betaFeedback

WorksPhysics and Astronomy

Radiative-recoil corrections at order α2(Zα)5α^2 (Z α)^5 to energy levels of muonium and positronium

Abstract

We have calculated radiative-recoil corrections at order α2(Zα)5α^2 (Z α)^5 to the energy levels of spin-1/2–spin-1/2 Coulombic bound systems. The corrections we consider involve two exchange photons with two-loop radiative corrections on the electron line. We obtain precise results for all values of the mass ratio. Our results are relevant for upcoming improved measurements of the muonium ground-state hyperfine splitting and 1S-2S interval.

Introduction

Muonium and positronium are purely leptonic atoms, formed from an electron bound to either a positive muon or a positron. The structure of these atoms is described almost entirely by QED, with only small contributions from strong and weak interaction effects. Consequently, their energy levels can be calculated to high precision. Corresponding high precision measurements allow for tests of current theory and improved determination of fundamental constants. Representative reviews are given in Refs. [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. Novel approaches and improved and prospective measurements of the muonium [10, 11, 12, 13, 14, 15, 16, 17, 18, 19] and positronium [20, 21, 22, 23, 24, 25, 26, 27] energy levels provide the incentive to complete the order α7\alpha^7 energy level corrections.

Of immediate importance is the completion of the order α7\alpha^7 corrections to the muonium ground-state hyperfine splitting. New measurements of this splitting by the MuSEUM collaboration are currently underway at the J-PARC Muon Science Facility (MUSE), with a precision goal of ∼10\sim10 Hz [14, 17, 18]. Current theory for this transition is [4, 28, 29, 30]

ΔEhfsth(Mu)=4 463 302 873(511)(70)(2) Hz,(1)\Delta E_{\mathrm{hfs}}^{\mathrm{th}}(\mathrm{Mu}) = 4\,463\,302\,873(511)(70)(2)\ \mathrm{Hz}, \tag*{(1)}

where the quoted uncertainties are due to the uncertainty in the muon mass, uncalculated theoretical corrections, and the uncertainty in the fine structure constant α\alpha. The theoretical uncertainties are dominated by pure recoil corrections of order mr(Zα)7m_r(Z\alpha)^7 and radiative-recoil corrections of orders mrα(Zα)6m_r\alpha(Z\alpha)^6 and mrα2(Zα)5m_r\alpha^2(Z\alpha)^5, where mrm_r is the muonium reduced mass and ZZ is an assumed charge of the μ+\mu^+ in units of the proton charge (used to keep track of the number of muon-photon interactions: for muonium, Z=1Z=1). Many of the order mrα2(Zα)5m_r\alpha^2(Z\alpha)^5 recoil corrections are already known [31, 33, 34]. We have completed the remaining recoil correction of this order, involving two-photon exchange with two-loop radiative corrections on the electron line, with negligible uncertainty. We have calculated the radiative-recoil corrections to muonium and positronium S-state energy levels coming from two-photon exchange with two-loop radiative corrections on the electron line as a function of the mass ratio x=m1/m2x=m_1/m_2 between constituent particles. Prior calculations [35, 36, 37, 38, 39, 40, 41, 42, 43, 44] took into account the reduced mass dependence of the wave function at zero distance ∣ψ(0)∣2=(mrZα)3πn3|\psi(0)|^2=\frac{(m_rZ\alpha)^3}{\pi n^3}, but otherwise, only the no-recoil limit x=0x=0 was considered. We have found high-precision results for systems with xx ranging from small (hydrogen, muonium, muonic hydrogen) to the maximum value x=1x=1 as appropriate for positronium and true muonium.

General Outline.—There are six sets of graphs that contribute to energy levels at order α2(Zα)5\alpha^2(Z\alpha)^5, as displayed in Ref. [31]. Five of these sets involved vacuum polarization or light-by-light fermion loops, while the sixth contains two-loop radiative corrections to the electron line with no vacuum polarization contributions. The five sets of graphs containing fermion loops have been studied in detail, including recoil corrections [32, 33, 34]. The last set of graphs is the one of our concern here. The graphs of interest are shown in Figure 1.

Nineteen Feynman diagrams contributing to the bound-state energy levels

Figure 1. The nineteen Feynman diagrams having no fermion loops that contribute to bound state energy levels at order α2(Zα)5\alpha^2(Z\alpha)^5. Also contributing, but not shown, are the nineteen graphs where the two exchange photons cross one another. Some of the graphs [(A4), (B1), (B2), (B3), (C1) and group (E)] must be doubled to account for reflection symmetry.

Our calculations were done in the framework of NRQED [45]. We have effectively calculated the dd matching coefficients of (1) of [45] or eq:22 of [42] that represent four-fermion contact terms. The corresponding energy corrections are given by ΔE=−∣ψ(0)∣2M\Delta E=-|\psi(0)|^2\mathcal{M} where M\mathcal{M} is the order α2(Zα)3\alpha^2(Z\alpha)^3 three-loop amplitude for scattering of the constituent particles at zero relative velocity (i.e. at threshold).

