WorksPhysics and Astronomy
Investigating Roles of Triple Excitations for High-precision Determination of Clock Properties of Alkaline Earth Metal Singly Charged Ions
Abstract
High-accuracy calculations of electric dipole polarizabilities and quadrupole moments () of the clock states of the singly charged calcium (Ca), strontium (Sr) and barium (Ba) alkaline-earth ions are estimated by employing relativistic coupled-cluster (RCC) theory. It demonstrates importance of the triple excitations in the RCC method for precise determination of the above quantities. We also observe a different trend of correlations in the values than an earlier study with respect to orbitals from higher angular momenta. Reliability of the results is verified by comparing the calculated energies, magnetic dipole hyperfine structure constants, and lifetimes of the atomic states with the experimental values of the Ca, Sr and Ba ions. Nuclear quadrupole moments of these isotopes are also estimated by combining calculations with the measured electric quadrupole hyperfine structure constants, showing large deviations from the literature values.
Singly charged alkaline-earth metal ions are instrumental in investigating many fundamental physics. Transitions between the ground () and metastable ( states, with principal quantum number , in the heavier alkaline ions play key roles in the clock frequency measurements [1, 2, 3], probing atomic parity violation (APV) [4, 5, 6] and analyzing isotope shift (IS) data to extract nuclear charge radii [7, 8, 9] and mass of a possible new vector boson from the non-linear effects in the King’s plot [10], constructing qubits [11–13] etc. to name a few. Owing to laser-accessible energy level structures, many of these ions can be efficiently manipulated with lasers, allowing for long interrogation times and thereby enabling high-precision measurements. In fact, a recent proposal aims to measure the APV amplitude in Ba to 0.1% precision using an ion trap technique [14].
From the perspective of carrying out above studies, high-accuracy calculations score as much as the precision measurements. It, therefore, necessitates to underscore roles of higher-order electron correlation effects in different atomic properties for more reliable outcomes. The prominent systematic effects in the high-precision atomic experiments are Stark shifts, blackbody radiation shifts, quadrupole shifts, and Zeeman shifts. In this regard, independent experiments are required to set-up to measure these systematic effects to the intended precision. Sometimes managing these experiments could be extremely challenging, time consuming and expensive. In such situation, high-accuracy atomic calculations to estimate systematics can be the alternative. In view of this, precise knowledge of electric dipole polarizabilities () and quadrupole moments () of the clock states of the Ca, Sr and Ba ions are of immense interest.
Over the past decades, several experimental techniques like resonant excitation, Stark ionization and single ion trapping methods have been applied to measure transition probabilities, lifetimes, and hyperfine structure constants of the clock states of the Ca, Sr and Ba ions [15–33]. However, only a few measured measurements of and of the considered ions are available in the literature [34–38]. Again, poor agreement between these values with the calculations reported using different theories are being observed [36–41]. This obscures improving accuracy of the clock transitions of the aforementioned ions. The calculations of were performed using lower-order methods, while other values were estimated in the semi-empirical approaches by adopting a mixture of methods utilizing some of the dominantly contributing electric dipole (E1) matrix elements from the calculations along with the experimental energies [3] [42, 43, 44, 45]. On the other hand, the values were estimated using the relativistic coupled-cluster (RCC) method [39, 41] and multi-configuration Dirac Hartree-Fock (MCDHF) method [40] approximating at the singles and doubles excitation levels. It has also been noticed that the extracted nuclear quadrupole moments () of Ca, Sr and Ba from the hyperfine structure studies of these atoms or their ions substantially differ from each other [46–49], which needs to be addressed.
