betaFeedback

WorksPhysics and Astronomy

Second-order hyperfine correction to H, D, and 3^3He energy levels

Abstract

The complete second-order hyperfine-interaction correction is calculated for centroid energy levels of H, D, and 3^3He atoms. For 3^3He, the corrections of −2.075-2.075~kHz and −0.305-0.305~kHz beyond the leading hyperfine-mixing contribution are obtained for the 21S2^1S and 23S2^3S states, respectively. These results shift the nuclear charge radii difference derived from the 3^3He -4^4He isotope shift and largely resolve the previously reported disagreement between the muonic and electronic helium determinations [van der Werf et al., arXiv:2306.02333 (2023); Schuhmann et al., arXiv:2305.11679 (2023)].

Introduction

The nuclear magnetic moment causes a shift of atomic energy levels, depending on the total angular momentum F⃗=I⃗+J⃗\vec{F} = \vec{I} + \vec{J}, where I⃗\vec{I} is the nuclear spin and J⃗\vec{J} represents the electron angular momentum [1]. This effect is known as hyperfine splitting (HFS). In hydrogen-like atoms, it is relatively small compared to the fine structure, which arises from the electron’s spin-orbit coupling.

Due to its small magnitude, hyperfine interaction is typically treated using first-order perturbation theory. The second-order HFS correction has rarely been addressed in the literature, and when it has, it is often treated incorrectly, primarily due to the presence of crossed diagrams [2] and the fact that it diverges in the point-nucleus limit. A proper analysis of the second-order HFS correction has been conducted only for hydrogen and muonic hydrogen (μH\mu\mathrm{H}) atoms so far, in the context of the two-photon exchange contribution [3, 4]. This was required because the hydrogen HFS has been measured precisely [5] and accurate measurements of HFS in light muonic atoms are underway [6, 7, 8, 9].

The second-order HFS corrections modify not only the HFS but also the centroid energies. The corresponding correction has usually been neglected due to its small size, with the leading order in hydrogenic systems starting at (Zα)6m3/M2(Z\alpha)^6m^3/M^2, where MM is the nuclear mass, mm is the electron mass, ZZ is the nuclear charge number, and α\alpha is the fine-structure constant. This correction forms part of a complete quadratic recoil correction, which has recently been derived in Ref. [10].

In helium-like atoms, the hyperfine interaction causes mixing between triplet and singlet states, which results in significant corrections. For instance, the HFS of the helium 23S2^{3}S state is impacted by it by as much as 60 kHz [11, 12], exceeding even the relativistic correction [12]. Similarly, the 23S−21S2^{3}S-2^{1}S mixing contributes a substantial correction to the 3He^{3}\mathrm{He}–4He^{4}\mathrm{He} isotope shift, around 80 kHz [13].

In light of the discrepancy in the nuclear charge radii difference between 3He^{3}\mathrm{He} and 4He^{4}\mathrm{He}, as determined by electronic versus muonic measurements [14, 15, 16], examining all possible corrections to the 3He^{3}\mathrm{He}–4He^{4}\mathrm{He} isotope shift has become critically important. Recently, Qi and collaborators [17] found that the mixing with n≥2n \ge2 excited states yields an additional correction to the helium isotope shift of about 1.37 kHz, which partially explains this discrepancy. In this work, we calculate the complete second-order HFS correction, finding an even larger correction of 1.77 kHz. Combined with an improved value of the field shift contribution, this result brings the electronic and muonic determinations of the nuclear charge radii difference [14, 15] into a 1.3 σ\sigma agreement.

Hydrogenic atom

Let us begin by considering the HFS interaction in hydrogenic systems. The Dirac Hamiltonian in the external electromagnetic field is (in natural units ℏ=c=1\hbar= c = 1)

HEM=α⃗⋅(p⃗−eA⃗hfs)+βm+eA0=HD+Vhfs,(1)H_{\mathrm{EM}} = \vec{\alpha} \cdot(\vec{p} - e\vec{A}_{\mathrm{hfs}}) + \beta m + eA^{0} = H_{\mathrm{D}} + V_{\mathrm{hfs}}, \tag*{(1)}

where ee is the electron charge, A0A_0 is the Coulomb field from the nucleus

V≡eA0=−Zαr,(2)V \equiv eA^{0} = -\frac{Z\alpha}{r}, \tag*{(2)}

and A⃗hfs\vec{A}_{\mathrm{hfs}} is the vector potential from the nuclear magnetic moment μ⃗\vec{\mu}

A⃗hfs=14πμ⃗×r⃗r3.(3)\vec{A}_{\mathrm{hfs}} = \frac{1}{4\pi}\vec{\mu} \times\frac{\vec{r}}{r^{3}}. \tag*{(3)}

The corresponding hyperfine interaction for a point nucleus is

Vhfs=−eα⃗⋅A⃗hfs=e4πμ⃗⋅α⃗×r⃗r3.(4)V_{\mathrm{hfs}} = -e\vec{\alpha} \cdot\vec{A}_{\mathrm{hfs}} = \frac{e}{4\pi}\vec{\mu} \cdot\vec{\alpha} \times\frac{\vec{r}}{r^{3}}. \tag*{(4)}

The second-order hyperfine correction to energy is ultraviolet divergent for a point nucleus. Therefore, we cannot ignore the charge and magnetic moment nuclear form factors, but for simplicity we assume that they are equal to each other, thus

ρE(r)=ρM(r)=ρ(r),(5)\rho_{E}(r) = \rho_{M}(r) = \rho(r), \tag*{(5)}

and neglect the higher order nuclear electromagnetic moments. Thus, potentials take the general form

V=−Zα[1r]fs≡−ZαrFC(r),(6)V=-Z\alpha\left[\frac{1}{r}\right]_{\mathrm{fs}}\equiv-\frac{Z\alpha}{r}F_C(r), \tag*{(6)}
Vhfs=e4πμ⋅α×[rr3]fs≡e4πμ⋅α×rr3Fhfs(r),(7)V_{\mathrm{hfs}}=\frac{e}{4\pi}\boldsymbol{\mu}\cdot\boldsymbol{\alpha}\times\left[\frac{\boldsymbol{r}}{r^3}\right]_{\mathrm{fs}}\equiv\frac{e}{4\pi}\boldsymbol{\mu}\cdot\frac{\boldsymbol{\alpha}\times\boldsymbol{r}}{r^3}F_{\mathrm{hfs}}(r), \tag*{(7)}

and FC(r)=Fhfs(r)=1F_C(r)=F_{\mathrm{hfs}}(r)=1 for the point nucleus. The second-order hyperfine correction to energy in the hydrogenic system is

Ehfs2=⟨ϕ∣Vhfs1(ED−HD)′Vhfs∣ϕ⟩.(8)E_{\mathrm{hfs}2}=\langle\phi|V_{\mathrm{hfs}}\frac{1}{(E_D-H_D)'}V_{\mathrm{hfs}}|\phi\rangle. \tag*{(8)}

The product of the nuclear magnetic moments in the above is to be replaced by μaμb→δabμ2/3\mu^a\mu^b\to\delta^{ab}\boldsymbol{\mu}^2/3, because here we are interested in corrections to energy levels but not to the hyperfine structure. Using the spectral representation of the Green function, we express Eq. (8) as

Ehfs2=13(Zα)2m3(g2M)2I(I+1)∑n,κn,μn1εa−εnE_{\mathrm{hfs}2}=\frac{1}{3}(Z\alpha)^2m^3\left(\frac{g}{2M}\right)^2I(I+1)\sum_{n,\kappa_n,\mu_n}\frac{1}{\varepsilon_a-\varepsilon_n}
×⟨naκaμa∣Fhfs(r)(α×r)ir3∣nκnμn⟩{}\times\langle n_a\kappa_a\mu_a|F_{\mathrm{hfs}}(r)\frac{(\boldsymbol{\alpha}\times\boldsymbol{r})_i}{r^3}|n\kappa_n\mu_n\rangle
×⟨nκnμn∣Fhfs(r)(α×r)ir3∣naκaμa⟩.(9){}\times\langle n\kappa_n\mu_n|F_{\mathrm{hfs}}(r)\frac{(\boldsymbol{\alpha}\times\boldsymbol{r})_i}{r^3}|n_a\kappa_a\mu_a\rangle. \tag*{(9)}

where we introduced the nuclear gg-factor gg as

μ=q2MgI,(10)\boldsymbol{\mu}=\frac{q}{2M}g\boldsymbol{I}, \tag*{(10)}

where q=Z∣e∣q=Z|e|. Performing the angular-momentum algebra in Eq. (9), we obtain, for the 1s reference state (κa=−1\kappa_a=-1),

Ehfs2=23(Zα)2m3(g2M)2I(I+1)E,(11)E_{\mathrm{hfs}2}=\frac{2}{3}(Z\alpha)^2m^3\left(\frac{g}{2M}\right)^2I(I+1)\mathcal{E}, \tag*{(11)}

where E\mathcal{E} is

E=∑κn=−1,2aκn∑n1εa−εn{∫0∞x2dxFhfs(x)x2[ga(x)fn(x)+fa(x)gn(x)]}2.(12)\mathcal{E}=\sum_{\kappa_n=-1,2}a_{\kappa_n}\sum_n\frac{1}{\varepsilon_a-\varepsilon_n} \left\{\int_0^\infty x^2dx\frac{F_{\mathrm{hfs}}(x)}{x^2}[g_a(x)f_n(x)+f_a(x)g_n(x)]\right\}^2. \tag*{(12)}

Here, a−1=2/3a_{-1}=2/3, a2=1/3a_2=1/3, and ga(x)g_a(x) and fa(x)f_a(x) are the upper and the lower radial components of the Dirac wave function of the reference state. For the dipole parametrization with rM=rCr_M=r_C, the magnetic and electric distribution functions are

Fhfs(x)=1−e−λx[1+λx+12(λx)2],F_{\mathrm{hfs}}(x)=1-e^{-\lambda x}\left[1+\lambda x+\frac{1}{2}(\lambda x)^2\right],
FC(x)=1−(1+λx2)e−λx,(13)F_C(x)=1-\left(1+\frac{\lambda x}{2}\right)e^{-\lambda x}, \tag*{(13)}

with λ=23/rC\lambda=2\sqrt{3}/r_C. We evaluate numerically Eq. (12) for the dipole nuclear model by representing the summation over the spectrum as a sum over the pseudostates in the basis of B-splines [18, 19]. A series of calculations were performed for a fixed nuclear charge ZZ but for different nuclear radii rCr_C. The obtained results were fitted to the form

E(rC)=(Zα)4[aln⁡rC+b+crC+…],(14)\mathcal{E}(r_C)=(Z\alpha)^4\left[a\ln r_C+b+cr_C+\ldots\right], \tag*{(14)}

which yielded the following result for the leading logarithmic coefficient,

a=5.00(1),(15)a=5.00(1), \tag*{(15)}

in agreement with the result of analytical calculations in Eq. (74). Next, we performed numerical calculations of E\mathcal{E} for different values of ZZ with a fixed value of rCr_C and fitted the results to the form

E(Zα)=(Zα)4[5ln⁡(ZαmrC)+B+ZαC+…].(16)\mathcal{E}(Z\alpha)=(Z\alpha)^4\left[5\ln(Z\alpha m r_C)+B+Z\alpha C+\ldots\right]. \tag*{(16)}

We obtained the nonlogarithmic coefficient BB as

B=2.700(2),(17)B=2.700(2), \tag*{(17)}

again in agreement with the analytical result in Eq. (74). The numerical results for the leading ZαZ\alpha-expansion coefficients serve as a valuable test of the correctness of the analytical derivation for few-electron atoms.