For illustration of the calculation, we consider the diagram shown in Figure 1(A1). We calculate in d=4−2ϵd=4-2\epsilon dimensions in order to regularize ultraviolet (UV) and infrared (IR) divergences. We use a timelike metric (g00=+1, gij=−δij)(g_{00}=+1,\,g_{ij}=-\delta_{ij}), natural units ℏ=c=1\hbar=c=1, and values of fundamental constants from [30]. The amplitude MA1\mathcal{M}_{A1}, coming from the diagram of Figure 1(A1) is given by

MA1=−i∫ddq ddr dds uˉ1f(−iq1γν)iγ(m1n−q)−m1(−iq1γκ)iγ(m1n−q−r)−m1(−iq1γμ1)×iγ(m1n−q−r+s)−m1(−iq1γμ2)iγ(m1n−q−r)−m1(−iq1γκ)iγ(m1n−q)−m1(−iq1γν)u1i×(−iq2)(−ir2)(−is2)2uˉ2f(−iq2γμ1)iγ(m2n−s)−m2(−iq2γμ2)u2i,(2)\begin{aligned} \mathcal{M}_{A1}=-i\int d^d q\,d^d r\,d^d s\,\bar{u}_{1f}(-iq_1\gamma^\nu)\frac{i}{\gamma(m_1 n-q)-m_1}(-iq_1\gamma^\kappa)\frac{i}{\gamma(m_1 n-q-r)-m_1}(-iq_1\gamma^{\mu_1}) \\ &\times\frac{i}{\gamma(m_1 n-q-r+s)-m_1}(-iq_1\gamma^{\mu_2})\frac{i}{\gamma(m_1 n-q-r)-m_1}(-iq_1\gamma_\kappa)\frac{i}{\gamma(m_1 n-q)-m_1}(-iq_1\gamma_\nu)u_{1i} \\ &\times\left(\frac{-i}{q^2}\right)\left(\frac{-i}{r^2}\right)\left(\frac{-i}{s^2}\right)^2\bar{u}_{2f}(-iq_2\gamma_{\mu_1})\frac{i}{\gamma(m_2 n-s)-m_2}(-iq_2\gamma_{\mu_2})u_{2i}, \tag*{(2)} \end{aligned}

where qkq_k and mkm_k are the charge and mass of the kthk^{\mathrm{th}} constituent, n=(1,0⃗)n=(1,\vec{0}) is a timelike unit vector, mknm_k n is the incoming and outgoing energy-momentum vector of the kthk^{\mathrm{th}} particle at threshold in the center-of-mass (CM) frame, and ddq≡ddq/(2π)dd^d q \equiv d^d q/(2\pi)^d. We have used qq and rr for the momenta of the radiative correction photons and ss for the momentum of the exchange photons, and ukfu_{kf} and ukiu_{ki} are the final and initial Dirac spinors of the kthk^{\mathrm{th}} constituent. This amplitude can be written as

MA1=i(q12)(q1q2)2∫ddq ddr dds uˉ1fγν(γ(m1n−q)+m1)γκ(γ(m1n−q−r)+m1)γμ1u1i(q2−2qnm1)2((q+r)2−2(q+r)nm1)2×(γ(m1n−q−r+s)+m1)γμ2(γ(m1n−q−r)+m1)γκ(γ(m1n−q)+m1)γνu1i((q+r−s)2−2(q+r−s)nm1)×uˉ2fγμ1(γ(m2n−s)+m2)γμ2u2i(s2−2snm2)(q2)(r2)(s2).(3)\begin{aligned} \mathcal{M}_{A1}=i(q_1^2)(q_1q_2)^2\int d^d q\,d^d r\,d^d s\, \frac{\bar{u}_{1f}\gamma^\nu\left(\gamma(m_1n-q)+m_1\right)\gamma^\kappa\left(\gamma(m_1n-q-r)+m_1\right)\gamma^{\mu_1}u_{1i}} {\left(q^2-2qn m_1\right)^2\left((q+r)^2-2(q+r)nm_1\right)^2} \\ &\times\frac{\left(\gamma(m_1n-q-r+s)+m_1\right)\gamma^{\mu_2}\left(\gamma(m_1n-q-r)+m_1\right)\gamma_\kappa\left(\gamma(m_1n-q)+m_1\right)\gamma_\nu u_{1i}} {\left((q+r-s)^2-2(q+r-s)nm_1\right)} \\ &\times\frac{\bar{u}_{2f}\gamma_{\mu_1}\left(\gamma(m_2n-s)+m_2\right)\gamma_{\mu_2}u_{2i}} {\left(s^2-2snm_2\right)(q^2)(r^2)(s^2)}. \tag*{(3)} \end{aligned}

We combine the traces using the identities [46]

u1TCu2=18{γ5−γϵm}(γn+1),(4)u_1^T C u_2=\frac{1}{\sqrt{8}}\left\{\begin{matrix}\gamma_5\\-\gamma_{\epsilon_m}\end{matrix}\right\}(\gamma n+1), \tag*{(4)}
C−1uˉ2Tu1=18{γ5γϵm∗}(γn+1),(5)C^{-1}\bar{u}_2^T u_1=\frac{1}{\sqrt{8}}\left\{\begin{matrix}\gamma_5\\\gamma_{\epsilon_m}^{*}\end{matrix}\right\}(\gamma n+1), \tag*{(5)}