In this work, we intend to perform first-principle calculations of and of the clock states of Ca, Sr, and Ba by employing the recently developed singles, doubles and triples approximated RCC (RCCSDT) method. In order to validate the calculations, we present energies and magnetic dipole hyperfine structure constants () of the clock states using the RCCSDT method and compare them with the experimental values. We also extract the values of the Ca, Sr and Ba isotopes by combining the precisely measured electric quadrupole hyperfine structure constants () with their corresponding calculations of the above ions. We consider as many as 40, 39, 38, 37, 36, 35, and 34 Gaussian-type orbital (GTO) basis functions for constructing Dirac-Hartree-Fock (DHF) orbitals of the , , , , , , and symmetries, respectively. The GTO parameters are optimized such that energies of core orbitals agree reasonably well with the numerical values from GRASP [50]. Contributions from the orbitals belonging to orbital angular momentum (quoted as +Basis) are extracted using the third-order many-body perturbation (MP) theory (refer to Supplemental Materials [51] for details). To account electron correlation effects, we have considered the Dirac-Coulomb (DC) Hamiltonian () in the RCCSDT method, while contributions from the Breit interactions (given as +Breit) and lower-order vacuum polarization (VP) effects (given as +VP) are estimated at the singles and doubles approximation in the RCC (RCCSD) method [62, 63] to improve accuracy of the calculations. Below, we present results from the DHF, RCCSD and RCCSDT methods to demonstrate how electron correlation effects arising through the triple excitations play important roles in improving the results.
| Method | Ca | Sr | Ba | ||||||
| DHF | 91440.02 | 72618.65 | 72594.55 | 84041.96 | 67384.80 | 67241.97 | 75339.35 | 68139.75 | 67665.52 |
| RCCSD | 95906.31 | 82293.80 | 82212.33 | 89094.93 | 74404.68 | 74101.58 | 80883.80 | 75738.94 | 74907.46 |
| RCCSDT | 95681.86 | 81912.20 | 81836.51 | 88859.83 | 74239.34 | 73953.86 | 80482.23 | 75420.56 | 74629.95 |
| +Basis | 10.34 | 77.39 | 77.14 | 16.57 | 81.17 | 79.88 | 30.06 | 125.91 | 121.71 |
| +Breit | −7.33 | 37.67 | 52.85 | −10.25 | 41.74 | 58.52 | −10.76 | 51.55 | 74.77 |
| +VP | 0.92 | −0.38 | −0.38 | 3.32 | −0.90 | −0.88 | 7.02 | −2.19 | −2.07 |
| Final | 95685.79 | 82026.88 | 81999.12 | 88869.47 | 74391.48 | 74091.38 | 80508.55 | 75595.83 | 74824.36 |
| Expt. [64] | 95751.87 | 82101.68 | 82040.99 | 88965.18 | 74409.28 | 74128.94 | 80686.30 | 75812.448 | 75011.493 |
| (%) | 0.07 | 0.09 | 0.05 | 0.11 | 0.02 | 0.05 | 0.22 | 0.29 | 0.25 |
| DHF | 98.90 | 104.92 | 102.24 | 127.03 | 148.90 | 139.42 | 185.52 | 91.29 | 88.78 |
| RCCSD | 76.33 | 32.01 | 31.87 | 92.05 | 64.53 | 63.32 | 125.37 | 52.45 | 52.56 |
| +Basis | −0.02 | −0.14 | −0.14 | −0.05 | −0.28 | −0.27 | −0.12 | −0.19 | −0.18 |
| +Breit | 0.0 | −0.18 | −0.22 | −0.01 | −0.52 | −0.57 | 0.0 | −0.23 | −0.28 |
| +VP | −0.03 | −0.03 | −0.03 | −0.03 | −0.05 | −0.04 | −0.10 | −0.04 | −0.04 |
| Final | 76.3(5) | 31.7(5) | 31.5(5) | 92.0(7) | 63.7(8) | 62.4(8) | 125.2(9) | 52.0(8) | 52.1(8) |
| Theory | 76.1(5) | 32.0(3) | 31.8(3) | 88.29(1.0) | 61.43(52) | 62.87(75) | 124.26(1.0) | 48.81(46) | 50.67(58) |
| 73.0(1.5) | 28.5(1.0) | 29.5(10) | 91.3(9) | 62.0(5) | 124.15 | ||||
| Expt. | 123.88(5) [34] | ||||||||
| 125.5(10) [35] |
Table 1. Calculated energies (in cm) and values (in a.u.) of the clock states of Ca, Sr, and Ba at the DHF and RCC methods. The final values are compared with experimental values (Expt.) and other precise calculations. values in a.u. [42]; [45]; [3]; [44]; [43].