NRQED approach for a few electron systems

For light few-electron systems, the best computational approach is that based on nonrelativistic quantum electrodynamics (NRQED), because we can account accurately for relativistic effects and electron correlations on the same footing. In fact, the derivation is similar to that of the internuclear spin-spin interaction in molecules by Ramsey [20], with the difference that here we have the same nucleus and thus have to include the hard electron-nucleus interaction, which is highly singular. Therefore, we will use the dimension regularization for the derivation of NRQED operators, see Appendix § 10 for details. Because we know that the shift is proportional to I2\boldsymbol{I}^2, we can assume initially that I=1/2I=1/2 and later after cancellation of 1/ε1/\varepsilon terms, replace it by an arbitrary II. Therefore, in d=3−2εd=3-2\varepsilon dimensions the vector potential takes the form

Ahfsi(q)=−iZeg4MσNijqjq2ρ~(q2),(18)A_{\mathrm{hfs}}^i(\boldsymbol{q})=-iZe\frac{g}{4M}\sigma_N^{ij}\frac{q^j}{q^2}\widetilde{\rho}(q^2), \tag*{(18)}

where ρ~\widetilde{\rho} is a Fourier transform of ρ\rho in Eq. (5) and σNij\sigma_N^{ij} is dd-dimensional spin operator for the nucleus defined in Eq. (84). In the coordinate representation it is

Ahfsi(r)=−iZeg4MσNij∫ddq(2π)deiq⋅rqjq2ρ~(q2)=Ze4πg4MσNij[rjr3]fs,ε.(19)\begin{aligned} A_{\mathrm{hfs}}^i(\boldsymbol{r})=-iZe\frac{g}{4M}\sigma_N^{ij}\int\frac{d^dq}{(2\pi)^d}e^{i\boldsymbol{q}\cdot\boldsymbol{r}}\frac{q^j}{q^2}\widetilde{\rho}(q^2) \\ &= \frac{Z e}{4\pi}\frac{g}{4M}\sigma_N^{ij}\left[\frac{r^j}{r^3}\right]_{\mathrm{fs},\varepsilon}. \tag*{(19)} \end{aligned}

In d=3d=3 dimensions, σij=ϵijkσk\sigma^{ij}=\epsilon^{ijk}\sigma^k and for a point nucleus A⃗hfs(r⃗)\vec{A}_{\mathrm{hfs}}(\vec{r}) becomes

A⃗hfs(r⃗)=−Ze4πg2MI⃗×r⃗r3=14πμ⃗×r⃗r3,(20)\vec{A}_{\mathrm{hfs}}(\vec{r})=-\frac{Z e}{4\pi}\frac{g}{2M}\vec{I}\times\frac{\vec{r}}{r^3}=\frac{1}{4\pi}\vec{\mu}\times\frac{\vec{r}}{r^3}, \tag*{(20)}

as it should.

The leading second-order hyperfine-interaction correction is of the order mα6(m/M)2m\alpha^6(m/M)^2. We represent it as a sum of three terms

Ehfs2=Esec+Ediam+Ehigh,(21)E_{\mathrm{hfs}2}=E_{\mathrm{sec}}+E_{\mathrm{diam}}+E_{\mathrm{high}}, \tag*{(21)}

which we calculate in the following three sections. Similarly to the first-order HFS correction, which is proportional to the Fermi coefficient EFE_F, the second-order HFS correction is proportional to the second-order Fermi coefficient EF2E_{F2},

EF2=α6m(Zg2M)2I(I+1).(22)E_{F2}=\alpha^6m\left(\frac{Z g}{2M}\right)^2 I(I+1). \tag*{(22)}

EsecE_{\mathrm{sec}}

EsecE_{\mathrm{sec}} is a second-order correction of the form

Esec=⟨Hhfs1E−HHhfs⟩,(23)E_{\mathrm{sec}}=\left\langle H_{\mathrm{hfs}}\frac{1}{E-H}H_{\mathrm{hfs}}\right\rangle, \tag*{(23)}

where HH is the nonrelativistic Hamiltonian

H=p122m+p222m+[αr12−Zαr1−Zαr2]ε,(24)H=\frac{p_1^2}{2m}+\frac{p_2^2}{2m}+\left[\frac{\alpha}{r_{12}}-\frac{Z\alpha}{r_1}-\frac{Z\alpha}{r_2}\right]_{\varepsilon}, \tag*{(24)}

and HhfsH_{\mathrm{hfs}} is the leading hyperfine interaction. It is obtained from the nonrelativistic expansion of VhfsV_{\mathrm{hfs}} for a point nucleus

Vhfs=−Zαg4MσNijαiσij[rjr3]ε,(25)V_{\mathrm{hfs}}=-\frac{Z\alpha g}{4M}\sigma_N^{ij}\alpha^i\sigma^{ij}\left[\frac{r^j}{r^3}\right]_{\varepsilon}, \tag*{(25)}

namely

Hhfs=−Zαg4MσNij{σi[rjr3]ε,σkpk2m},(26)H_{\mathrm{hfs}}=-\frac{Z\alpha g}{4M}\sigma_N^{ij}\left\{\sigma^i\left[\frac{r^j}{r^3}\right]_{\varepsilon},\frac{\sigma^k p^k}{2m}\right\}, \tag*{(26)}

where {A,B}=AB+BA\{A,B\}=AB+BA. In dd-dimensions

σiσk=12{σi,σk}+12[σi,σk]=δik+iσik,(27)\sigma^i\sigma^k=\frac{1}{2}\{\sigma^i,\sigma^k\}+\frac{1}{2}[\sigma^i,\sigma^k]=\delta^{ik}+i\sigma^{ik}, \tag*{(27)}

thus

Hhfs=−Zαg4mMσNij(σiki2[rjr3,pk]ε+[rjr3]εpi)=HhfsA+HhfsB+HhfsC,(28)\begin{aligned} H_{\mathrm{hfs}}=-\frac{Z\alpha g}{4mM}\sigma_N^{ij}\left(\sigma^{ik}\frac{i}{2}\left[\frac{r^j}{r^3},p^k\right]_{\varepsilon}+\left[\frac{r^j}{r^3}\right]_{\varepsilon}p^i\right) \\ &=H_{\mathrm{hfsA}}+H_{\mathrm{hfsB}}+H_{\mathrm{hfsC}}, \tag*{(28)} \end{aligned}

where after summing over all electrons

HhfsA=∑aZαg2mMσNijσaijπdδd(ra),(29)H_{\mathrm{hfsA}}=\sum_a\frac{Z\alpha g}{2mM}\sigma_N^{ij}\sigma_a^{ij}\frac{\pi}{d}\delta^d(r_a), \tag*{(29)}
HhfsB=∑aZαg8mMσNijσaik[δjkra3−3rajrakra5]ε,(30)H_{\mathrm{hfsB}}=\sum_a\frac{Z\alpha g}{8mM}\sigma_N^{ij}\sigma_a^{ik}\left[\frac{\delta^{jk}}{r_a^3}-3\frac{r_a^j r_a^k}{r_a^5}\right]_{\varepsilon}, \tag*{(30)}
HhfsCd=3=∑aZαg2mMI⃗⋅r⃗ara3×p⃗a.(31)H_{\mathrm{hfsC}}^{d=3}=\sum_a\frac{Z\alpha g}{2mM}\vec{I}\cdot\frac{\vec{r}_a}{r_a^3}\times\vec{p}_a. \tag*{(31)}

The corresponding second-order contributions

Esec=EsecA+EsecB+EsecC(32)E_{\mathrm{sec}}=E_{\mathrm{secA}}+E_{\mathrm{secB}}+E_{\mathrm{secC}} \tag*{(32)}

are angular averaged [see Eqs. (85) and (A3) in Appendix A] and expressed in atomic units. We proceed first with the calculation of EsecAE_{\mathrm{secA}},

EsecA=⟨HhfsA1(E−H)′HhfsA⟩=EF2d3(d−1)∑a,b⟨σaij4πδd(ra)1(E−H)′σbij4πδd(rb)⟩,(33)\begin{aligned} E_{\mathrm{secA}}=\left\langle H_{\mathrm{hfsA}}\frac{1}{(E-H)'}H_{\mathrm{hfsA}}\right\rangle\\ &=\frac{E_{F2}}{d^3(d-1)}\sum_{a,b}\left\langle\sigma_a^{ij}4\pi\delta^d(r_a)\frac{1}{(E-H)'}\sigma_b^{ij}4\pi\delta^d(r_b)\right\rangle, \tag*{(33)} \end{aligned}

where atomic units are used for the matrix elements, and EF2E_{F2} is assumed to be the dd-dimensional generalization of (22) with I(I+1)→⟨I⃗ 2⟩εI(I+1)\to\langle\vec{I}^{\,2}\rangle_\varepsilon. The expression in (33) is singular and needs to be transformed to the regular form by pulling out the 1/ε1/\varepsilon singularity from the second-order matrix elements. This is achieved by the use of the identity

Ha∣ϕ⟩≡{H−E,Qa}∣ϕ⟩+Va∣ϕ⟩,(34)H_a|\phi\rangle\equiv\{H-E,Q_a\}|\phi\rangle+V_a|\phi\rangle, \tag*{(34)}

where ϕ\phi is the reference state,

Ha=4πδd(ra),(35)H_a=4\pi\delta^d(r_a), \tag*{(35)}
Qa=2[1ra]ε,(36)Q_a=2\left[\frac{1}{r_a}\right]_{\varepsilon}, \tag*{(36)}
Va=−2[r⃗ara3]ε⋅∇⃗a,(37)V_a=-2\left[\frac{\vec{r}_a}{r_a^3}\right]_{\varepsilon}\cdot\vec{\nabla}_a, \tag*{(37)}

to obtain

⟨Ha1(E−H)′Hb⟩=⟨Va+1(E−H)′Vb⟩+⟨Ha⟩⟨Qb⟩+⟨Hb⟩⟨Qa⟩−⟨HaQb⟩−⟨HbQa⟩+⟨Qa(H−E)Qb⟩,(38)\left\langle H_a\frac{1}{(E-H)'}H_b\right\rangle =\left\langle V_a^+\frac{1}{(E-H)'}V_b\right\rangle+\langle H_a\rangle\langle Q_b\rangle+\langle H_b\rangle\langle Q_a\rangle-\langle H_aQ_b\rangle-\langle H_bQ_a\rangle+\langle Q_a(H-E)Q_b\rangle, \tag*{(38)}

where Va+V_a^+ is a Hermitian conjugation of VaV_a. The resulting form of EsecAE_{\mathrm{secA}} is

