, the tensor ∇akb\nabla_a k_b is symmetric. We use the projections

ρ=mamˉb∇bka,σ=mamb∇bka,μ=namb∇bmˉa,λ=namˉb∇bmˉa,(10)\rho=m^a\bar m^b\nabla_bk_a,\qquad\sigma=m^am^b\nabla_bk_a,\qquad\mu=n_am^b\nabla_b\bar m^a,\qquad\lambda=n_a\bar m^b\nabla_b\bar m^a, \tag*{(10)}

which correspond to the Newman–Penrose spin coefficients [23], up to signature-dependent signs. Here ρ\rho is real. The polarization connection is

Aa=−imˉb∇amb,(11)A_a=-i\bar m^b\nabla_am_b, \tag*{(11)}

a real covector with kaAa=0k^aA_a=0 for a parallel frame. Using m⋅m=0m\cdot m=0, m⋅mˉ=1m\cdot\bar m=1 and k⋅m=0k\cdot m=0, the gradient of mm decomposes as

∇ρmμ=iAρmμ−(n⋅∇ρm)kμ+(mν∇ρkν)nμ.(12)\nabla_\rho m^\mu=iA_\rho m^\mu-(n\cdot\nabla_\rho m)k^\mu+(m^\nu\nabla_\rho k_\nu)n^\mu. \tag*{(12)}

Writing Am=AamaA_m=A_am^a, it follows that

AaAa=2∣Am∣2,Km≡∇ρmˉμ∇ρmμ=2∣Am∣2+2Re⁡(ρμ+σλ).(13)A_aA^a=2|A_m|^2,\qquad K_m\equiv\nabla_\rho\bar m_\mu\nabla^\rho m^\mu=2|A_m|^2+2\operatorname{Re}(\rho\mu+\sigma\lambda). \tag*{(13)}

Projected source

Projecting □(Bm⊗s)\Box(\mathcal{B}m^{\otimes s}) onto mˉ⊗s\bar m^{\otimes s} and using mˉ⋅□m=i∇⋅A−Km\bar m\cdot\Box m=i\nabla\cdot A-K_m, we obtain

mˉ⊗s⋅□(Bm⊗s)=Es+iGs,(14)\bar m^{\otimes s}\cdot\Box(\mathcal{B}m^{\otimes s})=E_s+iG_s, \tag*{(14)}
Es=□B−B[sKm+s(s−1)AaAa],(15)E_s=\Box\mathcal{B}-\mathcal{B}\left[sK_m+s(s-1)A_aA^a\right], \tag*{(15)}
Gs=2sAa∇aB+sB∇aAa.(16)G_s=2sA^a\nabla_a\mathcal{B}+s\mathcal{B}\nabla_aA^a. \tag*{(16)}

For helicity −s-s the sign of GsG_s is reversed. The projections onto the opposite helicity vanish, because m⋅m=0m\cdot m=0 and m⋅∇ρm=0m\cdot\nabla_\rho m=0. For s=2s=2 these expressions reproduce the decomposition D±±=E±iGD_{\pm\pm}=E\pm iG, D+−=D−+=0D_{+-}=D_{-+}=0 of the gravitational TT source.

Phase and amplitude

Write the TT component of the first correction as mˉ⊗s⋅b1=iBφs\bar m^{\otimes s}\cdot b_1=iB\varphi_s. Since k⋅∇mˉ=0k\cdot\nabla\bar m=0, the projection commutes with D\mathcal{D}, and Eq. (II.4) gives

dφsdλ=Es±iGs2B+csCmˉmˉmˉmˉ,c0=c1=0,c2=1.(17)\frac{d\varphi_s}{d\lambda}=\frac{E_s\pm iG_s}{2B}+c_s C_{\bar m\bar m\bar m\bar m},\qquad c_0=c_1=0,\quad c_2=1. \tag*{(17)}

Since u≃BeiS/ϵ(1+iϵφs)e(h)u\simeq B e^{iS/\epsilon}(1+i\epsilon\varphi_s)e^{(h)}, the real part of φs\varphi_s corrects the phase and the imaginary part the amplitude. The amplitude correction ∓Gs/(2B)\mp G_s/(2B) is linear in AA and odd in the helicity. It is the field-level counterpart of the Berry term responsible for the spin-optical corrections [10, 11, 15]. The phase correction is helicity even.

The scalar term □B/(2B)\Box B/(2B) is common to all spins. It also dominates the approach to the caustic of the scattered congruence. We therefore consider the difference Δs≡Φs−Φ0\Delta_s\equiv\Phi_s-\Phi_0 between the post-eikonal phases of the spin-ss and scalar waves on the same congruence. From Eqs. (III.4)–(III.8),

dΔsdλ=ϵ[−s2∣Am∣2−sRe⁡(ρμ+σλ)+csCmˉmˉmˉmˉ].(18)\frac{d\Delta_s}{d\lambda}=\epsilon\left[-s^2|A_m|^2-s\operatorname{Re}(\rho\mu+\sigma\lambda)+c_s C_{\bar m\bar m\bar m\bar m}\right]. \tag*{(18)}

Helicity-squared law

Ricci identity

Differentiate μ\mu along the ray and commute the derivatives. Using k⋅∇n=k⋅∇m=0k\cdot\nabla n=k\cdot\nabla m=0 and mb∇bkc=ρmc+σmˉc−(n⋅V)kcm^b\nabla_b k^c=\rho m^c+\sigma\bar m^c-(n\cdot V)k^c, with Vc=mb∇bkcV^c=m^b\nabla_b k^c, one finds

dμdλ=−(ρμ+σλ)+Cnmˉkm.(19)\frac{d\mu}{d\lambda}=-(\rho\mu+\sigma\lambda)+C_{n\bar mkm}. \tag*{(19)}

Tracelessness of the Weyl tensor, contracted with gab=−kanb−nakb+mamˉb+mˉambg^{ab}=-k^a n^b-n^a k^b+m^a\bar m^b+\bar m^a m^b, gives Cmˉmˉmˉmˉ=−2Re⁡CnmˉkmC_{\bar m\bar m\bar m\bar m}=-2\operatorname{Re}C_{n\bar mkm}. Hence

Cmˉmˉmˉmˉ=−2dRe⁡μdλ−2Re⁡(ρμ+σλ).(20)C_{\bar m\bar m\bar m\bar m}=-2\frac{d\operatorname{Re}\mu}{d\lambda}-2\operatorname{Re}(\rho\mu+\sigma\lambda). \tag*{(20)}

The algebraic projection of the Weyl tensor that enters the gravitational equation is thus a total derivative plus the same optical combination that appears in KmK_m.

Two consequences follow at once. First, at order MM the products ρμ\rho\mu and σλ\sigma\lambda are of second order. The integral of CmˉmˉmˉmˉC_{\bar m\bar m\bar m\bar m} along the ray then reduces to the boundary term −2[Re⁡μ]-2[\operatorname{Re}\mu], which vanishes for scattering because Re⁡μ\operatorname{Re}\mu decays at both ends. This is the structural reason why the nominal ρ−2\rho^{-2} contribution of the Weyl term integrates to zero (Appendix A).