Our calculated results for the energies of various states of the alkaline-earth metal ions are presented in Table I. We find that in all the states across all the considered ions, correlation effects improve the results significantly over the DHF results and inclusion of triples contributions decrease the values by 0.2-0.5% from the RCCSD method compared to the overall correlation contributions, which are about 4-7% in the ground states and about 9-12% in the metastable states. +Basis and +Breit contributions are on the order of 0.1%, whereas the +VP corrections are generally found to be very small. Comparison of our final results with the experimental energies from the National Institute of Standards and Technology (NIST) database [64] show small percentage differences. The calculated values can be improved further after inclusion of correlation effects from the quadruple excitations and self-energy corrections, which we defer to our future work but the present level of accuracy is sufficient enough to estimate properties of our interest using the calculated wave functions to the intended accuracy.
In the presence of external DC electric field , the Stark shift due to the second-order correction to of the state is . Conventionally, the dependency of is taken care by expressing , where and are referred as scalar and tensor components of . Detailed computational procedures for evaluating and in linear response (LR) RCC method, are given in Refs. [65, 66] (also refer to Supplemental Materials [51]). In the bottom part of Table I, we report our results for . In these quantities, the +Breit corrections are found to be consistently larger than the +VP contributions, and also outweighs the +Basis contributions. We also compare our final values with the available measurements [34, 35] and other precise estimations [3] [42, 43, 44, 45] in the same table. In Table II, we present values obtained from the DHF and RCC methods which show similar behavior as for . In fact, we provide additional values from the MP methods based on the Rayleigh-Schrödinger (MP) and Brillouin-Wigner (MP) approaches, and random phase approximation (RPA) in the Supplemental Materials [51]. Due to unavailability of the experimental results, we compare our final values with other high-accuracy calculations from Refs. [3] [42, 45] and they seem to have good agreement overall. Apart from theoretical estimations in Refs. [3] [45], other calculations mentioned above make use of experimental energies and/or E1 matrix elements to offer the precise values. Very good agreement between our ab initio results with the semi-empirical and measured values demonstrate potential of our employed LR-RCC method. It is to be noted that contributions from the triples are included in the uncertainties listed in the table, but they are discussed later.
| Method | Ca | Ca | Ca | Ca | Sr | Sr | Sr | Sr | Ba | Ba | Ba | Ba |
| DHF | −66.39 | 1.712 | −90.68 | 2.451 | −93.83 | 2.469 | −118.32 | 3.559 | −45.50 | 2.733 | −58.63 | 3.994 |
| RCCSD | −17.05 | 1.288 | −23.99 | 1.846 | −35.72 | 2.038 | −48.02 | 2.948 | −22.45 | 2.278 | −30.14 | 3.353 |
| RCCSDT | − | 1.296 | − | 1.857 | − | 2.037 | − | 2.984 | − | 2.279 | − | 3.354 |
| +Basis | 0.10 | −0.003 | 0.14 | −0.004 | 0.19 | −0.005 | 0.26 | −0.007 | 0.12 | −0.006 | 0.16 | −0.009 |
| +Breit | 0.14 | −0.002 | 0.22 | −0.003 | 0.39 | −0.002 | 0.53 | −0.006 | 0.18 | −0.003 | 0.26 | −0.006 |
| +VP | 0.03 | 0.03 | 0.04 | 0.04 | 0.04 | 0.04 | 0.001 | |||||
| Final | −16.8(2) | 1.291(6) | −23.6(3) | 1.850(9) | −35.1(4) | 2.030(8) | −47.2(5) | 2.971(10) | −22.1(2) | 2.270(9) | −29.7(3) | 3.339(12) |
| Theory | −17.43(23) | 1.289(11) | −24.51(29) | 1.849(17) | −35.42(25) | 2.029(12) | −48.83(30) | 2.935(17) | −24.62(28) | 2.256(11) | −30.85(31) | 3.319(15) |
| Theory | −15.8(7) | 1.338 | −22.45(5) | 1.917 | 2.107 | −47.7(3) | 3.048 | 2.297 | 3.379 | |||
| Expt. | 1.83(1) [36] | 2.6(3) [37], 2.9730.026 | 0.033[38] |
Table II. TABLE II. and values (both are in a.u.) of the metastable states of Ca, Sr, and Ba from different contributions of the present calculation and their comparisons with the available experimental values (Expt.) and accurate calculations.