EsecA=EF21d3(d−1)∑a,bσaijσbij[⟨Va+1E−H′Vb⟩+⟨4πδ3(ra)⟩⟨2rb⟩+⟨4πδ3(rb)⟩⟨2ra⟩−⟨4πδd(ra) 2[1rb]ε⟩−⟨4πδd(rb) 2[1ra]ε⟩+2δab⟨[r⃗ara3]ε⋅[r⃗brb3]ε⟩].(39)E_{\mathrm{secA}}=E_{F2}\frac{1}{d^3(d-1)}\sum_{a,b}\sigma_a^{ij}\sigma_b^{ij}\left[\left\langle V_a^+\frac{1}{E-H'}V_b\right\rangle+\left\langle4\pi\delta^3(r_a)\right\rangle\left\langle\frac{2}{r_b}\right\rangle+\left\langle4\pi\delta^3(r_b)\right\rangle\left\langle\frac{2}{r_a}\right\rangle-\left\langle4\pi\delta^d(r_a)\,2\left[\frac{1}{r_b}\right]_{\varepsilon}\right\rangle-\left\langle4\pi\delta^d(r_b)\,2\left[\frac{1}{r_a}\right]_{\varepsilon}\right\rangle+2\delta_{ab}\left\langle\left[\frac{\vec r_a}{r_a^3}\right]_{\varepsilon}\cdot\left[\frac{\vec r_b}{r_b^3}\right]_{\varepsilon}\right\rangle\right]. \tag*{(39)}

For the second-order tensor contribution EsecBE_{\mathrm{secB}} we use Eq. (86) to contract indices, and

EsecB=⟨HhfsB1E−HHhfsB⟩=EF2d−22d2(d−1)2∑a,b⟨σakl[δijra3−3rairajra5]ε1E−Hσbkl[δijrb3−3rbirbjrb5]ε⟩,(40)E_{\mathrm{secB}}=\left\langle H_{\mathrm{hfsB}}\frac{1}{E-H}H_{\mathrm{hfsB}}\right\rangle=E_{F2}\frac{d-2}{2d^2(d-1)^2}\sum_{a,b}\left\langle\sigma_a^{kl}\left[\frac{\delta^{ij}}{r_a^3}-3\frac{r_a^i r_a^j}{r_a^5}\right]_{\varepsilon}\frac{1}{E-H}\sigma_b^{kl}\left[\frac{\delta^{ij}}{r_b^3}-3\frac{r_b^i r_b^j}{r_b^5}\right]_{\varepsilon}\right\rangle, \tag*{(40)}

we use the following identity to pull out 1/ε1/\varepsilon singularity from the matrix elements:

Haij∣ϕ⟩≡{H−E,Qaij}∣ϕ⟩+Vaij∣ϕ⟩,(41)H_a^{ij}|\phi\rangle\equiv\{H-E,Q_a^{ij}\}|\phi\rangle+V_a^{ij}|\phi\rangle, \tag*{(41)}

where

Haij=[δijra3−3rairajra5]ε=(−∇i∇j+δijd∇⃗2)V(ra),(42)H_a^{ij}=\left[\frac{\delta^{ij}}{r_a^3}-3\frac{r_a^i r_a^j}{r_a^5}\right]_{\varepsilon}=\left(-\nabla^i\nabla^j+\frac{\delta^{ij}}{d}\vec\nabla^2\right)\mathcal{V}(r_a), \tag*{(42)}
Qaij=2[δij6ra−rairaj2ra3]ε=2(−∇i∇j+δijd∇⃗2)V(2)(ra),(43)Q_a^{ij}=2\left[\frac{\delta^{ij}}{6r_a}-\frac{r_a^i r_a^j}{2r_a^3}\right]_{\varepsilon}=2\left(-\nabla^i\nabla^j+\frac{\delta^{ij}}{d}\vec\nabla^2\right)\mathcal{V}^{(2)}(r_a), \tag*{(43)}
Vaij∣ϕ⟩=[−δijrak+3δikraj+3δjkrai3ra3+3rairajrakra5]∇ak∣ϕ⟩,(44)V_a^{ij}|\phi\rangle=\left[-\frac{\delta^{ij}r_a^k+3\delta^{ik}r_a^j+3\delta^{jk}r_a^i}{3r_a^3}+\frac{3r_a^i r_a^j r_a^k}{r_a^5}\right]\nabla_a^k|\phi\rangle, \tag*{(44)}

with V\mathcal{V} and V(2)\mathcal{V}^{(2)} defined in Appendix A. We obtain

⟨Haij1E−HHbij⟩=⟨Va+ij1E−HVbij⟩−⟨HaijQbij⟩−⟨HbijQaij⟩+⟨Qaij(H−E)Qbij⟩.(45)\left\langle H_a^{ij}\frac{1}{E-H}H_b^{ij}\right\rangle=\left\langle V_a^{+ij}\frac{1}{E-H}V_b^{ij}\right\rangle-\left\langle H_a^{ij}Q_b^{ij}\right\rangle-\left\langle H_b^{ij}Q_a^{ij}\right\rangle+\left\langle Q_a^{ij}(H-E)Q_b^{ij}\right\rangle. \tag*{(45)}

The result for EsecBE_{\mathrm{secB}} is then

EsecB=EF2d−22d2(d−1)2∑a,bσastσbst[⟨Va+ij1E−HVbij⟩−2⟨[δijra3−3rairajra5]ε[δij6rb−rbirbj2rb3]ε⟩−2⟨[δijrb3−3rbirbjrb5]ε[δij6ra−rairaj2ra3]ε⟩+2δab⟨∇k[δij6ra−rairaj2ra3]ε∇bk[δij6rb−rbirbj2rb3]ε⟩].(46)\begin{aligned} E_{\mathrm{secB}}={}&E_{F2}\frac{d-2}{2d^2(d-1)^2}\sum_{a,b}\sigma_a^{st}\sigma_b^{st}\left[\left\langle V_a^{+ij}\frac{1}{E-H}V_b^{ij}\right\rangle-2\left\langle\left[\frac{\delta^{ij}}{r_a^3}-3\frac{r_a^i r_a^j}{r_a^5}\right]_{\varepsilon}\left[\frac{\delta^{ij}}{6r_b}-\frac{r_b^i r_b^j}{2r_b^3}\right]_{\varepsilon}\right\rangle\right.\\ &\left.-2\left\langle\left[\frac{\delta^{ij}}{r_b^3}-3\frac{r_b^i r_b^j}{r_b^5}\right]_{\varepsilon}\left[\frac{\delta^{ij}}{6r_a}-\frac{r_a^i r_a^j}{2r_a^3}\right]_{\varepsilon}\right\rangle+2\delta_{ab}\left\langle\nabla^k\left[\frac{\delta^{ij}}{6r_a}-\frac{r_a^i r_a^j}{2r_a^3}\right]_{\varepsilon}\nabla_b^k\left[\frac{\delta^{ij}}{6r_b}-\frac{r_b^i r_b^j}{2r_b^3}\right]_{\varepsilon}\right\rangle\right]. \tag*{(46)} \end{aligned}

Spin-orbit contribution EsecCE_{\mathrm{secC}} does not require any regularization,

EsecC=⟨HhfsC1E−HHhfsC⟩=EF213∑a,b⟨r⃗ara3×p⃗a1E−Hr⃗brb3×p⃗b⟩,(47)E_{\mathrm{secC}}=\left\langle H_{\mathrm{hfsC}}\frac{1}{E-H}H_{\mathrm{hfsC}}\right\rangle =E_{F2}\frac{1}{3}\sum_{a,b}\left\langle\frac{\vec r_a}{r_a^3}\times\vec p_a\frac{1}{E-H}\frac{\vec r_b}{r_b^3}\times\vec p_b\right\rangle, \tag*{(47)}

and can be calculated directly in d=3d=3 dimensions.

EdiamE_{\mathrm{diam}}

The diamagnetic contribution in natural units is

Ediam=∑ae22m⟨A⃗(r⃗a)2⟩,(48)E_{\mathrm{diam}}=\sum_a\frac{e^2}{2m}\left\langle\vec A(\vec r_a)^2\right\rangle, \tag*{(48)}

where

eA⃗(r⃗a)=Zαg4MσNij[rajra3]ε.(49)e\vec A(\vec r_a)=\frac{Z\alpha g}{4M}\sigma_N^{ij}\left[\frac{r_a^j}{r_a^3}\right]_{\varepsilon}. \tag*{(49)}

After transforming to atomic units and performing the angular average, it becomes

Ediam=EF21d∑a⟨1ra4⟩ε.(50)E_{\mathrm{diam}}=E_{F2}\frac{1}{d}\sum_a\left\langle\frac{1}{r_a^4}\right\rangle_{\varepsilon}. \tag*{(50)}

EdiamE_{\mathrm{diam}}

The high-energy contribution is obtained from the scattering amplitude, and is essentially the same as for the hydrogenic case. In momentum space and natural units, we have

V(q⃗)=−4πZαq⃗ 2ρ~(q2),(51)V(\vec{q})=-\frac{4\pi Z\alpha}{\vec{q}^{\,2}}\widetilde{\rho}(q^2), \tag*{(51)}
Vhfs(q⃗)=i4πZαg4Mαiσijqjq2ρ~(q2).(52)V_{\mathrm{hfs}}(\vec{q})=i4\pi Z\alpha\frac{g}{4M}\alpha^i\sigma^{ij}\frac{q^j}{q^2}\widetilde{\rho}(q^2). \tag*{(52)}

Then,

Ehigh=ϕ2(0)∫ddq1(2π)d∫ddq2(2π)dV(q⃗1−q⃗2)×⟨t∣γ0SF(q2−q1)γ0Vhfs(−q⃗1)SF(q2)γ0Vhfs(q⃗2)+γ0Vhfs(−q⃗1)SF(q1)γ0SF(q2)γ0Vhfs(q⃗2)+γ0Vhfs(−q⃗1)SF(q1)γ0Vhfs(q⃗2)SF(q1−q2)γ0∣t⟩,(53)\begin{aligned} E_{\mathrm{high}}={}&\phi^2(0)\int\frac{d^d q_1}{(2\pi)^d}\int\frac{d^d q_2}{(2\pi)^d}V(\vec{q}_1-\vec{q}_2)\\ &\times\langle t|\gamma^0 S_F(q_2-q_1)\gamma^0 V_{\mathrm{hfs}}(-\vec{q}_1)S_F(q_2)\gamma^0V_{\mathrm{hfs}}(\vec{q}_2)\\ &+\gamma^0V_{\mathrm{hfs}}(-\vec{q}_1)S_F(q_1)\gamma^0S_F(q_2)\gamma^0V_{\mathrm{hfs}}(\vec{q}_2)\\ &+\gamma^0V_{\mathrm{hfs}}(-\vec{q}_1)S_F(q_1)\gamma^0V_{\mathrm{hfs}}(\vec{q}_2)S_F(q_1-q_2)\gamma^0|t\rangle, \tag*{(53)} \end{aligned}

where SFS_F is the fermion propagator and t=(m,0⃗)t=(m,\vec{0}). Performing traces with Dirac matrices, one obtains