which are valid for {spin-0spin-1}\left\{\begin{matrix}\text{spin-0}\\\text{spin-1}\end{matrix}\right\} configurations of the constituent particles. Here CC is the charge conjugation matrix, which satisfies C−1γμTC=−γμC^{-1}\gamma_\mu^T C=-\gamma_\mu, γ5\gamma_5 is a matrix that anticommutes with the other gamma matrices γ5γμγ5=−γμ\gamma_5\gamma_\mu\gamma_5=-\gamma_\mu and satisfies γ52=1\gamma_5^2=1, and ϵm=(0,ϵ^m)\epsilon_m=(0,\hat{\epsilon}_m) is a polarization vector. We compute traces in dd dimensions using the standard Dirac anticommutation relations for gamma matrices, γμγμ=d\gamma_\mu\gamma^\mu=d, and tr⁡[1]=4\operatorname{tr}[1]=4. (There are no subtleties associated with γ5\gamma_5 since it always occurs in pairs.) One can rearrange terms to write

uˉ1fXu1i×uˉ2fYu2i=uˉ1fXu1i×uˉ2iCYC−1uˉ2fT=18tr⁡[X(γn+1){γ5Y~γ5−γϵmY~γϵm∗}(γn+1)],(6)\begin{aligned} \bar{u}_{1f}Xu_{1i}\times\bar{u}_{2f}Yu_{2i} =\bar{u}_{1f}Xu_{1i}\times\bar{u}_{2i}CY C^{-1}\bar{u}_{2f}^{T} \\ &=\frac{1}{8}\operatorname{tr}\left[X(\gamma n+1) \left\{\begin{matrix}\gamma_5\widetilde{Y}\gamma_5\\-\gamma_{\epsilon_m}\widetilde{Y}\gamma_{\epsilon_m}^{*}\end{matrix}\right\} (\gamma n+1)\right], \tag*{(6)} \end{aligned}

where Y~=C−1YTC\widetilde{Y}=C^{-1}Y^T C. For the spin-1 term, we average over the spin directions and use

γϵm⊗γϵm∗→γα⊗γβ1d−1∑m(ϵm)α(ϵm∗)β=γα⊗γβ1d−1(nαnβ−gαβ).(7)\begin{aligned} \gamma_{\epsilon_m}\otimes\gamma_{\epsilon_m}^{*}\to\gamma^\alpha\otimes\gamma^\beta\frac{1}{d-1}\sum_m(\epsilon_m)_\alpha(\epsilon_m^*)_\beta\\ &=\gamma^\alpha\otimes\gamma^\beta\frac{1}{d-1}(n_\alpha n_\beta-g_{\alpha\beta}). \tag*{(7)} \end{aligned}

The amplitude MA1\mathcal{M}_{A1} becomes

MA1=i(q12)(q1q2)2∫ddq ddr dds NA1DA1=α2(Zα)2(4π)2μˉ2ϵμ6ϵΦ3∫ddq ddr dds NA1DA1.(8)\begin{aligned} \mathcal{M}_{A1}=i(q_1^2)(q_1q_2)^2\int d^d q\,d^d r\,d^d s\,\frac{N_{A1}}{D_{A1}} \\ &=\frac{\alpha^2(Z\alpha)^2}{(4\pi)^2}\bar{\mu}^{2\epsilon}\mu^{6\epsilon}\Phi^3\int d^d q\,d^d r\,d^d s\,\frac{N_{A1}}{D_{A1}}. \tag*{(8)} \end{aligned}

We have used q1q2=−4πZαμˉ2ϵq_1q_2=-4\pi Z\alpha\bar{\mu}^{2\epsilon} for the product of the charges, and have included a mass parameter μˉ2ϵ\bar{\mu}^{2\epsilon} that is necessary to keep the dimensions consistent in dd dimensions. It is convenient to introduce an additional mass parameter μ\mu and a factor Φ≡(−i)(4π)−2ϵeϵγE\Phi\equiv(-i)(4\pi)^{-2\epsilon}e^{\epsilon\gamma_E}, defined so that μˉ2ϵ=(μ2eγE4π)ϵ=μ2ϵiΦ(4π)2\bar{\mu}^{2\epsilon}=\left(\frac{\mu^{2}e^{\gamma_E}}{4\pi}\right)^{\epsilon}=\mu^{2\epsilon}\frac{i\Phi}{(4\pi)^2}. We include one factor of Φ\Phi with each loop integral so that the results of the integrals, when expanded in the dimensional regularization parameter ϵ\epsilon, have as simple a form as possible without extraneous factors of γE\gamma_E or ln⁡(4π)\ln(4\pi). The numerator in MA1\mathcal{M}_{\mathrm{A1}} is

NA1=18tr⁡[γν(γ(m1n−q)+m1)γκ(γ(m1n−q−r)+m1)γμ1(γ(m1n−q−r+s)+m1)γμ2×(γ(m1n−q−r)+m1)γκ(γ(m1n−q)+m1)γν(γn+1){γμ2(γ(m2n−s)+m2)γμ1γϵmγμ2(γ(m2n−s)−m2)γμ1γϵ∗}(γn+1)],(9)N_{\mathrm{A1}}=\frac{1}{8}\operatorname{tr}\left[\gamma^\nu\left(\gamma(m_1 n-q)+m_1\right)\gamma^\kappa\left(\gamma(m_1 n-q-r)+m_1\right)\gamma^{\mu_1}\left(\gamma(m_1 n-q-r+s)+m_1\right)\gamma^{\mu_2}\right. \\ \left.\times\left(\gamma(m_1 n-q-r)+m_1\right)\gamma_\kappa\left(\gamma(m_1 n-q)+m_1\right)\gamma_\nu(\gamma n+1)\left\{\gamma_{\mu_2}\left(\gamma(m_2 n-s)+m_2\right)\gamma_{\mu_1}\right.\right. \\ \left.\left.\gamma_{\epsilon m}\gamma_{\mu_2}\left(\gamma(m_2 n-s)-m_2\right)\gamma_{\mu_1}\gamma_{\epsilon}^{*}\right\}(\gamma n+1)\right], \tag*{(9)}