Second, inserting Eq. (IV.2) into Eq. (III.9) and integrating along the ray, we obtain

Δs=ϵ{s2(12IC−IA)+s[Re⁡μ]inout+(cs−s(s−1)2)IC}.(21)\Delta_s=\epsilon\left\{s^2\left(\frac{1}{2}I_C-I_A\right)+s[\operatorname{Re}\mu]_{\mathrm{in}}^{\mathrm{out}}+\left(c_s-\frac{s(s-1)}{2}\right)I_C\right\}. \tag*{(21)}

where

IC=∫Cmˉmˉmˉmˉ dλ,IA=∫∣Am∣2 dλ.(22)I_C=\int C_{\bar m\bar m\bar m\bar m}\,d\lambda,\qquad I_A=\int|A_m|^2\,d\lambda. \tag*{(22)}

The last term vanishes both for Maxwell (c1=0c_1=0) and for linearized gravity (c2=1c_2=1). Therefore

Δs=ϵs2(12IC−IA)+ϵs[Re⁡μ]inout,s=1,2,(23)\Delta_s=\epsilon s^2\left(\frac{1}{2}I_C-I_A\right)+\epsilon s[\operatorname{Re}\mu]_{\mathrm{in}}^{\mathrm{out}},\qquad s=1,2, \tag*{(23)}

and, in particular,

Δ2=4Δ1−2ϵ[Re⁡μ]inout.(24)\Delta_2=4\Delta_1-2\epsilon[\operatorname{Re}\mu]_{\mathrm{in}}^{\mathrm{out}}. \tag*{(24)}

Equation (IV.5) is the central structural result of this work. Without the Weyl coupling, the optical term in Eq. (III.9) would carry the coefficient ss, while the connection term carries s2s^2. For s=1s=1 the two coincide. For s=2s=2 the direct Weyl coupling contributes exactly the missing s(s−1)/2s(s-1)/2 times ICI_C. The Weyl term is therefore not an independent helicity-even structure superimposed on the electromagnetic one: it is the term that makes the helicity-even sector of the gravitational wave scale as h2h^2. The operator structure is familiar from (∇−ihA)2=∇2−2ihA⋅∇−h2A2(\nabla-ihA)^2=\nabla^2-2ihA\cdot\nabla-h^2A^2. The linear term underlies spin optics, and the quadratic term is its helicity-even partner.

Frame dependence

The separate integrals in Eq. (IV.5) are defined in the parallel frame fixed on the incident screen. Under a local rotation m→eiψmm \to e^{i\psi}m, with B→Be−isψ\mathcal{B} \to\mathcal{B}e^{-is\psi}, the field b0b_0 is unchanged, and so is the total phase. Under a parallel null rotation m→m+ζkm \to m+\zeta k with constant ζ\zeta, ICI_C changes only at relative order ρ−1\rho^{-1} (Appendix A). The same holds for IAI_A: since δAm=−i(ζρ−ζˉσ)\delta A_m=-i(\zeta\rho-\bar{\zeta}\sigma), with ζ=O(M/b)\zeta=O(M/b) and ρ,σ=O(M)\rho,\sigma=O(M), the change is of the same relative order.

The TT phase Φs\Phi_s is not itself a gauge-invariant quantity. The invariant observables are field strengths evaluated at the detector, and they differ from the TT phase by endpoint terms that do not accumulate along the ray. For Maxwell, the Newman–Penrose scalar FnmˉF_{n\bar{m}} follows directly from Eq. (II.2):

Fnmˉ=−iϵBeiS/ϵ[1+iϵ(φ1+n⋅∇ln⁡B−Re⁡μ)−ϵ(An+Im⁡μ)+O(ϵ2)].(25)F_{n\bar{m}}=-\frac{i}{\epsilon}\mathcal{B}e^{iS/\epsilon}\left[1+i\epsilon\left(\varphi_1+n\cdot\nabla\ln\mathcal{B}-\operatorname{Re}\mu\right)-\epsilon\left(A_n+\operatorname{Im}\mu\right)+O(\epsilon^2)\right]. \tag*{(25)}

Endpoint terms for the gravitational amplitude

Equation Eq. (IV.7) shows that, for Maxwell, the phase of a field-strength component is the TT phase plus local terms, and that the local term −Re⁡μ-\operatorname{Re}\mu cancels the boundary contribution [Re⁡μ]inout[\operatorname{Re}\mu]_{\mathrm{in}}^{\mathrm{out}} in Δ1\Delta_1; for an incident plane wave Re⁡μin=0\operatorname{Re}\mu_{\mathrm{in}}=0. We now carry out the analogous calculation for linearized gravity.

Use tetrad indices 1=k1=k, 2=n2=n, 3=m3=m, 4=mˉ4=\bar{m}, with Tab…T_{ab\ldots} denoting the contraction of a tensor with the tetrad vectors after differentiation. For helicity +2+2,

hμν=eiS/ϵ(b0μν+ϵb1μν+O(ϵ2)),b0μν=Bmμmν.(26)h_{\mu\nu}=e^{iS/\epsilon}\left(b_{0\mu\nu}+\epsilon b_{1\mu\nu}+O(\epsilon^2)\right),\qquad b_{0\mu\nu}=\mathcal{B}m_\mu m_\nu. \tag*{(26)}

The only nonvanishing component of b0b_0 is b0 44=Bb_{0\,44}=\mathcal{B}, and the TT component of b1b_1 is b1 44=iBφ2b_{1\,44}=i\mathcal{B}\varphi_2. In vacuum the linearized Weyl tensor equals the linearized Riemann tensor, so that, up to background-curvature terms of relative order ϵ2\epsilon^2,

δCnmˉnmˉ=δR2424=12(∇2∇4h24+∇4∇2h24−∇4∇4h22−∇2∇2h44).(27)\delta C_{n\bar{m}n\bar{m}}=\delta R_{2424}=\frac{1}{2}\left(\nabla_2\nabla_4h_{24}+\nabla_4\nabla_2h_{24}-\nabla_4\nabla_4h_{22}-\nabla_2\nabla_2h_{44}\right). \tag*{(27)}

Each term follows from

∇a∇bhcd=eiS/ϵ[−1ϵ2kakbbcd+iϵ(∇akb bcd+kb∇abcd+ka∇bbcd)+O(1)],(28)\nabla_a\nabla_bh_{cd}=e^{iS/\epsilon}\left[-\frac{1}{\epsilon^2}k_ak_b b_{cd}+\frac{i}{\epsilon}\left(\nabla_a k_b\,b_{cd}+k_b\nabla_a b_{cd}+k_a\nabla_b b_{cd}\right)+O(1)\right], \tag*{(28)}

with b=b0+ϵb1b=b_0+\epsilon b_1. Although b0 24b_{0\,24} and b0 22b_{0\,22} vanish, their derivatives need not, because the frame varies. In the first two terms of Eq. (IV.9) the ϵ−2\epsilon^{-2} term vanishes because k⋅mˉ=0k\cdot\bar{m}=0; in the ϵ−1\epsilon^{-1} bracket only the contraction k⋅n=−1k\cdot n=-1 survives, and mˉb∇bb0 cd ncmˉd=Bn⋅∇mˉm=Bμˉ\bar{m}^b\nabla_b b_{0\,cd}\,n^c\bar{m}^d=\mathcal{B}n\cdot\nabla_{\bar{m}}m=\mathcal{B}\bar{\mu}. Hence