Ehigh=−ϕ2(0)(4πZα)3(g2M)24d(d−1)⟨I⃗ 2⟩ε∫ddq1(2π)d∫ddq2(2π)dρ~(q12)ρ~(q22)ρ~(q32)q12q22q32(1+(d−2)(q⃗1⋅q⃗2)(q⃗2⋅q⃗3)(q⃗3⋅q⃗1)q12q22q32),(54)E_{\mathrm{high}}=-\phi^2(0)(4\pi Z\alpha)^3\left(\frac{g}{2M}\right)^2\frac{4}{d(d-1)}\langle\vec{I}^{\,2}\rangle_\varepsilon \int\frac{d^d q_1}{(2\pi)^d}\int\frac{d^d q_2}{(2\pi)^d}\frac{\widetilde{\rho}(q_1^2)\widetilde{\rho}(q_2^2)\widetilde{\rho}(q_3^2)}{q_1^2q_2^2q_3^2} \left(1+(d-2)\frac{(\vec{q}_1\cdot\vec{q}_2)(\vec{q}_2\cdot\vec{q}_3)(\vec{q}_3\cdot\vec{q}_1)}{q_1^2q_2^2q_3^2}\right), \tag*{(54)}

where q⃗3=q⃗1−q⃗2\vec{q}_3=\vec{q}_1-\vec{q}_2. After transforming integral to the coordinate representation, it becomes

Ehigh=−ϕ2(0)(Zα)3(g4M)216d(d−1)⟨I⃗ 2⟩ε∫ddr[Vρ(r)3−(d−2)∇i∇jVρ(2)(r)∇j∇kVρ(2)(r)∇k∇iVρ(2)(r)]=EhighA+EhighB,(55)\begin{aligned} E_{\mathrm{high}}={}&-\phi^2(0)(Z\alpha)^3\left(\frac{g}{4M}\right)^2\frac{16}{d(d-1)}\langle\vec{I}^{\,2}\rangle_\varepsilon\int d^d r\left[V_\rho(r)^3-(d-2)\nabla^i\nabla^jV_\rho^{(2)}(r)\nabla^j\nabla^kV_\rho^{(2)}(r)\nabla^k\nabla^iV_\rho^{(2)}(r)\right]\\ ={}&E_{\mathrm{highA}}+E_{\mathrm{highB}}, \tag*{(55)} \end{aligned}

where Vρ(r)V_\rho(r) and Vρ(2)(r)V_\rho^{(2)}(r) are defined in Appendix A. EhighAE_{\mathrm{highA}} includes the r⃗\vec{r} integral over a small region of the nuclear charge distribution, where we can set d=3d=3. It is evaluated as

EhighA=−ϕ2(0)(Zα)3(g4M)283⟨I⃗ 2⟩∫Rd3r[Vρ(r)3−∇i∇jVρ(2)(r)∇j∇kVρ(2)(r)∇k∇iVρ(2)(r)].(56)E_{\mathrm{highA}}=-\phi^2(0)(Z\alpha)^3\left(\frac{g}{4M}\right)^2\frac{8}{3}\langle\vec{I}^{\,2}\rangle\int_R d^3r\left[V_\rho(r)^3-\nabla^i\nabla^jV_\rho^{(2)}(r)\nabla^j\nabla^kV_\rho^{(2)}(r)\nabla^k\nabla^iV_\rho^{(2)}(r)\right]. \tag*{(56)}

It is convenient at this point to use the explicit form of the nuclear charge distribution, which we chose to be dipole parametrization, the same as used in the hydrogenic case in Sec. II,

ρ~(p2)=λ4(λ2+p2)2.(57)\widetilde{\rho}(p^2)=\frac{\lambda^4}{(\lambda^2+p^2)^2}. \tag*{(57)}

Thus, d=3d=3 potentials become

4πρ(r)=4π∫d3p(2π)3ρ~(p2)eip⃗⋅r⃗=λ32e−λr,(58)4\pi\rho(r)=4\pi\int\frac{d^3p}{(2\pi)^3}\widetilde{\rho}(p^2)e^{i\vec{p}\cdot\vec{r}}=\frac{\lambda^3}{2}e^{-\lambda r}, \tag*{(58)}
Vρ(r)=4π∫d3p(2π)3ρ~(p2)p2eip⃗⋅r⃗=1r−e−λrr−λ2e−λr,(59)V_\rho(r)=4\pi\int\frac{d^3p}{(2\pi)^3}\frac{\widetilde{\rho}(p^2)}{p^2}e^{i\vec{p}\cdot\vec{r}}=\frac{1}{r}-\frac{e^{-\lambda r}}{r}-\frac{\lambda}{2}e^{-\lambda r}, \tag*{(59)}
Vρ(2)(r)=4π∫d3p(2π)3ρ~(p2)p4eip⃗⋅r⃗−1p4=−r2−2λ2r+12λe−λr+2λ2e−λrr.(60)V_\rho^{(2)}(r)=4\pi\int\frac{d^3p}{(2\pi)^3}\frac{\widetilde{\rho}(p^2)}{p^4}e^{i\vec{p}\cdot\vec{r}}-\frac{1}{p^4}=-\frac{r}{2}-\frac{2}{\lambda^2r}+\frac{1}{2\lambda}e^{-\lambda r}+\frac{2}{\lambda^2}\frac{e^{-\lambda r}}{r}. \tag*{(60)}

Now the rr-integration in Eq. (56) can be performed analytically, with the result

EhighA=−ϕ2(0)(4πZα)3(g4M)283⟨I⃗ 2⟩5π(−467540+2ln⁡25+3ln⁡310−ln⁡rCR+γ),(61)E_{\mathrm{highA}}=-\phi^2(0)(4\pi Z\alpha)^3\left(\frac{g}{4M}\right)^2\frac{8}{3}\langle\vec{I}^{\,2}\rangle5\pi\left(-\frac{467}{540}+\frac{2\ln2}{5}+\frac{3\ln3}{10}-\ln\frac{r_C}{R}+\gamma\right), \tag*{(61)}

where rC=23/λr_C=2\sqrt{3}/\lambda. It should be mentioned that the description of the nucleus through the elastic form-factors ρC\rho_C and ρM\rho_M here is not fully justified. It is because for high momenta ∼ℏ/rC\sim\hbar/r_C the electron sees individual nucleons, rather than the nucleus as a whole. The detailed analysis of this contribution is complicated and should be performed together with other three-photon exchange corrections [21]. We will use the above form-factor result for comparison with direct numerical calculations for hydrogenic systems described in Sec. II, keeping in mind that for the 23S−21S2^3S-2^1S transition in 3He{}^3\mathrm{He} it is negligible.

The part EhighBE_{\mathrm{highB}} of the high energy contribution involves integral outside of the nuclear region, thus we can assume the point nucleus and obtain

EhighB=−ϕ2(0)(Zα)3(g4M)216d(d−1)⟨I⃗ 2⟩ε∫Rddr[V(r)3−(d−2)∇i∇jV(2)(r)∇j∇kV(2)(r)∇k∇iV(2)(r)]=ϕ2(0)(Zα)3(g4M)2⟨T⃗2⟩ε[(4π)εΓ(1+ε)]2403π(2330+14ε+ln⁡R+γ),(62)\begin{aligned} E_{\mathrm{highB}}=-\phi^2(0)(Z\alpha)^3\left(\frac{g}{4M}\right)^2\frac{16}{d(d-1)}\langle\vec{I}^{\,2}\rangle_\varepsilon\int_R d^d r\left[V(r)^3-(d-2)\nabla^i\nabla^jV^{(2)}(r)\nabla^j\nabla^kV^{(2)}(r)\nabla^k\nabla^iV^{(2)}(r)\right] \\ &= \phi^{2}(0)(Z\alpha)^{3}\left(\frac{g}{4M}\right)^{2}\left\langle\vec{T}^{2}\right\rangle_{\varepsilon}\left[(4\pi)^{\varepsilon}\Gamma(1+\varepsilon)\right]^{2}\frac{40}{3}\pi\left(\frac{23}{30}+\frac{1}{4\varepsilon}+\ln R+\gamma\right), \tag*{(62)} \end{aligned}

and by convention the factor [(4π)εΓ(1+ε)]2[(4\pi)^{\varepsilon}\Gamma(1+\varepsilon)]^{2} is consistently pulled out from all the terms. The sum of both parts is independent on arbitrary cutoff RR, and is

Ehigh=(Zα)3(g2M)2⟨T⃗2⟩ε103π(881540−2ln⁡25−3ln⁡310+14ε+ln⁡rC)⟨δd(r)⟩.(63)E_{\mathrm{high}}=(Z\alpha)^{3}\left(\frac{g}{2M}\right)^{2}\left\langle\vec{T}^{2}\right\rangle_{\varepsilon}\frac{10}{3}\pi\left(\frac{881}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\frac{1}{4\varepsilon}+\ln r_{C}\right)\left\langle\delta^{d}(r)\right\rangle. \tag*{(63)}

Transforming to atomic units by r⃗→(mα)−1/(1+2ε)r⃗\vec{r}\to(m\alpha)^{-1/(1+2\varepsilon)}\vec{r}, one pulls out the factor m(1−2ε)/(1+2ε)α2/(1+2ε)m^{(1-2\varepsilon)/(1+2\varepsilon)}\alpha^{2/(1+2\varepsilon)} from HH and obtains the nonrelativistic Hamiltonian in atomic units. Similarly, for H(6)H^{(6)} the common factor in atomic units

η=m(1−10ε)/(1+2ε)α6/(1+2ε)(64)\eta=m^{(1-10\varepsilon)/(1+2\varepsilon)}\alpha^{6/(1+2\varepsilon)} \tag*{(64)}

is pulled out from all the terms, which corresponds to the replacement m→1m\to1, α→1\alpha\to1, and

Ehigh=EF2103π(881540−2ln⁡25−3ln⁡310+14ε+ln⁡mαrC)∑aZ⟨δd(ra)⟩.(65)E_{\mathrm{high}}=E_{F2}\frac{10}{3}\pi\left(\frac{881}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\frac{1}{4\varepsilon}+\ln m\alpha r_{C}\right)\sum_{a}Z\left\langle\delta^{d}(r_{a})\right\rangle. \tag*{(65)}

Complete Result

In order to eliminate 1/ε1/\varepsilon singularity, we transform all singular operators (in atomic units) to the same form,