and the denominator is

DA1=(−q2)(−r2)(−s2)2(−q2+2m1qn)2(−s2+2m2sn)[−(q+r)2+2m1(q+r)n]2[−(q+r−s)2+2m1(q+r−s)n].(10)D_{\mathrm{A1}}=(-q^2)(-r^2)(-s^2)^2(-q^2+2m_1qn)^2(-s^2+2m_2sn)\left[-(q+r)^2+2m_1(q+r)n\right]^2\left[-(q+r-s)^2+2m_1(q+r-s)n\right]. \tag*{(10)}

For the crossed graph everything is the same except that μ1↔μ2\mu_1\leftrightarrow\mu_2 and s→−ss\to-s in the bracketed term for the muon line, which becomes

{γμ1(γ(m2n+s)+m2)γμ2γϵmγμ1(γ(m2n+s)−m2)γμ2γϵ∗}.(11)\left\{\gamma_{\mu_1}\left(\gamma(m_2 n+s)+m_2\right)\gamma_{\mu_2} \gamma_{\epsilon m}\gamma_{\mu_1}\left(\gamma(m_2 n+s)-m_2\right)\gamma_{\mu_2}\gamma_{\epsilon}^{*}\right\}. \tag*{(11)}

Our calculations were done using Mathematica [47]. After first evaluating the traces with FeynCalc [48, 49, 50], we used the integration by parts (IBP) identities [51, 52] and the program FIRE [53] to reduce the expressions to a sum of 139 master integrals. Many of the three-loop master integrals, having up to eight denominators, were difficult to evaluate analytically, so we elected to pursue a numerical evaluation for specified values of x=m1/m2x=m_1/m_2 using the AMFlow program [54]. This program is able to evaluate dimensionally regularized Feynman integrals order by order in ϵ\epsilon, up to a specified maximum order, to a specified numerical precision. The program generates sets of differential equations satisfied by the required integral and companion integrals using an auxiliary mass parameter as the independent variable. The program solves the differential equations by a power series approach, using analytic continuation to move from a point where the boundary conditions are convenient to the physical point specified by the user. The boundary conditions are themselves Feynman integrals that are dealt with by the program recursively and automatically. The coefficients we found that multiplied the master integrals diverged, at most, as 1/ϵ41/\epsilon^4 for ϵ\epsilon approaching zero, so we had AMFlow evaluate the master integrals up to order ϵ4\epsilon^4, for eight total orders in ϵ\epsilon from ϵ−3\epsilon^{-3} to ϵ4\epsilon^4. We specified 42 or more significant figures of precision from AMFlow to overcome any severe cancellation between terms with the concomitant loss of precision.

After evaluating the master integrals, the complete contribution of the diagrams of Figure 1 was obtained for the spin-0 and spin-1 states individually. The total result contained 1/ϵ31/\epsilon^3 divergences that required renormalization. We used renormalized perturbation theory [55, 56, 57, 58], which contains vertex and self-mass counterterms, to arrive at the complete set of required graphs that, together, produce the finite renormalized scattering amplitudes. Appropriate Feynman rules can be found in Ref. [59] for the counterterm graphs of Figure 2.

Counterterm diagrams for the radiative-recoil correction

Figure 2. Counterterm diagrams for the radiative-recoil correction of order α2(Zα)5\alpha^2(Z\alpha)^5. Graphs with crossed exchange photons must also be included. Symmetry factors of two are included for diagrams (B2), (C2), and (D1)–(D4).

All calculations using FeynCalc and FIRE were independently confirmed by multiple authors. The AMFlow integrals were confirmed by comparison against analytic results for cases where the analytic results were available.

Results.— We present our results as functions of the mass ratio xx with appropriate factors. For the hyperfine splitting we have

ΔEhfs=α2(Zα)πn3EFRhfs(x),(12)\Delta E_{\mathrm{hfs}}=\frac{\alpha^2(Z\alpha)}{\pi n^3}E_F R_{\mathrm{hfs}}(x), \tag*{(12)}

where the Fermi energy is

EF=83m1(Zα)4(m1m2)(mrm1)3.(13)E_F = \frac{8}{3}m_1(Z\alpha)^4\left(\frac{m_1}{m_2}\right)\left(\frac{m_r}{m_1}\right)^3. \tag*{(13)}

For the average energy we have

ΔEavg=m1α2(Zα)5πn3(mrm1)3Ravg(x).(14)\Delta E_{\mathrm{avg}} = \frac{m_1\alpha^2(Z\alpha)^5}{\pi n^3}\left(\frac{m_r}{m_1}\right)^3 R_{\mathrm{avg}}(x). \tag*{(14)}

We assume m1≤m2m_1 \leq m_2. Particle one is typically the electron, while particle two might be the proton, positive muon, or positron. Plots of Rhfs(x)R_{\mathrm{hfs}}(x) and Ravg(s)R_{\mathrm{avg}}(s) are shown in Figure 3 and Figure 4.