∇2∇4h24+∇4∇2h24=eiS/ϵ[−2iϵBμˉ+O(1)].(29)\nabla_2\nabla_4h_{24}+\nabla_4\nabla_2h_{24}=e^{iS/\epsilon}\left[-\frac{2i}{\epsilon}\mathcal{B}\bar{\mu}+O(1)\right]. \tag*{(29)}

The third term is O(1)O(1), since b0 22=0b_{0\,22}=0 and mˉ⋅k=0\bar{m}\cdot k=0. In the last term, (n⋅∇b0)44=n⋅∇B+2iBAn(n\cdot\nabla b_0)_{44}=n\cdot\nabla\mathcal{B}+2i\mathcal{B}A_n, and

∇2∇2h44=eiS/ϵ[−Bϵ2−iϵ(Bφ2+2n⋅∇B−Bnanb∇akb+4iBAn)+O(1)].(30)\nabla_2\nabla_2h_{44}=e^{iS/\epsilon}\left[-\frac{\mathcal{B}}{\epsilon^2}-\frac{i}{\epsilon}\left(\mathcal{B}\varphi_2+2n\cdot\nabla\mathcal{B}-\mathcal{B}n^an^b\nabla_a k_b+4i\mathcal{B}A_n\right)+O(1)\right]. \tag*{(30)}

Collecting terms,

δCnmˉnmˉ=B2ϵ2eiS/ϵ[1+iϵ(φ2+2n⋅∇ln⁡B−nanb∇akb−2Re⁡μ)−ϵ(4An+2Im⁡μ)+O(ϵ2)].(31)\delta C_{n\bar{m}n\bar{m}}=\frac{\mathcal{B}}{2\epsilon^2}e^{iS/\epsilon}\left[1+i\epsilon\left(\varphi_2+2n\cdot\nabla\ln\mathcal{B}-n^an^b\nabla_a k_b-2\operatorname{Re}\mu\right)-\epsilon\left(4A_n+2\operatorname{Im}\mu\right)+O(\epsilon^2)\right]. \tag*{(31)}

On a curved background the linearized Weyl tensor changes under a gauge transformation by LξCabcd\mathcal{L}_{\xi}C_{abcd}. For transformations compatible with the WKB form, ξ=ϵeiS/ϵξ0\xi=\epsilon e^{iS/\epsilon}\xi_0, this change is of relative order ϵ2\epsilon^2, so the phase in Eq. (IV.13) is gauge invariant at the order considered.

The structure is that of Eq. (IV.7), with the weight s=2s=2 in front of n⋅∇ln⁡B−Re⁡μn\cdot\nabla\ln\mathcal{B}-\operatorname{Re}\mu. The terms 2n⋅∇ln⁡B2n\cdot\nabla\ln\mathcal{B} and nanb∇akbn^an^b\nabla_a k_b are local: they are evaluated at the detector, do not accumulate along the ray, and decay in the far zone. The term −2Re⁡μ-2\operatorname{Re}\mu cancels the boundary contribution 2[Re⁡μ]inout2[\operatorname{Re}\mu]_{\mathrm{in}}^{\mathrm{out}} in Eq. (IV.6). This cancellation is exact. It does not rely on the decay of Re⁡μ\operatorname{Re}\mu, which is guaranteed only at first order in MM: at second order, on the outgoing branch before the caustic, the right-hand side of Eq. (IV.1) is dominated by the approximately constant product σλ\sigma\lambda, and Re⁡μ\operatorname{Re}\mu grows with the distance from the lens. For both fields, therefore, the accumulated phase of the invariant field strength, measured relative to the scalar wave, is

Δsinv=ϵs2(12IC−IA),s=1,2,(32)\Delta^{\mathrm{inv}}_s=\epsilon s^2\left(\frac{1}{2}I_C-I_A\right),\qquad s=1,2, \tag*{(32)}

up to local terms at the detector, and Δ2inv=4Δ1inv\Delta^{\mathrm{inv}}_2=4\Delta^{\mathrm{inv}}_1. The non-TT part of b1b_1 does not affect this result. The bulk calculation of § 0.5 and the partial-wave calculation of Sec. VI therefore evaluate the same invariant quantity.

Schwarzschild: bulk evaluation

From now on we use coordinates in units of MM, with ϵ=1/(Mω)\epsilon=1/(M\omega) and ρ=b/M\rho=b/M, and consider the weak-deflection regime ρ≫1\rho\gg1. The turning point rmr_m satisfies rm=b−M+O(M2/b)r_m=b-M+O(M^2/b), so the leading coefficients below are the same whether expressed in bb or in rmr_m.

Weyl integral

In the parallel frame fixed on the incident screen, the weak-deflection evaluation of Appendix A gives

IC=−3π8ρ3+O(ρ−4).(33)I_C=-\frac{3\pi}{8\rho^3}+O(\rho^{-4}). \tag*{(33)}

The straight-line contribution at order ρ−2\rho^{-2} is a total derivative and cancels between past and future infinity. The first nonvanishing term comes from the bending of the trajectory.

Connection integral

Commuting derivatives in the transport of Eq. (III.2) gives

dAadλ=−(∇akb)Ab+iCmmˉka.(34)\frac{\mathrm{d}A_a}{\mathrm{d}\lambda}=-(\nabla_a k^b)A_b+iC_{m\bar mka}. \tag*{(34)}

Since Aa=O(M)A_a=O(M) and ∇akb=O(M)\nabla_a k^b=O(M), the first term is of second order, and ∣Am∣2\lvert A_m\rvert^2 can be computed to O(M2)O(M^2) along the unperturbed ray. Take k=∂t+∂zk=\partial_t+\partial_z, the ray through (x,y,z)=(b,0,λ)(x,y,z)=(b,0,\lambda), and m=(x^+iy^)/2m=(\hat{x}+i\hat{y})/\sqrt{2}. In the static frame the vacuum Weyl tensor is determined by its electric part, Eij=(M/r3)(δij−3ninj)E_{ij}=(M/r^3)(\delta_{ij}-3n_i n_j), and

Cmmˉkm=Eijz^imj=−3Mbλ2r5.(35)C_{m\bar mkm}=E_{ij}\hat{z}^{i}m^{j}=-\frac{3Mb\lambda}{\sqrt{2}r^5}. \tag*{(35)}

Integrating Eq. (V.2) from the incident screen, where Aa=0A_a=0,

Am=iMb2r3,IA=M2b22∫−∞∞dλ(b2+λ2)3=3π16ρ3+O(ρ−4).(36)A_m=\frac{iMb}{\sqrt{2}r^3},\qquad I_A=\frac{M^2b^2}{2}\int_{-\infty}^{\infty}\frac{\mathrm{d}\lambda}{(b^2+\lambda^2)^3}=\frac{3\pi}{16\rho^3}+O(\rho^{-4}). \tag*{(36)}

The identities Cmmˉmmˉ=EzzC_{m\bar m m\bar m}=E_{zz}, Cmˉmmmˉ=−2Re⁡CnmˉkmC_{\bar m m m\bar m}=-2\operatorname{Re}C_{n\bar mkm} and Cmmˉkm=EzmC_{m\bar mkm}=E_{zm} used here have been checked numerically with the full vacuum Weyl tensor.