⟨[δijr3−3rirjr5]ε[δij6r−rirj2r3]ε⟩=(1−ε)⟨1r4⟩ε,(66)\left\langle\left[\frac{\delta^{ij}}{r^{3}}-3\frac{r^{i}r^{j}}{r^{5}}\right]_{\varepsilon}\left[\frac{\delta^{ij}}{6r}-\frac{r^{i}r^{j}}{2r^{3}}\right]_{\varepsilon}\right\rangle=(1-\varepsilon)\left\langle\frac{1}{r^{4}}\right\rangle_{\varepsilon}, \tag*{(66)}
⟨∇k[δij6r−rirj2r3]ε∇k[δij6r−rirj2r3]ε⟩=(d−1)(d2−2d+4)4d⟨1r4⟩ε=(76−3118ε)⟨1r4⟩ε+O(ε),(67)\left\langle\nabla^{k}\left[\frac{\delta^{ij}}{6r}-\frac{r^{i}r^{j}}{2r^{3}}\right]_{\varepsilon}\nabla^{k}\left[\frac{\delta^{ij}}{6r}-\frac{r^{i}r^{j}}{2r^{3}}\right]_{\varepsilon}\right\rangle=\frac{(d-1)(d^{2}-2d+4)}{4d}\left\langle\frac{1}{r^{4}}\right\rangle_{\varepsilon}=\left(\frac{7}{6}-\frac{31}{18}\varepsilon\right)\left\langle\frac{1}{r^{4}}\right\rangle_{\varepsilon}+O(\varepsilon), \tag*{(67)}

and use [22]

⟨1r4⟩ε=⟨1r4⟩+Z⟨πδd(r)⟩(−2ε+8)+O(ε),(68)\left\langle\frac{1}{r^{4}}\right\rangle_{\varepsilon}=\left\langle\frac{1}{r^{4}}\right\rangle+Z\left\langle\pi\delta^{d}(r)\right\rangle\left(-\frac{2}{\varepsilon}+8\right)+O(\varepsilon), \tag*{(68)}

where ⟨1/r4⟩\langle1/r^{4}\rangle is defined as integral from a small radius aa to infinity with the 1/a1/a and ln⁡a+γ\ln a+\gamma terms subtracted out. The 1/ε1/\varepsilon singularity cancels out, and we obtain the final result for the hydrogenic atom as

Ehfs2=EsecA+EsecB+Ediam+Ehigh=EF2[19⟨V+1(E−H)′V⟩+112⟨V+ij1E−HVij⟩+169⟨πδ3(r)⟩⟨1r⟩+512⟨1r4⟩\begin{aligned} E_{\mathrm{hfs2}}=E_{\mathrm{secA}}+E_{\mathrm{secB}}+E_{\mathrm{diam}}+E_{\mathrm{high}} \\ &=E_{F2}\left[\frac{1}{9}\left\langle V^{+}\frac{1}{(E-H)'}V\right\rangle+\frac{1}{12}\left\langle V^{+ij}\frac{1}{E-H}V^{ij}\right\rangle+\frac{16}{9}\left\langle\pi\delta^{3}(r)\right\rangle\left\langle\frac{1}{r}\right\rangle+\frac{5}{12}\left\langle\frac{1}{r^{4}}\right\rangle \right. \end{aligned}
+Z4n3103(1223540−2ln⁡25−3ln⁡310+ln⁡mαrC)].(69)\left. +\frac{Z^{4}}{n^{3}}\frac{10}{3}\left(\frac{1223}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\ln m\alpha r_{C}\right)\right]. \tag*{(69)}

Using the following expectation values with hydrogenic nSnS-state wave functions,

⟨1r4⟩=8Z4n3[−53+12n+16n2+γ+Ψ(n)−ln⁡n2+ln⁡Z],(70)\left\langle\frac{1}{r^{4}}\right\rangle=\frac{8Z^{4}}{n^{3}}\left[-\frac{5}{3}+\frac{1}{2n}+\frac{1}{6n^{2}}+\gamma+\Psi(n)-\ln\frac{n}{2}+\ln Z\right], \tag*{(70)}
⟨V+1(E−H)′V⟩=Z4(−563n5−24n4+83n3),(71)\left\langle V^{+}\frac{1}{(E-H)'}V\right\rangle=Z^{4}\left(-\frac{56}{3n^{5}}-\frac{24}{n^{4}}+\frac{8}{3n^{3}}\right), \tag*{(71)}
⟨V+ij1(E−H)Vij⟩=Z4(29n5−23n3),(72)\left\langle V^{+ij}\frac{1}{(E-H)}V^{ij}\right\rangle=Z^{4}\left(\frac{2}{9n^{5}}-\frac{2}{3n^{3}}\right), \tag*{(72)}

we obtain for the hydrogenic nSnS state

Ehfs2(n)=m(Zα)6(gm2M)2I(I+1)103n2[181270−310n+112n2+310ln⁡43−ln⁡(n)+γ+Ψ(n)+ln⁡mZαrC].(73)E_{\mathrm{hfs2}}(n)=m(Z\alpha)^{6}\left(\frac{gm}{2M}\right)^{2}I(I+1)\frac{10}{3n^{2}}\left[\frac{181}{270}-\frac{3}{10n}+\frac{1}{12n^{2}}+\frac{3}{10}\ln\frac{4}{3}-\ln(n)+\gamma+\Psi(n)+\ln mZ\alpha r_{C}\right]. \tag*{(73)}

Specifically, for 1S1S state it reduces to

Ehfs2(1)=m(Zα)6(gm2M)2I(I+1)103(0.540 008+ln⁡mZαrC)(74)E_{\mathrm{hfs2}}(1)=m(Z\alpha)^6\left(\frac{gm}{2M}\right)^2 I(I+1)\frac{10}{3}\left(0.540\,008+\ln mZ\alpha r_C\right) \tag*{(74)}

which is in perfect agreement with the results of our numerical calculations described in § 1.1. Namely, the analytical constant term 0.540 008 in Eq. (73) agrees with the result obtained by fitting the all-order numerical data, 2.700(2)/5=0.540 0(4)2.700(2)/5=0.540\,0(4).

The final result for the helium atom takes the form (in atomic units)

Ehfs2=EF2{∑a[512⟨1ra4⟩+103(1223540−2ln⁡25−3ln⁡310+ln⁡mαrC)Zπ⟨δ3(ra)⟩]+∑a,b[127⟨σ⃗aVa+1(E−H)′σ⃗bVb⟩+136⟨σ⃗aVa+ij1E−Hσ⃗bVbij⟩+13⟨r⃗ara3×p⃗a1E−Hr⃗brb3×p⃗a⟩]+1627∑a,bσ⃗a⋅σ⃗b⟨πδ3(ra)⟩⟨1rb⟩−1627∑a≠bσ⃗a⋅σ⃗b⟨πδ3(ra)rb⟩+118∑a≠bσ⃗a⋅σ⃗b⟨1ra3rb−3r⃗a⋅r⃗bra5rb3⟩},(75)\begin{aligned} E_{\mathrm{hfs2}}={}&E_{F2}\Bigg\{\sum_a\left[\frac{5}{12}\left\langle\frac{1}{r_a^4}\right\rangle+\frac{10}{3}\left(\frac{1223}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\ln m\alpha r_C\right)Z\pi\left\langle\delta^3(r_a)\right\rangle\right]\\ &+\sum_{a,b}\left[\frac{1}{27}\left\langle\vec{\sigma}_a V_a^+\frac{1}{(E-H)'}\vec{\sigma}_b V_b\right\rangle+\frac{1}{36}\left\langle\vec{\sigma}_a V_a^{+ij}\frac{1}{E-H}\vec{\sigma}_b V_b^{ij}\right\rangle+\frac{1}{3}\left\langle\frac{\vec r_a}{r_a^3}\times\vec p_a\frac{1}{E-H}\frac{\vec r_b}{r_b^3}\times\vec p_a\right\rangle\right]\\ &+\frac{16}{27}\sum_{a,b}\vec{\sigma}_a\cdot\vec{\sigma}_b\left\langle\pi\delta^3(r_a)\right\rangle\left\langle\frac{1}{r_b}\right\rangle-\frac{16}{27}\sum_{a\ne b}\vec{\sigma}_a\cdot\vec{\sigma}_b\left\langle\frac{\pi\delta^3(r_a)}{r_b}\right\rangle+\frac{1}{18}\sum_{a\ne b}\vec{\sigma}_a\cdot\vec{\sigma}_b\left\langle\frac{1}{r_a^3r_b}-3\frac{\vec r_a\cdot\vec r_b}{r_a^5r_b^3}\right\rangle\Bigg\}, \tag*{(75)} \end{aligned}

where EF2E_{F2} is defined in Eq. (22), the operator VaV_a is defined in Eq. (37), and VaijV_a^{ij} is defined in Eq. (44).

Numerical Results

We first consider the hydrogenlike atoms. Using Eq. (73), we obtain the second-order HFS correction to the 1S−2S1S-2S transition to be 0.86 kHz for H, 0.05 kHz for D, and 6.84 kHz for He+. Consequently, the contribution for the H-D isotope shift of the 1S−2S1S-2S transition energy is 0.81 kHz, which is twice the previously assumed uncertainty [21]. We, however, do not update the charge radius difference determined from the H-D isotope shift, as the corresponding inelastic three-photon exchange contribution must also be considered, which could be as large as the elastic one calculated here. Notably, the inelastic two-photon exchange contribution has already been accounted for the H-D isotope shift in Refs. [21, 23, 24].

Now let us consider the helium atom. For the singlet states the total electron spin vanishes and, therefore, σ⃗1⋅σ⃗2=−3\vec{\sigma}_1\cdot\vec{\sigma}_2=-3, (σ⃗1−σ⃗2)2=12(\vec{\sigma}_1-\vec{\sigma}_2)^2=12. Using notations from Table I and Table II, we write the complete result for the second-order HFS correction for singlet SS states as

21S2^{1}S23S2^{3}S
Q1Q_14πδ3(r1)4\pi\delta^3(r_1)16.455 16916.592 071
Q3Q_34πδ3(r1)/r24\pi\delta^3(r_1)/r_25.593 7444.648 724
Q53Q_{53}1/r11/r_11.135 4081.154 664
Q57Q_{57}1/r141/r_1^425.531 68025.511 838
Q65Q_{65}[r12r22−3(r⃗1⋅r⃗2)2]/(r15r23)[r_1^2r_2^2-3(\vec r_1\cdot\vec r_2)^2]/(r_1^5r_2^3)0.015 0240.001 629

Table I. Expectation values of various operators for the 21S2^{1}S and 23S2^{3}S states, in atomic units.