Figure 3. Plot of Rhfs(x)R_{\mathrm{hfs}}(x) showing the recoil dependence of the hyperfine structure contribution of order α2(Zα)5\alpha^{2}(Z\alpha)^{5}. Uncertainties are much too small to be visible.

Figure 4. Plot of Ravg(x)R_{\mathrm{avg}}(x) showing the recoil dependence of the spin-average contribution of order α2(Zα)5\alpha^{2}(Z\alpha)^{5}. Uncertainties are much too small to be visible.

The recoil behavior of the hyperfine structure energy shift is shown in Figure 3. The data for this plot is available in the Ancillary Material [60] that accompanies this article. The data points are quite precise, with ones near the center of the plot having up to 32 significant figures. We can compare the no-recoil limit of our result, −0.673259-0.673259 (see Table I), against prior work. Kinoshita and Nio obtained an early estimate of Rhfs(0)R_{\mathrm{hfs}}(0) of -0.63(4) in 1994 [37]. Eides and Shelyuto obtained -0.6726(4) in 1995 [41], and Kinoshita and Nio improved their earlier estimate in 1996, finding the value -0.6764(79) [42]. A significantly more precise value, -0.67326(2), can be extracted from the work of Mondéjar, Piclum, and Czarnecki, published in 2010 [43]. Our no-recoil limit is consistent with these prior results. All of these calculations, including ours, are based on the evaluation of a two-body scattering amplitude at threshold. The Kinoshita-Nio calculation was done using Feynman gauge and a non-zero photon mass λ\lambda for IR regulation, with the uncertainty dominated by the challenge of taking the λ→0\lambda\to0 limit. Eides and Shelyuto used the Fried-Yennie gauge, which has better IR

Systemx=m1/m2x = m_1/m_2Rhfs(x)R_{\mathrm{hfs}}(x)Ravg(x)R_{\mathrm{avg}}(x)
no recoil0-0.673259-7.723827
hydrogenme/mp≈0.00054m_e/m_p \approx0.00054-0.675641-7.718773
muoniumme/mμ≈0.0048m_e/m_\mu\approx0.0048-0.687745-7.679183
muonic hydrogenmμ/mp≈0.11m_\mu/m_p \approx0.11-0.762530-6.801971
positronium1-0.479248-3.510425

Table 1. Order α2(Zα)5\alpha^{2}(Z\alpha)^{5} radiative-recoil corrections for various physical systems composed of spin-12\frac{1}{2} particles. Our results are given for both the hyperfine splitting [Rhfs(x)R_{\mathrm{hfs}}(x)] and average [Ravg(x)R_{\mathrm{avg}}(x)] correction functions. Uncertainties are less than one in the last digit.

behavior that allowed for higher precision. Mondéjar et al. used the integration by parts identities to express their result in terms of a few master integrals, which were evaluated numerically.

The function Ravg(x)R_{\mathrm{avg}}(x) for the average energy correction is presented in Figure 4. The no-recoil limit of this plot is -7.723827 as shown in Table I. The first calculation of this no-recoil correction was done by Pachucki [38]. His result was -7.61(16), done using a hybrid method where some diagrams (a gauge-invariant set) were calculated in Fried-Yennie gauge, while the others were done using Feynman gauge and a mixed analytic/numeric approach. Eides and Shelyuto [41] achieved a higher precision result -7.724(1) using the Fried-Yennie gauge. Dowling et al. [44] found the high-precision result -7.72381(4) by utilizing the integration by parts identities and a numerical evaluation of the resulting master integrals.

Conclusions.—We calculated the radiative-recoil corrections of order α2(Zα)5\alpha^{2}(Z\alpha)^{5} to the hyperfine splitting and average energies of SS states of two-body QED bound systems for all values of the mass ratio. Our muonium ground state hyperfine result is

ΔEhfs=α2(Zα)πEFRhfs(xM)=−378.886(1) Hz,(15)\Delta E_{\mathrm{hfs}} = \frac{\alpha^{2}(Z\alpha)}{\pi} E_{\mathrm{F}} R_{\mathrm{hfs}}(x_{\mathrm{M}}) = -378.886(1)\ \mathrm{Hz}, \tag*{(15)}

where xM=me/mμx_{\mathrm{M}} = m_{e}/m_{\mu}. This corrects the no-recoil estimate of Mondéjar et al. [43] by −7.98(1) Hz-7.98(1)\ \mathrm{Hz}, where the no-recoil value was found using Rhfs(0)≈−0.67326(2)R_{\mathrm{hfs}}(0) \approx-0.67326(2). The −7.98(1) Hz-7.98(1)\ \mathrm{Hz} recoil correction is relevant given the expected precision of ∼10 Hz\sim10\ \mathrm{Hz} of the MuSEUM experiment [14, 17] and will reduce the theoretical uncertainty of ±70 Hz\pm70\ \mathrm{Hz} cited by Eides [28, 29] or ±85 Hz\pm85\ \mathrm{Hz} cited by Mohr et al. [30]. For positronium the α2(Zα)5\alpha^{2}(Z\alpha)^{5} correction is

ΔEhfs=2×α2(Zα)πEFRhfs(1)=−13.85 kHz.(16)\Delta E_{\mathrm{hfs}} = 2 \times\frac{\alpha^{2}(Z\alpha)}{\pi} E_{\mathrm{F}} R_{\mathrm{hfs}}(1) = -13.85\ \mathrm{kHz}. \tag*{(16)}

(The contribution was doubled to account for two-loop radiative corrections on the positron line.)