Result

Combining Eqs. (IV.5), (V.1) and (V.4),

Δs=ϵs2(−3π16−3π16)ρ−3=−3π8s2Mωρ3+O(1Mωρ4).(37)\Delta_s=\epsilon s^2\left(-\frac{3\pi}{16}-\frac{3\pi}{16}\right)\rho^{-3}=-\frac{3\pi}{8}\frac{s^2}{M\omega\rho^3}+O\left(\frac{1}{M\omega\rho^4}\right). \tag*{(37)}

The Weyl integral contributes, through the Ricci identity, one half of the electromagnetic coefficient; the connection integral contributes the other half. The gravitational coefficient is four times the electromagnetic one.

Schwarzschild: partial-wave evaluation

We now compute the same quantity from the scattering amplitudes. Nothing in this section depends on a choice of frame, tetrad, or gauge.

Constraint at first order in MM

In the static linearized field hμν=2U diag⁡(1,1,1,1)h_{\mu\nu}=2U\,\operatorname{diag}(1,1,1,1), with U=M/rU=M/r, the Born amplitude is determined by hμνTμν=2U(T00+Tii)h_{\mu\nu}T^{\mu\nu}=2U(T^{00}+T^{ii}). For a massless scalar T00+Tii→2ω2T^{00}+T^{ii}\to2\omega^2, independent of the angle. For the Maxwell field T00+Tii=2T00→2ω2cos⁡2(θ/2)T^{00}+T^{ii}=2T^{00}\to2\omega^2\cos^2(\theta/2) between states of equal helicity. For gravitons the helicity-preserving amplitude carries cos⁡4(θ/2)\cos^4(\theta/2) [22, 24]. Hence

fs(1)(θ)=cos⁡2sθ2f0(1)(θ),cos⁡2θ2=1−q24ω2,(38)f_s^{(1)}(\theta)=\cos^{2s}\frac{\theta}{2}f_0^{(1)}(\theta),\qquad\cos^2\frac{\theta}{2}=1-\frac{q^2}{4\omega^2}, \tag*{(38)}

where q=2ωsin⁡(θ/2)q=2\omega\sin(\theta/2). Since f0(1)∝ω2/q2f_0^{(1)}\propto\omega^2/q^2, the difference fs(1)−f0(1)f_s^{(1)}-f_0^{(1)} is a polynomial in q2q^2. It is a contact term in impact-parameter space. At first order in MM there is therefore no spin-dependent phase at b≠0b\ne0, to all orders in 1/ω1/\omega.

Phase shifts and angular functions

In Schwarzschild spacetime the helicity-preserving amplitude takes the standard form [25, 26, 27]

fs(θ)=12iω∑l(2l+1)e2iδl(s)dss l(θ)(39)f_s(\theta)=\frac{1}{2i\omega}\sum_l(2l+1)e^{2i\delta_l^{(s)}}d_{ss}^{\,l}(\theta) \tag*{(39)}

The phase shifts δl(s)\delta_l^{(s)} are those of the Regge–Wheeler equation d2ψ/dr∗2+(ω2−Vs)ψ=0\mathrm{d}^2\psi/\mathrm{d}r_*^2+(\omega^2-V_s)\psi=0, with

Vs=f[l(l+1)r2+(1−s2)2Mr3],f=1−2Mr.(40)V_s=f\left[\frac{l(l+1)}{r^2}+(1-s^2)\frac{2M}{r^3}\right],\qquad f=1-\frac{2M}{r}. \tag*{(40)}

For s=2s=2 the axial and polar phase shifts are related by Chandrasekhar’s transformation [28]. Their relative phase is of order Mω/l4M\omega/l^4, far beyond the order considered here, so helicity reversal does not enter.

We count orders in ϵ=1/(Mω)\epsilon=1/(M\omega) at fixed ρ\rho, with L=l+12=ωbL=l+\frac{1}{2}=\omega b, and keep all terms of first order in ϵ\epsilon. These are the terms (Mω)n/Ln+1=ϵ/ρn+1(M\omega)^n/L^{n+1}=\epsilon/\rho^{n+1}.

Radial part. Relative to ω2\omega^2, the spin-dependent part of VsV_s is of order ϵ2\epsilon^2. Its effect on the phase shift is therefore given exactly, at order ϵ\epsilon, by first-order WKB perturbation:

2δl(s)−2δl(0)=2s2M∫rt∞drr3ω2−fL2/r2=2s2Mωb2J(Mb),(41)2\delta_l^{(s)}-2\delta_l^{(0)}=2s^2M\int_{r_t}^{\infty}\frac{\mathrm{d}r}{r^3\sqrt{\omega^2-fL^2/r^2}}=\frac{2s^2M}{\omega b^2}\mathcal{J}\left(\frac{M}{b}\right), \tag*{(41)}

with

J(μ)=∫0utu du1−u2+2μu3=1+3π4μ+O(μ2),(42)\mathcal{J}(\mu)=\int_0^{u_t}\frac{u\,\mathrm{d}u}{\sqrt{1-u^2+2\mu u^3}}=1+\frac{3\pi}{4}\mu+O(\mu^2), \tag*{(42)}

where utu_t is the root near unity. We obtained the coefficient 3π/43\pi/4 numerically, to ten significant figures, by Richardson extrapolation. Higher-order WKB corrections and the Langer term are spin independent at this order.