StateTermExpressionSymmetryValue
21S2^{1}SS0S_0⟨21S∣4π[δ3(r1)−δ3(r2)]∣23S⟩2[E0(21S)−E0(23S)]−1\langle2^{1}S|4\pi[\delta^3(r_1)-\delta^3(r_2)]|2^{3}S\rangle^2[E_0(2^{1}S)-E_0(2^{3}S)]^{-1}29 015.120
21S2^{1}SS(3S)S(^{3}S)⟨(V1++V2+)1E0−H0(V1−V2)⟩\left\langle(V_1^+ + V_2^+)\frac{1}{E_0-H_0}(V_1-V_2)\right\rangle3S^{3}S27 703.473 2
21S2^{1}SS(3D)S(^{3}D)⟨(V1ij+−V2ij+)1E0−H0(V1ij−V2ij)⟩\left\langle(V_1^{ij+}-V_2^{ij+})\frac{1}{E_0-H_0}(V_1^{ij}-V_2^{ij})\right\rangle3De^{3}D^e−7.356-7.356 2
21S2^{1}SS(1Pe)S(^{1}P^e)⟨(r⃗1r13×p⃗1+r⃗2r23×p⃗2)1E0−H0′(r⃗1r13×p⃗1+r⃗2r23×p⃗2)⟩\left\langle\left(\frac{\vec r_1}{r_1^3}\times\vec p_1+\frac{\vec r_2}{r_2^3}\times\vec p_2\right)\frac{1}{E_0-H_0'}\left(\frac{\vec r_1}{r_1^3}\times\vec p_1+\frac{\vec r_2}{r_2^3}\times\vec p_2\right)\right\rangle1Pe^{1}P^e−0.012-0.012 469
23S2^{3}SS(3S)S(^{3}S)⟨(V1++V2+)1E0−H0′(V1+V2)⟩\left\langle(V_1^+ + V_2^+)\frac{1}{E_0-H_0'}(V_1+V_2)\right\rangle3S^{3}S−658.648-658.648 8
23S2^{3}SS(1S)S(^{1}S)⟨(V1+−V2+)1E0−H0′(V1−V2)⟩\left\langle(V_1^+ - V_2^+)\frac{1}{E_0-H_0'}(V_1-V_2)\right\rangle1S^{1}S−29 425.198-29\,425.198 9
23S2^{3}SS(3D)S(^{3}D)⟨(V1ij++V2ij+)1E0−H0′(V1ij+V2ij)⟩\left\langle(V_1^{ij+}+V_2^{ij+})\frac{1}{E_0-H_0'}(V_1^{ij}+V_2^{ij})\right\rangle3De^{3}D^e−7.463-7.463 4
23S2^{3}SS(1D)S(^{1}D)⟨(V1ij+−V2ij+)1E0−H0′(V1ij−V2ij)⟩\left\langle(V_1^{ij+}-V_2^{ij+})\frac{1}{E_0-H_0'}(V_1^{ij}-V_2^{ij})\right\rangle1De^{1}D^e−7.458-7.458 3
23S2^{3}SS(3Pe)S(^{3}P^e)⟨(r⃗1r13×p⃗1+r⃗2r23×p⃗2)1E0−H0′(r⃗1r13×p⃗1+r⃗2r23×p⃗2)⟩\left\langle\left(\frac{\vec r_1}{r_1^3}\times\vec p_1+\frac{\vec r_2}{r_2^3}\times\vec p_2\right)\frac{1}{E_0-H_0'}\left(\frac{\vec r_1}{r_1^3}\times\vec p_1+\frac{\vec r_2}{r_2^3}\times\vec p_2\right)\right\rangle3Pe^{3}P^e−0.004-0.004 694

Table II. Second-order matrix elements for 21S2^{1}S and 23S2^{3}S states of He, in atomic units.

Ehfs2=EF2{19S(3S)+112S(3D)+13S(1Pe)+53(1223540−2ln⁡25−3ln⁡310+ln⁡mαrC)ZQ1+89Q3+56Q57−13Q65}.(76)\begin{aligned} E_{\mathrm{hfs2}}=E_{F2}\Bigg\{&\frac{1}{9}S(^{3}S)+\frac{1}{12}S(^{3}D)+\frac{1}{3}S(^{1}P^{e})\\ &+\frac{5}{3}\left(\frac{1223}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\ln m\alpha r_C\right)ZQ_1\\ &+\frac{8}{9}Q_3+\frac{5}{6}Q_{57}-\frac{1}{3}Q_{65}\Bigg\}. \tag*{(76)} \end{aligned}

Our numerical calculations are carried out with the fully correlated exponential basis set by the method described in Ref. [25]. Numerical results for the first- and second-order matrix elements are summarized in Table I and Table II. The contribution to the ionization energy is obtained by evaluating Eq. (76) and subtracting the corresponding result for He+(1S)({}^{1}S). Our final result for the 21S{}^{2}{}^{1}S state of helium is expressed in the form

Ehfs2[He]−Ehfs2[He+]=EF29S0[1+δ(21S)],(77)E_{\mathrm{hfs2}}[\mathrm{He}]-E_{\mathrm{hfs2}}[\mathrm{He}^{+}]=\frac{E_{F2}}{9}S_0\left[1+\delta(^{2}{}^{1}S)\right], \tag*{(77)}

where S0S_0 describes the leading n=2n=2 hyperfine mixing and is defined in Table II. The correction δ\delta corresponds to the hyperfine mixing with n≠2n\ne2 states,

δ(21S)=−0.034 25,(78)\delta(^{2}{}^{1}S)=-0.034\,25, \tag*{(78)}

with the corresponding energy shift δEmix=−2.075\delta E_{\mathrm{mix}}=-2.075 kHz.

For the triplet states the total electron spin is equal to 1, thus σ⃗1⋅σ⃗2=1\vec{\sigma}_1\cdot\vec{\sigma}_2=1, (σ⃗1−σ⃗2)2=4(\vec{\sigma}_1-\vec{\sigma}_2)^2=4, (σ⃗1+σ⃗2)2=8(\vec{\sigma}_1+\vec{\sigma}_2)^2=8. Using notations from Table I and Table II, the complete result for the second-order HFS correction for triplet SS states can be expressed as

Ehfs2=EF2{227S(3S)+127S(1S)+118S(3D)+136S(1D)+53(1223540−2ln⁡25−3ln⁡310+ln⁡mαrC)ZQ1+13S(3Pe)+56Q57+3227Q1Q53−827Q3+19Q65}.(79)\begin{aligned} E_{\mathrm{hfs2}}=E_{F2}\Bigg\{&\frac{2}{27}S(^{3}S)+\frac{1}{27}S(^{1}S)+\frac{1}{18}S(^{3}D)+\frac{1}{36}S(^{1}D)\\ &+\frac{5}{3}\left(\frac{1223}{540}-\frac{2\ln2}{5}-\frac{3\ln3}{10}+\ln m\alpha r_C\right)ZQ_1\\ &+\frac{1}{3}S(^{3}P^{e})+\frac{5}{6}Q_{57}+\frac{32}{27}Q_1Q_{53}-\frac{8}{27}Q_3+\frac{1}{9}Q_{65}\Bigg\}. \tag*{(79)} \end{aligned}

With numerical results for the first- and second-order matrix elements from Table I and Table II, we obtain the contribution to the ionization energy of the 23S{}^{2}{}^{3}S state as

Ehfs2[He]−Ehfs2[He+]=−EF227S0[1+δ(23S)],(80)E_{\mathrm{hfs2}}[\mathrm{He}]-E_{\mathrm{hfs2}}[\mathrm{He}^{+}]=-\frac{E_{F2}}{27}S_0\left[1+\delta(^{2}{}^{3}S)\right], \tag*{(80)}

where

δ(23S)=0.015 08,(81)\delta(^{2}{}^{3}S)=0.015\,08, \tag*{(81)}

with the corresponding energy shift δEmix=−0.305\delta E_{\mathrm{mix}}=-0.305 kHz for the 23S{}^{2}{}^{3}S state.

23S2^{3}S-21S2^{1}S ISOTOPE SHIFT

In our previous calculation of the isotope shift of the 23S−21S2^{3}S-2^{1}S transition energy of helium [22], the second-order HFS contribution was accounted for approximately by taking the leading n=2n=2 mixing contributions S0S_0 and neglecting δ(21S)\delta(2^{1}S) and δ(23S)\delta(2^{3}S) in Eqs. (77) and (80). It was recently pointed out by Qi and coworkers [17] that these contributions are in fact significant and shift the theoretical predictions for the isotope shift beyond the previously estimated error bars. We now extend our previous theory by accounting for the complete second-order HFS correction calculated in the previous sections.

The individual theoretical contributions to the 3^3He-4^4He isotope shift of the 21S2^{1}S-23S2^{3}S centroid energies are summarized in Table III, in comparison with results of Qi and coworkers [17]. All numerical values apart from δEmix\delta E_{\mathrm{mix}} are taken from our previous investigation [22]. Note that the entries α6\alpha^6, α7\alpha^7, and α5Epol\alpha^5E_{\mathrm{pol}} were not independently reproduced by Qi and coworkers but taken from our work [22]. The deviation from Qi and coworkers for δEmix\delta E_{\mathrm{mix}} is due to the fact that they did not calculate the complete second-order HFS correction but used an approximate representation for it. Otherwise, we observe very good agreement with their calculation.

The primary source of uncertainty arises from the α7\alpha^7 term, which is estimated basing on the mass scaling of the hydrogenic A72A_{72} coefficient. A calculation of the full α7\alpha^7 correction for the helium atom is quite challenging and has not been attempted so far. Another open challenge is the inelastic second-order HFS contribution. However, given that the leading (electric dipole) polarizability correction is only 0.2 kHz, it can be argued to be negligible at the current level of precision.

The theory for the point nucleus summarized in Table III needs to be complemented by the description of the energy shift arising due to the nuclear mean square charge radius rCr_C. Generalizing the expression for the finite nuclear size (fs) effect for the hydrogenic energy levels [21] to the helium case, we obtain

Efs[AHe]=23Zα4mrC2χ2∑a⟨πδ3(ra)⟩M×{1−(Zα)2[ln⁡(ZαrCχ)+β]+α(Zα)(4ln⁡2−5)}≡CArC2,(82)E_{\mathrm{fs}}[{}^{A}\mathrm{He}]=\frac{2}{3}Z\alpha^4m\frac{r_C^2}{\chi^2}\sum_a\langle\pi\delta^3(r_a)\rangle_M \times\left\{1-(Z\alpha)^2\left[\ln\left(Z\alpha\frac{r_C}{\chi}\right)+\beta\right]+\alpha(Z\alpha)(4\ln2-5)\right\}\equiv C_A r_C^2, \tag*{(82)}

where χ=386.159\chi=386.159 fm is the reduced Compton wavelength, the expectation value of the δ\delta function is calculated for the case of the finite-mass nucleus, and the nonlogarithmic relativistic correction β=−0.41\beta=-0.41 is obtained from Ref. [21], assuming the 1S^{1}S state and the dipole charge distribution model. The last term in the brackets of Eq. (82) is the radiative fs correction [28].