We also have new results for the muonium and positronium 1S-2S transition energies

ΔEavg(1S−2S)=−7meα78π(mrme)3Ravg(x).(17)\Delta E_{\mathrm{avg}}(1S - 2S) = -\frac{7m_{e}\alpha^{7}}{8\pi}\left(\frac{m_{r}}{m_{e}}\right)^{3}R_{\mathrm{avg}}(x). \tag*{(17)}

For muonium, our radiative-recoil result is 287.02445(4) kHz, which corrects the no-recoil result of Dowling et al. [44] by −1.668(2)-1.668(2) kHz (to be compared with the ∼10\sim10 kHz goal of the Mu-MASS experiment [15]), while for positronium the contribution at this order is 33.28 kHz.

We have given high-precision results for the order α2(Zα)5\alpha^{2}(Z\alpha)^{5} radiative-recoil corrections for all values of the mass ratio. These results are important given the expected uncertainties of ongoing experiments, and they help to complete the calculation of all corrections to these systems of order α7\alpha^{7}. In addition, our calculation demonstrates the usefulness of the program AMFlow [54], which allows for the convenient evaluation of dimensionally regularized multiloop integrals to high precision.

Acknowledgements.— We acknowledge the assistance of Lecheng Ni in the early stages of this project. This work was completed in part using the F&M research cluster, supported by F&M’s Research Computing team and NSF CC* Cyber-infrastructure award 1925192. Crucial support was provided by the National Science Foundation grant No. PHY-2308792 and by the National Institute of Standards and Technology grant 60NANB23D230.

Data Availability.— The data that supports the findings of this article are available in the accompanying Ancillary Materials.