Angular part. The functions dss ld_{ss}^{\,l} obey an equation whose potential differs from that of PlP_l by s2/cos⁡2(θ/2)s^2/\cos^2(\theta/2) (Appendix B). In the oscillatory region Lθ≫1L\theta\gg1, their phase differs from that of PlP_l by

Φsang−Φ0ang=−s2Ltan⁡θ2.(43)\Phi_s^{\mathrm{ang}}-\Phi_0^{\mathrm{ang}}=-\frac{s^2}{L}\tan\frac{\theta}{2}. \tag*{(43)}

Stationary phase

For attractive scattering the dominant branch of Eq. (VI.2) is e2iδl+iLθe^{2i\delta_l+iL\theta}, and the stationary point is fixed by θ=−2 dδl/dL\theta=-2\,d\delta_l/dL. At fixed scattering angle, and to first order in the spin-dependent pieces, the displacement of the stationary point does not affect the phase. Therefore

arg⁡fs(θ)−arg⁡f0(θ)=s2ω[2Mb2J(Mb)−1btan⁡θ(b)2].(44)\arg f_s(\theta)-\arg f_0(\theta)=\frac{s^2}{\omega}\left[\frac{2M}{b^2}\mathcal{J}\left(\frac{M}{b}\right)-\frac{1}{b}\tan\frac{\theta(b)}{2}\right]. \tag*{(44)}

with the null deflection angle

θ(b)=4Mb+15π4M2b2+O(M3b3).(45)\theta(b)=\frac{4M}{b}+\frac{15\pi}{4}\frac{M^2}{b^2}+O\left(\frac{M^3}{b^3}\right). \tag*{(45)}

The terms of first order in MM cancel, 2M/b2−2M/b2=02M/b^2-2M/b^2=0, as required by Eq. (VI.1). At second order,

arg⁡fs−arg⁡f0=s2(3π2−15π8)1Mωρ3=−3π8s2Mωρ3+O(1Mωρ4,ϵ3).(46)\arg f_s-\arg f_0=s^2\left(\frac{3\pi}{2}-\frac{15\pi}{8}\right)\frac{1}{M\omega\rho^3}=-\frac{3\pi}{8}\frac{s^2}{M\omega\rho^3}+O\left(\frac{1}{M\omega\rho^4},\epsilon^3\right). \tag*{(46)}

in exact agreement with Eq. (V.5). The factor s2s^2 enters here through the coefficient (1−s2)(1-s^2) of the Regge–Wheeler potential and through the angular potential s2/cos⁡2(θ/2)s^2/\cos^2(\theta/2).

Equation (VI.7) holds at first order in ϵ\epsilon for all ρ\rho. Its values, computed with the exact deflection angle and the exact integral J\mathcal{J}, are listed in Table I. They approach −3π/8-3\pi/8 with a correction of order ρ−1\rho^{-1}; a Richardson extrapolation of the last two entries gives −1.178088-1.178088.

ρ=b/M\rho=b/MJ(1/ρ)\mathcal{J}(1/\rho)θ(b)\theta(b)Mωρ3Δs/s2M\omega\rho^3\Delta s/s^2
101.3694194080.5903957876−3.019852
201.1426978470.2361359954−1.739966
401.0644441420.1081040765−1.412052
801.0307661350.0519284942−1.285951
1601.0150464660.0254708775−1.229990
4001.0059409740.0100743045−1.198394
8001.0029578040.0050184915−1.188171
16001.0014757540.0025046124−1.183116
32001.0007370930.0012511518−1.180602

Table I. Invariant post-eikonal phase difference in Schwarzschild, Eq. (VI.7), evaluated with the exact deflection angle θ(b)\theta(b) and the exact integral J\mathcal{J}. The last column is Mωρ3(arg⁡fs−arg⁡f0)/s2M\omega\rho^3(\arg f_s-\arg f_0)/s^2, whose limit is −3π/8=−1.178097…-3\pi/8=-1.178097\ldots

Numerical checks

We integrated the Regge–Wheeler equations for s=0s=0 and s=1s=1 numerically. Each integration starts inside the centrifugal barrier with the growing WKB solution and continues to r=4000Mr=4000M, where the phase is read off with the local wave number. The resulting phase-shift differences are compared with Eq. (VI.4) in Table II. The relative differences are between 4×10−64\times10^{-6} and 5×10−55\times10^{-5}. At b≃20Mb\simeq20M the second-order part of Eq. (VI.4) accounts for about 12%12\% of the total, so it is confirmed to better than 10−310^{-3}. The angular phase (VI.6) was checked against the zeros of d11ld^l_{11} and d22ld^l_{22} relative to those of PlP_l, for l=400l=400 and θ=0.05\theta=0.05, 0.20.2, with agreement at the level of 10−410^{-4}. The Python scripts used for these checks and for Table I were written with the assistance of an AI tool (see the Acknowledgments). They use standard libraries (SciPy and mpmath) and are provided upon request.

MωM\omegallb/Mb/MnumericalEq. (VI.4)relative difference
1020020.055.682852×10−45.682852\times10^{-4}5.682874×10−45.682874\times10^{-4}−3.8×10−6-3.8\times10^{-6}
1040040.051.327065×10−41.327065\times10^{-4}1.327125×10−41.327125\times10^{-4}−4.5×10−5-4.5\times10^{-5}
2040020.022.849049×10−42.849049\times10^{-4}2.849075×10−42.849075\times10^{-4}−9.2×10−6-9.2\times10^{-6}

Table II. Phase-shift difference 2(δl(1)−δl(0))2(\delta_l^{(1)}-\delta_l^{(0)}) from direct integration of the Regge–Wheeler equations, compared with the first-order WKB expression, Eq. (VI.4).

Discussion

Helicity-odd and helicity-even sectors

The first post-eikonal correction separates cleanly under helicity reversal. The helicity-odd part, carried by GsG_s, is linear in the polarization connection and in hh. It underlies the spin Hall effect [10, 11, 15, 17]. The helicity-even part, measured relative to the scalar wave, is quadratic in hh for both the electromagnetic and the gravitational field, Eq. (IV.5). The direct Weyl coupling does not produce a separate gravitational structure. Through the Ricci identity it is precisely what upgrades the optical term from ss to s2s^2. A description of the gravitational wave obtained by weighting the electromagnetic helicity-even sector with h2h^2 is therefore complete at this order, whereas a weighting linear in ss would miss exactly the contribution of the Weyl term.

The coefficient in Eq. (I.2) has a simple origin in each calculation. Along the ray, half of it comes from the Weyl integral and half from the transverse twist of the parallel frame. In partial waves it is the difference between the second-order radial term, 3π/23\pi/2, and the kinematic term produced by the angular functions at the second-order deflection, 15π/815\pi/8.

Relation to partial-wave and wave-optics studies

Partial-wave treatments of scalar, electromagnetic and gravitational scattering by black holes are well developed [25, 26, 27]. Wave-optics studies of gravitational lensing by rotating black holes have recently been carried out with the Teukolsky equation. In particular, Kubota et al. [29] solved the Teukolsky equation for gravitational waves emitted by a binary and lensed by a Kerr black hole, in the regime where the wavelength is comparable to the gravitational radius. They found a helicity-dependent, small-period oscillation in the power spectrum of the forward amplification factor, which grows with the black-hole spin for prograde configurations. That work addresses the full wave problem numerically, in a regime where MωM\omega is of order unity, and its central effect distinguishes the two helicities. The present work addresses a different question. We work in the high-frequency, weak-deflection regime, isolate the helicity-even correction at first post-eikonal order, and compare fields of different spin on the same background. Our result is an analytic coefficient that depends on s2s^2 and is the same for both helicities. Low-frequency comparisons between Teukolsky scattering and scattering amplitudes [24] are also complementary to ours.

Two further lines of work deserve comment. Beyond-geometric-optics corrections to gravitational lensing have been studied for gravitational waves at linear order in the lens potential [20] and, for scalar waves, up to second order in GG [21]. In the scalar case the first correction appears at order G2G^2 and grows with the distance from the lens. This is the behavior of the common term □B/(2B)\Box B/(2B), which cancels in Δs\Delta_s. The spin-dependent integrands considered here decay as r−6r^{-6}, and their integrals converge. The absence of a spin-dependent phase at order GG is consistent with the linear-order analysis of Ref. [20].