It should be pointed out that, because of the nuclear mass dependence, the fs effect to the isotope shift is not exactly proportional to the difference of the squared radii (as was assumed in Ref. [17]) but has a small contribution.

beyond this. We therefore write the fs contribution to the 3He{}^{3}\mathrm{He}–4He{}^{4}\mathrm{He} isotope shift as

Efs[3He−4He]=C3r32−C4r42=C3+C42[r32−r42]+C3−C42[r32+r42]≡C[r32−r42]+D[r32+r42],(83)\begin{aligned} E_{\mathrm{fs}}[{}^{3}\mathrm{He}-{}^{4}\mathrm{He}] = C_{3}r_{3}^{2} - C_{4}r_{4}^{2} \\ &= \frac{C_{3}+C_{4}}{2}\left[r_{3}^{2}-r_{4}^{2}\right] + \frac{C_{3}-C_{4}}{2}\left[r_{3}^{2}+r_{4}^{2}\right] \\ &\equiv C\left[r_{3}^{2}-r_{4}^{2}\right] + D\left[r_{3}^{2}+r_{4}^{2}\right], \tag*{(83)} \end{aligned}

where rA≡rC[AHe]r_{A}\equiv r_{C}[{}^{A}\mathrm{He}] and the last line is the definition of the coefficients CC and DD. Their numerical values are given in Table IV.

E(3He, 21SF=1/2−23SF=3/2)E(^{3}\mathrm{He},\,^{2}{}^{1}S^{F=1/2}-{}^{2}{}^{3}S^{F=3/2})192 504 914 418.96(17)Exp. [15]
−E(4He, 21S−23S)-E(^{4}\mathrm{He},\,^{2}{}^{1}S-{}^{2}{}^{3}S)−192-192 510 702 148.72(20)Exp. [30]
δEhfs(23S3/2)\delta E_{\mathrm{hfs}}(^{2}{}^{3}S^{3/2})−2-2 246 567.059(5)Exp. [31, 32]
−δEiso(21S−23S)-\delta E_{\mathrm{iso}}({}^{2}{}^{1}S-{}^{2}{}^{3}S) (point nucleus)8 034 067.607 (136)Theory, Table 3
Sum−229.21(14)-229.21(14)
CC−214.758 kHz/fm2-214.758\ \mathrm{kHz/fm}^{2}this work
DD0.017 kHz/fm20.017\ \mathrm{kHz/fm}^{2}this work
δr2=r2(3He)−r2(4He)\delta r^{2}=r^{2}(^{3}\mathrm{He})-r^{2}(^{4}\mathrm{He})1.067 8(7) fm21.067\,8(7)\ \mathrm{fm}^{2}this work
1.069 3(15) fm21.069\,3(15)\ \mathrm{fm}^{2}Ref. [17]
1.075 7(15) fm21.075\,7(15)\ \mathrm{fm}^{2}previous [22, 33]
1.063 6(6)exp(30)theo fm21.063\,6(6)_{\mathrm{exp}}(30)_{\mathrm{theo}}\ \mathrm{fm}^{2}μ3,4He+\mu^{3,4}\mathrm{He}^{+} Lamb shift [14]

Table IV. Determination of the nuclear charge difference δr2\delta r^{2} from the measurement of 3He−4He{}^{3}\mathrm{He}-{}^{4}\mathrm{He} isotope shift in the 21S−23S2{}^{1}S-2{}^{3}S transition, in kHz unless specified otherwise. Physical constants are from Ref. [28].

Contribution(m/M)1(m/M)^{1}(m/M)2(m/M)^{2}(m/M)3(m/M)^{3}SumQi et al.
α2\alpha^{2}−8 026 758.512-8\,026\,758.512−4 958.331-4\,958.3315.0705.070−8 031 711.773-8\,031\,711.773−8 031 711.78-8\,031\,711.78
α4\alpha^{4}−2 496.229-2\,496.2292.0762.076—−2 494.153-2\,494.153−2 494.16-2\,494.16
α5\alpha^{5}56.60556.6050.000(47)0.000(47)—56.605(47)56.605(47)56.5956.59
α6\alpha^{6}2.732a2.732^{a}——2.7322.7322.73b2.73^{b}
α6Emix\alpha^{6}E_{\mathrm{mix}}—80.76580.7650.000(70)0.000(70)80.765(70)80.765(70)80.7680.76
α6δEmix\alpha^{6}\delta E_{\mathrm{mix}}—−1.770-1.770—−1.770-1.770−1.37-1.37
α7\alpha^{7}−0.210(105)-0.210(105)——−0.210(105)-0.210(105)−0.21(11)b-0.21(11)^{b}
α5Epol\alpha^{5}E_{\mathrm{pol}}0.198(20)0.198(20)——0.198(20)0.198(20)0.20(2)b0.20(2)^{b}
Total−8 034 067.607(136)-8\,034\,067.607(136)−8 034 067.24(19)-8\,034\,067.24(19)

Table 3. TABLE III. Breakdown of theoretical contributions to the 3He{}^{3}\mathrm{He}–4He{}^{4}\mathrm{He} isotope shift of the 21S2^{1}S–23S2^{3}S centroid transition frequencies, for the point nucleus, in kHz. EmixE_{\mathrm{mix}} is the mixing contribution induced by S0S_{0}. δEmix\delta E_{\mathrm{mix}} comes from δ(21S)\delta(2^{1}S) and δ(23S)\delta(2^{3}S) terms in Eqs. (77) and (80). EpolE_{\mathrm{pol}} is the leading nuclear polarizability contribution at the order α5\alpha^{5} (two-photon exchange), see Refs. [26, 27]. Physical constants are from Ref. [28].

We are now in the position to determine the mean square charge radius difference of the helium isotopes, δr2=r2[3He]−r2[4He]\delta r^{2}=r^{2}[{}^{3}\mathrm{He}]-r^{2}[{}^{4}\mathrm{He}], by comparing the theory prediction with the measured transition frequencies. Table IV summarizes all experimental and theoretical input required for the determination. Our result for the mean square charge radius difference of δr2=1.0678(7) fm2\delta r^{2}=1.0678(7)\,\mathrm{fm}^{2} agrees within 1.3 σ1.3\,\sigma with the muonic value of 1.0636(31) fm21.0636(31)\,\mathrm{fm}^{2}[14].

Summary

We have derived the complete elastic second-order hyperfine-interaction correction of order α6\alpha^{6} to the centroid energy levels of hydrogen-like and helium-like ions. The resulting formulas have been verified through comparisons with our all-order numerical calculations for hydrogen-like ions. Numerical calculations have been performed for the 21S2^{1}S and 23S2^{3}S states of atomic helium, extending the previous approximate treatment of this effect by Qi et al. [17].

The obtained results impact the determination of the nuclear charge radii difference between 3He{}^{3}\mathrm{He} and 4He{}^{4}\mathrm{He}, largely resolving the previously reported disagreement between the muonic and electronic helium determinations [14, 15]. From the measured isotope shift of the 21S2^{1}S–23S2^{3}S transition energy we obtain the mean square charge radius difference of 3He{}^{3}\mathrm{He} and 4He{}^{4}\mathrm{He}, which is only 1.3 σ1.3\,\sigma away from the muonic value. Given the difficulty of estimating the nuclear polarizability uncertainties in muonic atoms, we consider this to be good agreement.

Remarkably, the determination of δr2\delta r^{2} from the electronic helium spectroscopy is four times more accurate than that from the muonic helium. This is the consequence of the fact that the inelastic nuclear effects for muonic atoms are much more significant than in the electronic ones. The observed agreement between electronic and muonic measurements supports the reliability of nuclear charge radius determinations from muonic atoms, suggesting that they could be successfully extended to heavier nuclei, which is the objective of the QUARTET collaboration [29].

Acknowledgments

We gratefully acknowledge interesting discussions with F. Hagelstein, A. Antognini, R. Pohl, K. Eikema, and F. Merkt.

Dimensional regularization

This appendix is based on Ref. [22]. In order to extend spin-12\frac{1}{2} into dd dimensions, we define antisymmetric tensor

σij=i2[γi,γj],(84)\sigma^{ij}=\frac{i}{2}[\gamma^{i},\gamma^{j}], \tag*{(84)}

which in the three-dimensional limit simplifies to

σij→d→32 ϵijksk.(85)\sigma^{ij}\xrightarrow{d\to3}2\,\epsilon^{ijk}s^{k}. \tag*{(85)}

Since we consider only corrections to energy and neglect the hyperfine splitting, we may perform the angular average

σaijσbkl→σaijσbijδikδjl−δilδjkd(d−1)(86)\sigma_{a}^{ij}\sigma_{b}^{kl}\to\sigma_{a}^{ij}\sigma_{b}^{ij}\frac{\delta^{ik}\delta^{jl}-\delta^{il}\delta^{jk}}{d(d-1)} \tag*{(86)}
σaijσbil→σaijσbijδjld(87)\sigma_{a}^{ij}\sigma_{b}^{il}\to\sigma_{a}^{ij}\sigma_{b}^{ij}\frac{\delta^{jl}}{d} \tag*{(87)}

For a=ba=b,

σaijσaij=d(d−1).(88)\sigma_{a}^{ij}\sigma_{a}^{ij}=d(d-1). \tag*{(88)}

We will not always have spin 1/2 for the nucleus. In the dimensional regularization, we will pretend that I=1/2I=1/2 and use σNij\sigma_{N}^{ij}. However, instead of the above identity, we will consistently use the notation

σNijσNij=8⟨I2⟩ε,(89)\sigma_{N}^{ij}\sigma_{N}^{ij}=8\langle\mathbf{I}^{2}\rangle_{\varepsilon}, \tag*{(89)}

and drop the subscript ε\varepsilon once all 1/ε1/\varepsilon terms are canceled out.