In Table II, we report electric quadrupole (E2) moments of the metastable clock states of the considered ions and compare them with the literature values [36–40]. Though our calculations and measured values are close, yet they are outside the quoted error bars. Compared to other calculations, our results are more precise and do not sometime agree with others due to inclusion of triples contributions. More importantly, analysis of MP, RPA and +Basis results, in the Supplemental Materials [51], suggests that the calculations reported using the B-spline basis functions in Ref. [39] and GTOs in the present case show very different trends of contributions from orbitals belonging to different -quantum number. For example, Ref. [39] reported a large contribution from basis extrapolation with , amounting to nearly 0.9% in some cases. By contrast, we find that the basis extrapolation effects are generally smaller than our quoted uncertainties. It may be attributed to the way continuum are generated in both the works.
Inclusion of triples effects in the evaluation of and is strenuous in the LR-RCC method and it requires to determine amplitudes for both the unperturbed and perturbed wave operators. being a vector operator, the perturbed RCC operators generates unusually large number of amplitudes, especially for the metastable states, compared to their unperturbed counterparts at the given set of active orbitals. This is why we only estimated triples contributions to the ground state of
. However to gauge their influences to the values of the states, we also estimated them in the lighter Ca ion. From Table III, we find that triples contributions to are smaller than 0.5% in the ground states and 1.5% in the states. To understand the reason for small contributions arising from triples to the ground state values, we analyzed changes in the excitation energies and E1 matrix elements. We find that when moving from RCCSD to RCCSDT, the values of the E1 matrix elements decrease, from 2.912 atomic units (a.u.) and 4.119 a.u. to 2.903 a.u. and 4.109 a.u., and at the same time energies also decrease, from 25419.02 cm and 25509.06 cm to 25145.23 cm and 25374.59 cm, for the and transitions, respectively, in Ca. Since polarizability depends on the square of the E1 matrix elements and inversely on the transition energies, similar correlation trends in both the properties largely compensate to produce almost similar values in the RCCSD and RCCSDT methods.
| Quantity | Ca | Ca | Ca | Sr | Ba |
| −0.25 | 0.42 | 0.40 | −0.47 | −0.08 | |
| 0.0 | −0.31 | −0.30 | 0.0 | 0.0 |
Table III. TABLE III. Demonstration of triples contributions to the and values (in a.u.) of the ground states of the three considered ions and metastable states of Ca.