In the scattering-amplitude literature, the bending angle of massless particles has been computed at one loop, and it contains spin-dependent terms of the form bℏsℏG2M/b3b\hbar^s \hbar G^2M/b^3 [30, 31]. These terms originate in quantum loops, are linear in MM, and are obtained for ℏω≪M\hbar\omega\ll M at first order in ℏ\hbar. The effect found here is of a different nature. It is a classical wave effect, proportional to G2M2/ωG^2M^2/\omega. In particle variables, with E=ℏωE=\hbar\omega, it contributes to the deflection angle only at order ℏ2\hbar^2, beyond the truncation of Refs. [30, 31]. At the classical level the helicity-preserving eikonal phase of general relativity is universal at second post-Minkowskian order [32]. Finally, a helicity-even response of gravitons has recently been identified in a quantum-geometric gradient expansion [33]. That effect is sourced by the Ricci tensor and vanishes in vacuum, whereas the effect described here arises entirely from the Weyl curvature.

Magnitude of the effect

For a gravitational wave, Eq. (I.2) gives ∣Δ2∣=3π/(2Mωρ3)|\Delta_2|=3\pi/(2M\omega\rho^3). With GM⊙/c3≃4.9×10−6 sGM_\odot/c^3\simeq4.9\times10^{-6}\,\mathrm{s}, a wave of frequency f=100 Hzf=100\ \mathrm{Hz} passing a black hole of 106M⊙10^6M_\odot has Mω≃3×103M\omega\simeq3\times10^3. At ρ=10\rho=10 this gives ∣Δ2∣∼10−6|\Delta_2|\sim10^{-6} rad, and at ρ=100\rho=100, ∣Δ2∣∼10−9|\Delta_2|\sim10^{-9} rad. For a black hole of 108M⊙10^8M_\odot the corresponding values are 10−810^{-8} and 10−1110^{-11} rad. At ρ=10\rho=10 the higher-order terms in 1/ρ1/\rho increase these numbers by a factor of about 2.6 (Table I), which does not change the conclusion. For cosmological lenses the relevant impact parameters are of the order of the Einstein radius, b/M≳109b/M\gtrsim10^9, and the effect is entirely negligible. Larger values would require Mω≲1M\omega\lesssim1, where the post-eikonal expansion no longer applies and the full wave problem must be solved [29]. The helicity-even phase difference is therefore far below the phase resolution of present and planned detectors. It also enters the observed frequency-dependent phase together with the spin-independent term □B/(2B)\Box B/(2B), which is larger and grows with the distance from the lens [21].

The interest of the result is thus structural rather than observational. Equation (IV.5) fixes how the helicity-even part of the first post-eikonal correction depends on the spin of the field, and it complements the helicity-odd spin-optical sector. The coefficient −3πs2/8-3\pi s^2/8, together with the vanishing of the first-order term, also provides an analytic benchmark for numerical treatments of wave propagation in black-hole spacetimes, such as Teukolsky or wave-optics lensing codes, in the high-frequency regime.

Kerr

The derivation of Sec. IV uses only the vacuum Einstein equations and a parallel tetrad, so Eq. (IV.5) holds in Kerr spacetime as well. In the weak-deflection expansion the rotation parameter first enters the Weyl integral at order ρ−4\rho^{-4} (Appendix A). Gravitomagnetic corrections to AmA_m are of relative order a/ba/b, so they also affect IAI_A only at order ρ−4\rho^{-4}. The leading coefficient −3πs2/8-3\pi s^2/8 is therefore expected to be unchanged for a≠0a \ne0. An invariant confirmation, and the rotation-dependent terms at order ρ−4\rho^{-4}, require an invariant evaluation, for example from the Teukolsky equation with spin-weighted spheroidal harmonics. This is left for future work.

Conclusions

We have studied the helicity-even part of the first post-eikonal correction for scalar, electromagnetic, and linearized gravitational waves on Ricci-flat backgrounds. The direct Weyl coupling acts on the circular TT states as 2Cmmˉmmˉ2C_{m\bar m m\bar m} times the identity, so it neither converts nor splits the helicities. In a parallel null tetrad the Ricci identity for μ\mu rewrites CmmˉmmˉC_{m\bar m m\bar m} as a total derivative plus the optical combination −2Re⁡(ρμ+σλ)-2\operatorname{Re}(\rho\mu+\sigma\lambda). With this identity the helicity-even phase of a spin-ss wave, measured relative to the scalar wave, equals s2s^2 times a common bulk integral plus endpoint terms; for the invariant field strengths the endpoint terms cancel, apart from local terms at the detector. In the gravitational case the Weyl coupling supplies the part of the s2s^2 scaling that the frame-derivative terms alone would miss.

For weak-deflection scattering by a Schwarzschild black hole the helicity-even phase difference vanishes at first order in MM and at second order is Δs=−3πs2/(8Mωρ3)\Delta_s=-3\pi s^2/(8M\omega\rho^3). We obtained this coefficient in two ways: by integrating the bulk quantities along the ray, and from the Regge–Wheeler phase shifts together with the asymptotics of dssℓd^\ell_{ss}. The second route involves no frame or gauge choice, and direct numerical integration of the radial equations confirms it. The spin-optical and post-eikonal corrections of electromagnetic and gravitational waves are therefore organized by the first and second powers of the helicity. Extending the invariant evaluation to the rotation-dependent terms in Kerr spacetime remains open.

Acknowledgments

The author wishes to express sincere gratitude to the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and Fundação Cearense de Apoio ao Desenvolvimento Científico e Tecnológico (FUNCAP) for their valuable support through grants Project UNI-00210-00230.01.00/23 (FUNCAP), 303438/2026-6 (CNPq/PQ) and 420854/2025-8 (CNPq).

Use of AI tools.—This work made use of Claude (Anthropic, model claude-opus-5-5), accessed via the Claude desktop application in September 2026. Based on author prompts, the tool was used to assist with analytical derivations, numerical checks via Python scripts, and drafting and editing the manuscript. All analytical results were independently verified by the author, and all numerical outputs were reproduced by rerunning the scripts. The author takes full responsibility for the content of this paper.

Data Availability Statement

The code used in this work is available from the corresponding author upon reasonable request.

Conflict of Interest Statement

The author declares that there is no conflict of interest regarding the publication of this paper.