Throughout our calculations, we extensively used the following result for the general dd-dimensional integral,

Ωd=2πd/2Γ(d/2)(90)\Omega_{d}=\frac{2\pi^{d/2}}{\Gamma(d/2)} \tag*{(90)}
∫ddp(2π)d4πpneip⋅r=22−nπ1−d/2Γ(d−n2)Γ(n2)rn−d.(91)\int\frac{d^{d}p}{(2\pi)^{d}}\frac{4\pi}{p^{n}}e^{i\mathbf{p}\cdot\mathbf{r}}=2^{2-n}\pi^{1-d/2}\frac{\Gamma\left(\frac{d-n}{2}\right)}{\Gamma\left(\frac{n}{2}\right)}r^{n-d}. \tag*{(91)}

Two special cases are of particular importance (with d=3−2εd=3-2\varepsilon):

V(r)=∫ddp(2π)d4πp2eip⋅r=C1r1−2ε(92)\mathcal{V}(r)=\int\frac{d^{d}p}{(2\pi)^{d}}\frac{4\pi}{p^{2}}e^{i\mathbf{p}\cdot\mathbf{r}}=\frac{C_{1}}{r^{1-2\varepsilon}} \tag*{(92)}
V(2)(r)=∫ddp(2π)d4πp4eip⋅r=C2r1+2ε(93)\mathcal{V}^{(2)}(r)=\int\frac{d^{d}p}{(2\pi)^{d}}\frac{4\pi}{p^{4}}e^{i\mathbf{p}\cdot\mathbf{r}}=C_{2}r^{1+2\varepsilon} \tag*{(93)}

where

C1=14πε−1/2Γ(1/2−ε),(94)C_{1}=\frac{1}{4}\pi^{\varepsilon-1/2}\Gamma(1/2-\varepsilon), \tag*{(94)}
C2=14πε−1/2Γ(−1/2−ε).(95)C_{2}=\frac{1}{4}\pi^{\varepsilon-1/2}\Gamma(-1/2-\varepsilon). \tag*{(95)}

We define also the associated potentials:

Vρ=4π∫ddp(2π)dρ~(p2)p2eip⋅r,(96)\mathcal{V}_{\rho}=4\pi\int\frac{d^{d}p}{(2\pi)^{d}}\frac{\widetilde{\rho}(p^{2})}{p^{2}}e^{i\mathbf{p}\cdot\mathbf{r}}, \tag*{(96)}
Vρ(2)=4π∫ddp(2π)dρ~(p2)p4eip⋅r,(97)\mathcal{V}_{\rho}^{(2)}=4\pi\int\frac{d^{d}p}{(2\pi)^{d}}\frac{\widetilde{\rho}(p^{2})}{p^{4}}e^{i\mathbf{p}\cdot\mathbf{r}}, \tag*{(97)}

and in d=3d=3

Vρ=4π∫d3p(2π)3ρ~(p2)p2eip⋅r,(98)\mathcal{V}_{\rho}=4\pi\int\frac{d^{3}p}{(2\pi)^{3}}\frac{\widetilde{\rho}(p^{2})}{p^{2}}e^{i\mathbf{p}\cdot\mathbf{r}}, \tag*{(98)}
Vρ(2)=4π∫d3p(2π)3ρ~(p2)p4eip⋅r−1r.(99)\mathcal{V}_{\rho}^{(2)}=4\pi\int\frac{d^{3}p}{(2\pi)^{3}}\frac{\widetilde{\rho}(p^{2})}{p^{4}}e^{i\mathbf{p}\cdot\mathbf{r}}-\frac{1}{r}. \tag*{(99)}

Asymptotic forms of these potentials at r→∞r\to\infty are:

Vρ→V+local terms,(100)\mathcal{V}_{\rho}\to\mathcal{V}+\text{local terms}, \tag*{(100)}
Vρ(2)→V(2)+ρ~′(0)V+local terms,(101)\mathcal{V}_{\rho}^{(2)}\to\mathcal{V}^{(2)}+\widetilde{\rho}'(0)\mathcal{V}+\text{local terms}, \tag*{(101)}

and in d=3d=3

Vρ→1r+local terms,(102)\mathcal{V}_{\rho}\to\frac{1}{r}+\text{local terms}, \tag*{(102)}
Vρ(2)→−r2+ρ~′(0)r+local terms.(103)\mathcal{V}_{\rho}^{(2)}\to-\frac{r}{2}+\frac{\widetilde{\rho}'(0)}{r}+\text{local terms}. \tag*{(103)}

References

  1. [1]G. W. F. Drake, editor, Handbook of Atomic, Molecular, and Optical Physics, Springer, Berlin, 2005.
  2. [2]K. Pachucki, Finite nuclear mass correction to the hyper- fine splitting in hydrogenic systems, Phys. Rev. A 109, 052822 (2024).arxiv.org/abs/2403.11663
  3. [3]D. Ruth, K. Slifer, J.-P. Chen, C. E. Carlson, F. Hagel- stein, V. Pascalutsa, A. Deur, S. Kuhn, M. Ripani, X. Zheng et al., New spin structure constraints on hyperfine splitting and proton Zemach radius, Phys. Lett. B 859, 139116 (2024).
  4. [4]F. Hagelstein, V. Lensky, and V. Pascalutsa, Chiral perturbation theory of the hyperfine splitting in (muonic) hydrogen, Eur. Phys. J. C 83, 762 (2023).arxiv.org/abs/2305.09633
  5. [5]L. Essen, R. Donaldson, E. Hope, and M. Bangham, Hydrogen Maser Work at the National Physical Laboratory, Metrologia 9, 128 (1973).
  6. [6]P. Strasser, S. Fukumura, R. Iwai, S. Kanda, S. Kawamura, M. Kitaguchi, S. Nishimura, S. Seo, H. M. Shimizu, K. Shimomura et al., Improved Measurements of Muonic Helium Ground-State Hyperfine Structure at a Near-Zero Magnetic Field, Phys. Rev. Lett. 131, 253003 (2023).DOI
  7. [7]J. Nuber, A. Adamczak, M. Abdou Ahmed, L. Affolter, F. D. Amaro, P. Amaro, A. Antognini, P. Carvalho, Y.-H. Chang, T.-L. Chen et al., Diffusion of muonic hydrogen in hydrogen gas and the measurement of the 1s hyperfine splitting of muonic hydrogen, SciPost Phys. Core 6, 057 (2023).arxiv.org/abs/2211.08297
  8. [8]A. Antognini, F. Hagelstein, and V. Pascalutsa, The Proton Structure in and out of Muonic Hydrogen, Ann. Rev. Nucl. Part. Sci. 72, 389 (2022).DOI
  9. [9]A. Vacchi, E. Mochiutti, A. Adamczak, D. Bakalov, G. Baldazzi, M. Baruzzo, R. Benocci, R. Bertoni, M. Bonesini, H. Cabrera et al., Investigating the Proton Structure: The FAMU Experiment, Nuclear Physics News 33, 9 (2023).DOI
  10. [10]K. Pachucki and V. A. Yerokhin, Heavy particle quantum electrodynamics, Phys. Rev. A 110, 032804 (2024).arxiv.org/abs/2407.20703
  11. [11]M.M. Sternheim, Correction to the (He3, 2^3S1) to (He3+, 2^2S1/2) Hyperfine-Structure Ratio Phys. Rev. Lett. 15, 232 (1965).
  12. [12]K. Pachucki, Hyperfine splitting of 2^3S1 state in He3, J. Phys. B 34, 3357 (2001).
  13. [13]K. Pachucki and V. A. Yerokhin, Theory of the Helium Isotope Shift, J. Phys. Chem. Ref. Data 44, 031206 (2015).arxiv.org/abs/1503.07727
  14. [14]K. Schuhmann, L. M. P. Fernandes, F. Nez, M. A. Ahmed, F. D. Amaro, P. Amaro, F. Biraben, T.-L. Chen, D. S. Covita, A. J. Dax et al., The helion charge radius from laser spectroscopy of muonic helium-3 ions, arXiv:2305.11679 (2023).arxiv.org/abs/2305.11679
  15. [15]Y. van der Werf, K. Steinbach, R. Jannin, H. Bethlem, and K. Eikema, The alpha and helion particle charge radius difference from spectroscopy of quantum-degenerate helium, arXiv:2306.02333 (2023).
  16. [16]K. Pachucki, V. Lensky, F. Hagelstein, S. S. Li Muli, S. Bacca, and R. Pohl, Comprehensive theory of the Lamb shift in light muonic atoms, Rev. Mod. Phys. 96, 015001 (2024).DOI
  17. [17]X.-Q. Qi, P.-P. Zhang, Z.-C. Yan, L.-Y. Tang, A.-X. Chen, T.-Y. Shi, and Z.-X. Zhong, Revised 3He nuclear charge radius due to electronic hyperfine mixing, arXiv:2409.09279 (2024).DOI
  18. [18]W. R. Johnson, S. A. Blundell, and J. Sapirstein, Finite basis sets for the Dirac equation constructed from B splines, Phys. Rev. A 37, 307 (1988).pubmed.ncbi.nlm.nih.gov/9899658
  19. [19]V. M. Shabaev, I. I. Tupitsyn, V. A. Yerokhin, G. Plunien, and G. Soff, Dual Kinetic Balance Approach to Basis-Set Expansions for the Dirac Equation, Phys. Rev. Lett. 93, 130405 (2004).pubmed.ncbi.nlm.nih.gov/15524687
  20. [20]N. F. Ramsey, Electron Coupled Interactions between Nuclear Spins in Molecules, Phys. Rev. 91, 303 (1953).DOI
  21. [21]K. Pachucki, V. Patkós, and V. A. Yerokhin, Three-photon-exchange nuclear structure correction in hydrogenic systems, Phys. Rev. A 97, 062511 (2018).DOI
  22. [22]V. Patkós, V. A. Yerokhin, and K. Pachucki, Higher-order recoil corrections for singlet states of the helium atom, Phys. Rev. A 95, 012508 (2017).DOI
  23. [23]J. L. Friar and G. L. Payne, Nuclear polarizabilities and logarithmic sum rules, Phys. Rev. C 55, 2764 (1997).DOI
  24. [24]J. L. Friar and G. L. Payne, Higher-order nuclear-polarizability corrections in atomic hydrogen, Phys. Rev. C 56, 619 (1997).DOI
  25. [25]V. A. Yerokhin, V. Patkós, and K. Pachucki, Atomic Structure Calculations of Helium with Correlated Exponential Functions, Symmetry 13, 1246 (2021).DOI
  26. [26]K. Pachucki and A. M. Moro, Nuclear polarizability of helium isotopes in atomic transitions, Phys. Rev. A 75, 032521 (2007).DOI
  27. [27]S. S. L. Muli, T. R. Richardson, and S. Bacca, Revisiting the helium isotope-shift puzzle with improved uncertainties from nuclear structure corrections, arXiv:2401.13424 (2024).DOI
  28. [28]P. Mohr, D. Newell, B. Taylor, E. Tiesinga, CODATA recommended values of the fundamental physical constants: 2022, arXiv:2409.03787 (2024).arxiv.org/abs/2409.03787
  29. [29]B. Ohayon, A. Abeln, S. Bara, T. E. Cocolios, O. Eizenberg, A. Fleischmann, L. Gastaldo, C. Godinho, M. Heines, D. Hengstler et al., Towards Precision Muonic X-ray Measurements of Charge Radii of Light Nuclei, Physics 6, 206 (2024).DOI
  30. [30]R. Rengelink, Y. van der Werf, R. Notermans, R. Jannin, K. Eikema, M. Hoogerland, and W. Vassen, Precision spectroscopy of helium in a magic wavelength optical dipole trap, Nature Physics 14, 1132 (2018).DOI
  31. [31]H. A. Schluesser, E. N. Fortson, and H. G. Dehmelt, Hyperfine Structure of the Ground State of 3He+ by the Ion-Storage Exchange-Collision Technique, Phys. Rev. 187, 5 (1969), [(E) Phys. Rev. A 2, 1612 (1970)].
  32. [32]S. D. Rosner and F. M. Pipkin, Hyperfine Structure of the 2^3S1 State of He3, Phys. Rev. A 1, 571 (1970), (E) Phys. Rev. A 3, 521 (1971).
  33. [33]K. Pachucki, V. Patkós, and V. A. Yerokhin, Testing fundamental interactions on the helium atom, Phys. Rev. A 95, 062510 (2017).DOI

Paper details

vlnc:W_8whc7yuf
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

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