References

References

  1. [1]J. R. Sapirstein and D. R. Yennie, Theory of hydrogenic bound states, in Quantum Electrodynamics, ed. by T. Kinoshita (World Scientific, Singapore, 1990), pp. 560-672.DOI
  2. [2]S. G. Karshenboim, Precision study of positronium: testing bound state QED theory, Int. J. Mod. Phys. A 19, 3879-3896 (2004), arXiv:hep-ph/0310099.DOI
  3. [3]S. G. Karshenboim, Precision physics of simple atoms: QED tests, nuclear structure and fundamental constants, Phys. Rep. 422, 1-63 (2005), arXiv:hep-ph/0509010.DOI
  4. [4]M. I. Eides, H. Grotch, and V. A. Shelyuto, Theory of light hydrogenic bound states, (Springer, Berlin, 2007).
  5. [5]K. P. Jungmann, Precision muonium spectroscopy, J. Phys. Soc. Jpn. 85, 091004 (2016), arXiv:1603.01195.DOI
  6. [6]D. B. Cassidy, Experimental progress in positronium laser physics, Eur. Phys. J. D 72, 53 (2018).DOI
  7. [7]C. Frugiuele, J. Pérez-Ríos, and C. Peset, Current and future perspectives of positronium and muonium spectroscopy as dark sectors probe, Phys. Rev. D 100, 015010 (2019), arXiv:1902.08585.DOI
  8. [8]G. S. Adkins, D. B. Cassidy, and J. Pérez-Ríos, Precision spectroscopy of positronium: Testing bound-state QED theory and the search for physics beyond the Standard Model, Phys. Rep. 975, 1-61 (2022).DOI
  9. [9]S. D. Bass, S. Mariazzi, P. Moskal, and E. Stepién, Colloquium: Positronium physics and biomedical applications, Rev. Mod. Phys. 95, 021002 (2023), arXiv:2302.09246.DOI
  10. [10]P. Blumer, S. Geissmann, A. J. Vargas, G. Janka, B. Ohayon, and P. Crivelli, Muonium fine structure: theory update, tests of Lorentz violation, and experimental prospects, Eur. Phys. J. D 79, 24 (2025), arXiv:2412.19580.arxiv.org/abs/2412.19580
  11. [11]C. Delaunay, B. Ohayon, and Y. Soreq, Towards an independent determination of muon g − 2 from muonium spectroscopy, Phys. Rev. Lett. 127, 251801 (2021), arXiv:2106.11998.DOI
  12. [12]B. Ohayon, G. Janka, I. Cortinovis, Z. Burkley, L. de Sousa Borges, E. Depero, A. Golovizin, X. Ni, Z. Salman, A. Suter, C. Vigo, T. Prokscha, and P. Crivelli (Mu-MASS Collaboration), Precision measurement of the Lamb shift in muonium, Phys. Rev. Lett. 128, 011802 (2022), arXiv:2108.12891.DOI
  13. [13]G. Janka, B. Ohayon, I. Cortinovis, Z. Burkley, L. de Sousa Borges, E. Depero, A. Golovizin, X. Ni, Z. Salman, A. Suter, T. Prokscha, and P. Crivelli (Mu-MASS Collaboration), Measurement of the transition frequency from 2S1/2, F=0 to 2P1/2, F=1 states in muonium, Nature Commun. 13, 7273 (2022).DOI
  14. [14]R. Iwai, M. Abe, S. Fukumura, M. Hiraishi, S. Kanda, S. Kawamura, S. Nishimura, H. Okabe, K. Sasaki, P. Strasser, K. Shimomura, H. Tada, N. Teshima and H. A. Torii, Precise measurement of the hyperfine splitting in muonium with a high intensity pulsed muon beam at J-PARC, J. Phys.: Conf. Series 2462, 012019 (2023).DOI
  15. [15]I. Cortinovis, B. Ohayon, L. de Sousa Borges, G. Janka, A. Golovizin, N. Zhadnov, and P. Crivelli, Update of muonium 1S-2S transition frequency, Eur. Phys. J. D 77, 66 (2023), arXiv:2301.12883.DOI
  16. [16]E. Javary, E. Thorpe-Woods, I. Cortinovis, M. Mähring, L. de Sousa Borges, and P. Crivelli, Two-photon optical Ramsey-Doppler spectroscopy of positronium and muonium, Eur. Phys. J. D 79, 15 (2025), arXiv:2411.19872.arxiv.org/abs/2411.19872
  17. [17]P. Strasser et al. (MuSEUM Collaboration), Precision measurements of muonium and muonic helium hyperfine structure at J-PARC, Eur. Phys. J. D 79, 20 (2025), arXiv:2501.02736DOI
  18. [18]H. A. Torii, S. Nishimura, S. Kanda, T. Yamazaki, P. Strasser, and K. Shimomura, Hyperfine spectroscopy of muonium and muonic atoms for precise determination of the muon magnetic moment (mini review), Interactions 247, 336 (2026).DOI
  19. [19]P. Blumer, G. Janka, S. Geissmann, I. Ignatov, M. Mähring, Z. Salman, A. Suter, E. Thorpe-Woods, T. Prokscha, B. Ohayon, and P. Crivelli (Mu-MASS Collaboration), Microwave spectroscopy of the muonium 2S1/2−2P3/2 fine-structure interval, Phys. Rev. A 113, 012817 (2026), arXiv:2509.04674.arxiv.org/abs/2509.04674
  20. [20]R. E. Sheldon, T. J. Babij, S. H. Reeder, S. D. Hogan, and D. B. Cassidy, Microwave spectroscopy of positronium atoms in free space, Phys. Rev. A 107, 042810 (2023).DOI
  21. [21]R. E. Sheldon, T. J. Babij, S. H. Reeder, S. D. Hogan, and D. B. Cassidy, Precision microwave spectroscopy of the positronium 2 3S1 → 2 3P2 interval, Phys. Rev. Lett. 131, 043001 (2023).
  22. [22]L. T. Glöggler et al. (AEgIS Collaboration), Positronium laser cooling via the 1 3S − 2 3P transition with a broadband laser pulse, Phys. Rev. Lett. 132, 083402 (2024), arXiv:23010.08760.arxiv.org/abs/2310.08760
  23. [23]K. Shu et al., Cooling positronium to ultralow velocities with a chirped laser pulse train, Nature 633, 793-797 (2024).
  24. [24]M. W. Heiss, L. de Sousa Borges, A. Golovizin, and P. Crivelli, Measurement of the 1 3S1 → 2 3S1 interval in positronium using a field ionization of Rydberg states, Phys. Rev. A 111, 012810 (2025), arXiv:2407.02443.
  25. [25]L. de Sousa Borges, E. Thorpe-Woods, E. Javary, and P. Crivelli, Precision continuous-wave laser measurement of the 1 3S1 → 2 3S1 interval in positronium, Phys. Rev. A 113, 032804 (2026), arXiv:2512.16018.
  26. [26]D. M. Newson and D. B. Cassidy, Phase-variation Ramsey spectroscopy of the 2 3S1 → 2 3P2 interval in positronium, Phys. Rev. Lett. 136, 033001 (2026).
  27. [27]D. M. Newson and D. B. Cassidy, Precision measurement of the 2 3S1 → 2 3P2 interval in positronium using phase-variation Ramsey spectroscopy, Phys. Rev. A 114, 022821 (2026).