Weak-deflection evaluation of the Weyl integral

We work in units of MM and normalize the energy of the reference null geodesic to unity. For Schwarzschild, and for the equatorial Kerr problem with α=a/M\alpha= a/M, the radial equation reads (dρ/dλ)2=B(ρ)(d\rho/d\lambda)^2 = B(\rho), with

B(ρ)=1−ℓ2−α2ρ2+2(ℓ−α)2ρ3.(47)B(\rho) = 1 - \frac{\ell^2-\alpha^2}{\rho^2} + \frac{2(\ell-\alpha)^2}{\rho^3}. \tag*{(47)}

Here ℓ\ell is the impact parameter in units of MM, and the Schwarzschild case is α=0\alpha= 0. The turning point ρm\rho_m, defined by B(ρm)=0B(\rho_m) = 0, satisfies ℓ=ρm+1+O(ρm−1)\ell= \rho_m + 1 + O(\rho_m^{-1}). With x=ρ/ρmx = \rho/\rho_m,

B=B0+ρm−1B1+O(ρm−2),B0=1−1x2,B1=2(1x3−1x2),(48)B = B_0 + \rho_m^{-1}B_1 + O(\rho_m^{-2}), \qquad B_0 = 1 - \frac{1}{x^2}, \qquad B_1 = 2\left(\frac{1}{x^3} - \frac{1}{x^2}\right), \tag*{(48)}

so α\alpha does not enter at this order. The affine distance from the turning point is

Λ(ρ)=∫ρmρdρ′B=ρmL0(x)+L1(x)+O(ρm−1),L0=x2−1,L1=x−1x+1.(49)\Lambda(\rho) = \int_{\rho_m}^{\rho} \frac{d\rho'}{\sqrt{B}} = \rho_m L_0(x) + L_1(x) + O(\rho_m^{-1}), \qquad L_0 = \sqrt{x^2 - 1}, \qquad L_1 = \sqrt{\frac{x-1}{x+1}}. \tag*{(49)}

In the parallel frame fixed on the incident screen, the quantity κ=(ρ2+α2)e2ϕ−αe2t\kappa= (\rho^2+\alpha^2)e_2^\phi-\alpha e_2^t obeys dκ/dλ=−1d\kappa/d\lambda=-1. Denote by Λm\Lambda_m its value at the turning point. Then κ=Λm±Λ(ρ)\kappa=\Lambda_m\pm\Lambda(\rho) on the incoming and outgoing branches, and

Cmˉmmˉm=ρ2−3[Λm±Λ(ρ)]2ρ5.(50)C_{\bar m m \bar m m}=\frac{\rho^2-3[\Lambda_m\pm\Lambda(\rho)]^2}{\rho^5}. \tag*{(50)}

Adding both branches at the same radius cancels the terms linear in ΛΛm\Lambda\Lambda_m:

IC=2∫ρm∞ρ2−3(Λ2+Λm2)ρ5B dρ.(51)I_C=2\int_{\rho_m}^{\infty}\frac{\rho^2-3(\Lambda^2+\Lambda_m^2)}{\rho^5\sqrt{B}}\,d\rho. \tag*{(51)}

Since Λm=O(1)\Lambda_m=O(1), its contribution enters at O(ρm−4)O(\rho_m^{-4}). More precisely, a parallel null rotation e2→e2+[δ/(ℓ−α)]ke_2\to e_2+[\delta/(\ell-\alpha)]k changes the integral by exactly −3(2Λmδ+δ2)J5-3(2\Lambda_m\delta+\delta^2)J_5, with J5=2∫ρm∞dρ/(ρ5B)=4/(3ρm4)+⋯J_5=2\int_{\rho_m}^{\infty}d\rho/(\rho^5\sqrt{B})=4/(3\rho_m^4)+\cdots. Expanding, IC=I0ρm−2+I1ρm−3+O(ρm−4)I_C=I_0\rho_m^{-2}+I_1\rho_m^{-3}+O(\rho_m^{-4}). The leading coefficient,

I0=2∫1∞3−2x2x4x2−1 dx=0,(52)I_0=2\int_1^\infty\frac{3-2x^2}{x^4\sqrt{x^2-1}}\,dx=0, \tag*{(52)}

vanishes because, along the straight line ρ2=ρm2+λ2\rho^2=\rho_m^2+\lambda^2,

Cmˉmmˉm≃ρm2−2λ2(ρm2+λ2)5/2=ddλ[λ(ρm2+λ2)3/2].(53)C_{\bar m m \bar m m}\simeq\frac{\rho_m^2-2\lambda^2}{(\rho_m^2+\lambda^2)^{5/2}}=\frac{d}{d\lambda}\left[\frac{\lambda}{(\rho_m^2+\lambda^2)^{3/2}}\right]. \tag*{(53)}

This is the explicit form of the boundary term −2[Re⁡μ]-2[\operatorname{Re}\mu] in Eq. (IV.2). At the next order,

I1=−2∫1∞6(x−1)+2x2−3x(x+1)x4x2−1 dx=−3π8,(54)I_1=-2\int_1^\infty\frac{6(x-1)+\frac{2x^2-3}{x(x+1)}}{x^4\sqrt{x^2-1}}\,dx=-\frac{3\pi}{8}, \tag*{(54)}

which gives Eq. (V.1).

A numerical evaluation of the same integral for the equatorial Kerr problem with a/M=0.5a/M=0.5, with ℓ\ell determined from the turning radius and both branches integrated to infinity, gives ρm3IC=−2.071951, −1.554767, −1.352459, −1.262133, −1.219368, −1.194431, −1.186235\rho_m^3 I_C=-2.071951,\,-1.554767,\,-1.352459,\,-1.262133,\,-1.219368,\,-1.194431,\,-1.186235 and −1.182159-1.182159 for ρm=10, 20, 40, 80, 160, 400, 800\rho_m=10,\,20,\,40,\,80,\,160,\,400,\,800 and 16001600. These values converge to −3π/8-3\pi/8, consistently with the absence of α\alpha at this order.

Angular phase of dssld^l_{ss}

The rotation functions with equal indices can be written in terms of Jacobi polynomials,

dssl(θ)=cos⁡2sθ2Pl−s(0,2s)(cos⁡θ),(55)d^l_{ss}(\theta)=\cos^{2s}\frac{\theta}{2}P_{l-s}^{(0,2s)}(\cos\theta), \tag*{(55)}

and the Hilb-type asymptotics of Jacobi polynomials [34] give dssl(θ)≃θ/sin⁡θJ0(Lθ)d^{l}_{ss}(\theta) \simeq\sqrt{\theta/\sin\theta} J_0(L\theta) at leading order, with L=l+12L=l+\frac{1}{2} for every ss. The spin dependence appears at the next order. The function sin⁡θ dssl\sqrt{\sin\theta}\,d^{l}_{ss} obeys

d2udθ2+[L2+14sin⁡2θ−s2cos⁡2(θ/2)]u=0,(56)\frac{\mathrm{d}^2u}{\mathrm{d}\theta^2}+\left[L^2+\frac{1}{4\sin^2\theta}-\frac{s^2}{\cos^2(\theta/2)}\right]u=0, \tag*{(56)}

which reduces to the Legendre case for s=0s=0. In the oscillatory region the WKB phase difference is

Φsang−Φ0ang≃−s22L∫0θdθ′cos⁡2(θ′/2)=−s2Ltan⁡θ2.(57)\Phi^{\mathrm{ang}}_s-\Phi^{\mathrm{ang}}_0\simeq-\frac{s^2}{2L}\int_0^\theta\frac{\mathrm{d}\theta'}{\cos^2(\theta'/2)}=-\frac{s^2}{L}\tan\frac{\theta}{2}. \tag*{(57)}

The integration constant is fixed by the small-angle expansion dssl=1−[l(l+1)−s2]θ2/4+…d^{l}_{ss}=1-[l(l+1)-s^2]\theta^2/4+\ldots, which corresponds to J0(L2−s2 θ)J_0(\sqrt{L^2-s^2}\,\theta). Corrections are of relative order L−2L^{-2} and (Lθ)−2(L\theta)^{-2}. In the scattering problem these correspond to order ϵ3\epsilon^3.

References

References

  1. [1]R. A. Isaacson, Gravitational radiation in the limit of high frequency. i. the linear approximation and geometrical optics, Physical Review 166, 1263 (1968).DOI
  2. [2]R. A. Isaacson, Gravitational radiation in the limit of high frequency. ii. nonlinear terms and the effective stress tensor, Physical Review 166, 1272 (1968).DOI
  3. [3]S. R. Dolan, Geometrical optics for scalar, electromagnetic and gravitational waves on curved spacetime, International Journal of Modern Physics D 27, 1843010 (2018).DOI
  4. [4]S. R. Dolan, Higher-order geometrical optics for electromagnetic waves on a curved spacetime (2018), arXiv:1801.02273 [gr-qc].arxiv.org/abs/1801.02273
  5. [5]A. I. Harte, Gravitational lensing beyond geometric optics: I. formalism and observables, General Relativity and Gravitation 51, 14 (2019).DOI
  6. [6]M. V. Berry, Quantal phase factors accompanying adiabatic changes, Proceedings of the Royal Society of London A 392, 45 (1984).
  7. [7]V. P. Frolov and A. A. Shoom, Spinoptics in a stationary spacetime, Physical Review D 84, 044026 (2011).DOI
  8. [8]N. Yamamoto, Spin hall effect of gravitational waves, Physical Review D 98, 061701 (2018).DOI
  9. [9]V. P. Frolov, Maxwell equations in a curved spacetime: Spin optics approximation, Physical Review D 102, 084013 (2020).DOI
  10. [10]M. A. Oancea, J. Joudioux, I. Y. Dodin, D. E. Ruiz, C. F. Paganini, and L. Andersson, Gravitational spin hall effect of light, Physical Review D 102, 024075 (2020).DOI
  11. [11]L. Andersson, J. Joudioux, M. A. Oancea, and A. Raj, Propagation of polarized gravitational waves, Physical Review D 103, 044053 (2021).DOI
  12. [12]P. K. Dahal, Spin optics for gravitational waves, Astronomy 1, 271 (2022).DOI
  13. [13]A. I. Harte and M. A. Oancea, Spin hall effects and the localization of massless spinning particles, Physical Review D 105, 104061 (2022).DOI
  14. [14]L. Andersson and M. A. Oancea, Spin hall effects in the sky, Classical and Quantum Gravity 40, 154002 (2023).DOI
  15. [15]V. P. Frolov and A. A. Shoom, Gravitational spinoptics in a curved space-time, Journal of Cosmology and Astroparticle Physics 2024 (10), 039.arxiv.org/abs/2406.17905
  16. [16]V. P. Frolov, Spinoptics in the schwarzschild spacetime, Journal of Cosmology and Astroparticle Physics 2024 (12), 051.arxiv.org/abs/2410.19413
  17. [17]V. P. Frolov and A. Koek, Spinoptics in the kerr spacetime: Polarized wave scattering, Physical Review D 111, 104081 (2025).DOI
  18. [18]R. Ito, K. Mameda, and N. Yamamoto, Spin hall effect and berry curvature of gravitons from quantum field theory (2026), preprint, arXiv:2605.19817 [gr-qc].arxiv.org/abs/2605.19817
  19. [19]P. K. Dahal, Rayleigh propagation in Kerr spacetime (2023), arXiv:2301.08250 [gr-qc].DOI
  20. [20]G. Cusin and M. Lagos, Gravitational wave propagation beyond geometric optics, Physical Review D 101, 044041 (2020).DOI
  21. [21]E. Bruyère and C. Pitrou, Gravitational lensing beyond the eikonal approximation (2026), arXiv:2601.10239 [gr-qc].arxiv.org/abs/2601.10239
  22. [22]W. K. D. Logi and S. J. Kovács, Gravitational scattering of zero-rest-mass plane waves, Physical Review D 16, 237 (1977).
  23. [23]E. T. Newman and R. Penrose, An approach to gravitational radiation by a method of spin coefficients, Journal of Mathematical Physics 3, 566 (1962).
  24. [24]Y. F. Bautista, A. Guevara, C. Kavanagh, and J. Vines, Scattering in black hole backgrounds and higher-spin amplitudes. Part I, Journal of High Energy Physics 2023, 136 (2023).arxiv.org/abs/2107.10179
  25. [25]J. A. H. Futterman, F. A. Handler, and R. A. Matzner, Scattering from Black Holes (Cambridge University Press, Cambridge, 1988).
  26. [26]S. R. Dolan, Scattering and absorption of gravitational plane waves by rotating black holes, Classical and Quantum Gravity 25, 235002 (2008).DOI
  27. [27]L. C. B. Crispino, S. R. Dolan, and E. S. Oliveira, Electromagnetic wave scattering by Schwarzschild black holes, Physical Review Letters 102, 231103 (2009).DOI
  28. [28]S. Chandrasekhar, The Mathematical Theory of Black Holes (Clarendon Press, Oxford, 1983).
  29. [29]K. Ichiro Kubota, S. Arai, H. Motohashi, and S. Mukohyama, Spin wave optics for gravitational waves lensed by a Kerr black hole, Physical Review D 110, 124011 (2024).DOI
  30. [30]N. E. J. Bjerrum-Bohr, J. F. Donoghue, B. R. Holstein, L. Planté, and P. Vanhove, Bending of light in quantum gravity, Physical Review Letters 114, 061301 (2015).
  31. [31]N. E. J. Bjerrum-Bohr, J. F. Donoghue, B. R. Holstein, L. Planté, and P. Vanhove, Light-like scattering in quantum gravity, Journal of High Energy Physics 2016, 117 (2016).
  32. [32]M. A. Huber, A. Brandhuber, S. D. Angelis, and G. Travaglini, Eikonal phase matrix, deflection angle, and time delay in effective field theories of gravity, Physical Review D 102, 046014 (2020).DOI
  33. [33]M. Mehraeen, Quantum geometry of gravitons (2026), arXiv:2609.15237.arxiv.org/abs/2609.15237
  34. [34]G. Szegő, Orthogonal Polynomials, American Mathematical Society Colloquium Publications, Vol. 23 (American Mathematical Society, Providence, RI, 1939).

Paper details

vlnc:W_3ke66dnv
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

downloads · 0

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

  • Physics and Astronomy

Claude Opus 5.5 (bare)

View statement
Looking for related papers…
Contents