  28. [28]M. I. Eides, Hyperfine splitting in muonium: Accuracy of the theoretical prediction, Phys. Lett. B 795, 113-116 (2019), arXiv:1812.10881.
  29. [29]M. I. Eides, Muonium hyperfine splitting uncertainty revisited, Mod. Phys. Lett. A 41, 2650074 (2026).arxiv.org/abs/2510.07281
  30. [30]P. J. Mohr, D. B. Newell, B. N. Taylor, and E. Tiesinga, CODATA recommended values of the fundamental physical constants: 2022, Rev. Mod. Phys. 97, 025002 (2025), arXiv:2409.03787.
  31. [31]S. G. Karshenboim, V. A. Shelyuto, and M. I. Eides, The first corrections of order α2(Zα)EF to hyperfine splitting in muonium and hydrogen, Sov. J. Nucl. Phys. 50, 1015-1020 (1989) [Yad. Fiz. 50, 1636-1645 (1989)].
  32. [32]M. I. Eides and V. A. Shelyuto, Hard three-loop corrections to hyperfine splitting in positronium and muonium, Phys. Rev. D 92, 013010 (2015).DOI
  33. [33]M. I. Eides and V. A. Shelyuto, Hard three-loop corrections to hyperfine splitting, Int. J. Mod. Phys. A 31, 1641030 (2016).DOI
  34. [34]M. I. Eides and V. A. Shelyuto, Three-loop corrections to the Lamb shift in muonium and positronium, Phys. Rev. A 105, 012803 (2022), arXiv:2110.13279.
  35. [35]M. I. Eides, S. G. Karshenboim, and V. A. Shelyuto, First corrections of order α2(Zα)5 to hyperfine splitting and Lamb shift induced by two-loop insertions in the electron line, Phys. Lett. B 312, 358-366 (1993).
  36. [36]M. I. Eides, S. G. Karshenboim, and V. A. Shelyuto, First corrections to hyperfine splitting and Lamb shift induced by diagrams with two external photons and second order radiative insertions in the electron line, Yad. Fiz. 57, 1309-1325 (1994), arXiv:hep-ph/9306266.
  37. [37]T. Kinoshita and M. Nio, Improved theory of the muonium hyperfine structure, Phys. Rev. Lett. 72, 3803-3806 (1994), arXiv:hep-ph/9402260.DOI
  38. [38]K. Pachucki, Complete two-loop binding correction to the Lamb shift, Phys. Rev. Lett. 72, 3154-3157 (1994).
  39. [39]M. I. Eides, S. G. Karshenboim, and V. A. Shelyuto, Corrections to hyperfine splitting and Lamb shift induced by the overlapping two-loop electron self-energy insertion in the electron line, Physics of Atomic Nuclei 57, 2158-2170 (1994). [Yad. Fiz. 57, 2246-2259 (1994).]
  40. [40]M. I. Eides and V. A. Shelyuto, New corrections to hyperfine splitting and Lamb shift and the value of the Rydberg constant, JETP Lett. 61, 478-484 (1995) [Pis’ma Zh. Eksp. Teor. Fiz. 61, 465-470 (1995)].
  41. [41]M. I. Eides and V. A. Shelyuto, Corrections of order α2(Zα)5 to the hyperfine splitting and the Lamb shift, Phys. Rev. A 52, 954-961 (1995).
  42. [42]T. Kinoshita and M. Nio, Radiative corrections to the muonium hyperfine structure: The α2(Zα) correction, Phys. Rev. D 53, 4909-4929 (1996), arXiv:hep-ph/9512327.
  43. [43]J. Mondéjar, J. H. Piclum, and A. Czarnecki, Radiative-nonrecoil corrections of order α2(Zα)EF to the hyperfine splitting of muonium, Phys. Rev. A 81, 062511 (2010), arXiv:1005.1944.
  44. [44]M. Dowling, J. Mondéjar, J. H. Piclum, and A. Czarnecki, Radiative-nonrecoil corrections of order α2(Zα)5 to the Lamb shift, Phys. Rev. A 81, 022509 (2010), arXiv:0911.4078.
  45. [45]W. E. Caswell and G. P. Lepage, Effective Lagrangians for bound state problems in QED, QCD, and other field theories, Phys. Lett. B 167, 437-442 (1986).
  46. [46]A. Czarnecki, K. Melnikov, and A. Yelkhovsky, Positronium S-state spectrum: Analytic results at O(mα6), Phys. Rev. A 59, 4316-4330 (1999), arXiv:hep-ph/9901394.DOI
  47. [47]Wolfram Research, Inc., Mathematica, Version 14.3, Champaign, IL (2025), Mathematica.
  48. [48]R. Mertig, M. Böhm, and A. Denner, FeynCalc - computer-algebraic calculation of Feynman amplitudes, Comput. Phys. Commun. 64, 345-359 (1991).
  49. [49]V. Shtabovenko, R. Mertig, and F. Orellana, New developments in FeynCalc 9.0, Comput. Phys. Commun. 207, 432-444 (2016), arXiv:1601.01167.DOI
  50. [50]V. Shtabovenko, R. Mertig, and F. Orellana, FeynCalc 9.3: New features and improvements, Comput. Phys. Commun. 256, 107478 (2020), arXiv:2001.04407.DOI
  51. [51]F. V. Tkachov, A theorem on analytical calculability of 4-loop renormalization group functions, Phys. Lett. B 100, 65-68 (1981).
  52. [52]K. G. Chetyrkin and F. V. Tkachov, Integration by parts: The algorithm to calculate β-functions in 4 loops, Nucl. Phys. B 192, 159-204 (1981).
  53. [53]A. V. Smirnov, FIRE5: A C++ implementation of Feynman Integral REduction, Comput. Phys. Commun. 189, 182-191 (2015), arXiv:1408.2372.DOI
  54. [54]X. Liu and Y.-Q. Ma, AMFlow: A Mathematica package for Feynman integrals computation via auxiliary mass flow, Comput. Phys. Commun. 283, 108565 (2023), arXiv:2201.11669.DOI
  55. [55]C. Itzykson and J.-B. Zuber, Quantum Field Theory, (McGraw-Hill, New York, 1980).
  56. [56]M. Kaku, Quantum Field Theory: A Modern Introduction, (Oxford University Press, New York, 1993).
  57. [57]S. Weinberg, The Quantum Theory of Fields Vol. I: Foundations, (Cambridge University Press, Cambridge, 1995).
  58. [58]M. E. Peskin and D. V. Schroeder, An Introduction to Quantum Field Theory, (Addison-Wesley, Reading, Massachusetts, 1995).
  59. [59]G. S. Adkins, E. Mitchell, and A. V. Turan, Analytic results for the two-loop renormalization constants of Feynman gauge QED, to appear in Phys. Rev. D (2026).
  60. [60]See the Ancillary Material for data tables (with associated uncertainties) giving the values of the spin-0 and spin-1 amplitudes as well as R_hfs(x) and R_avg(x) as functions of the mass ratio x = m_1/m_2 for the values of x shown in the plots.

Paper details

vlnc:W_g56kzrha
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

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