We now intend to fathom the observed inconsistencies in the lifetime studies, both from theoretical and experimental, of the metastable states of the considered clock candidates that are mainly governed by the forbidden magnetic dipole (M1) and E2 transition amplitudes. By evaluating the M1 () and E2 () transition rates (in s) for the transition, its lifetime is determined as , where the sum for runs over all possible decay modes (E2 and M1) and sum for runs for all possible decay states. We present our calculated E2 and M1 matrix elements from the DHF, RCCSD and RCCSDT methods along
| Method | Ca | Ca | Ca | Sr | Sr | Sr | Ba | Ba | Ba |
| DHF | −588.26 | −33.21 | −14.15 | −734.39 | −31.14 | −12.98 | 2935.35 | 128.41 | 51.5 |
| RCCSD | −810.77 | −47.55 | −4.32 | −1009.14 | −46.15 | 0.99 | 4094.24 | 189.74 | −5.99 |
| RCCSDT | −805.41 | −47.60 | −3.25 | −997.32 | −46.26 | 2.64 | 4026.31 | 191.21 | −14.06 |
| +Basis | −0.34 | −0.06 | −0.03 | −0.83 | −0.10 | −0.05 | 8.14 | 0.65 | 0.29 |
| +Breit | −0.72 | −0.12 | −0.04 | −1.55 | −0.22 | −0.06 | 6.70 | 1.19 | 0.36 |
| +VP | −0.38 | ∼ 0.0 | 0.01 | −1.12 | −0.01 | 0.02 | 8.67 | 0.11 | −0.14 |
| +BW | 0.88 | 0.02 | −0.01 | 3.15 | 0.05 | −0.05 | −36.46 | −0.58 | 0.60 |
| Final | −806(3) | −47.8(3) | −3.3(4) | −998(4) | −46.5(2) | 2.5(2) | 4013(9) | 193(4) | −12.9(9) |
| Theory [46] | −806.4(2.5) | −47.3(3) | −3.6(3) | −1001.203 | −45.604 | 2.145 | 190.89 | −11.99 | |
| Theory [69] | −808.126 | −45.294 | −4.008 | −990.638 | −44.320 | 2.168 | 4021.721 | 185.013 | −12.593 |
| Theory [40] | −47.27 | −4.84 | −45.60 | −2.77 | 192.99 | −9.39 | |||
| Expt. | −806.40207 | −48.3(1.6) | −3.8(6) | −1000.5(1.0) | 2.1743(14) | 4018.2 | 189.7296(7) | −12.028(11) | |
| −805(2) | 47.3(2) | −3.8931(2) | −1000.47367 | ||||||
| DHF | 54.45 | 77.15 | 80.35 | 110.04 | 133.57 | 171.09 | |||
| RCCSD | 68.63 | 95.90 | 116.48 | 160.16 | 194.51 | 256.34 | |||
| RCCSDT | 69.38 | 95.83 | 115.47 | 157.87 | 191.61 | 251.23 | |||
| +Basis | 0.10 | 0.13 | 0.26 | 0.35 | 0.53 | 0.66 | |||
| +Breit | 0.01 | 0.10 | −0.06 | 0.10 | −0.30 | −0.07 | |||
| +VP | −0.01 | ∼ 0.0 | −0.01 | ∼ 0.0 | −0.02 | −0.01 | |||
| Final | 69.5(8) | 96.1(9) | 115.7(8) | 158.3(10) | 191.8(20) | 251.8(25) | |||
| Expt. | −3.7(1.9) | −3.9(6.0) | −4.241(4) | 49.11(6) | 44.5408(17) | 59.533(43) | |||
| This work | −0.0441(4) | 0.3102(19) | 0.2344(17) | ||||||
| Other work | −0.0444(6) [46] | 0.305(2) [46] | 0.246(1) [46] | ||||||
| −0.0479(6) [47] | 0.323(20) [48] | 0.245(4) [49] | |||||||
| −0.0408(8) [49] | 0.335(20) [49] |
Table VI. Calculated and values of the ground and metastable states of , , and using the DHF and many-body methods, and their comparisons with the literature values. The extracted values of the above isotopes from precise measurements of are also given.
Ref: [24], [25]; [26], [27], [28], [29], [30], [31], [32], [33]
with other corrections in Table IV. These values from the MP methods and RPA can be found in the Supplemental Materials [51]. Comparing our final results with other calculations [67, 68], we find the E2 and M1 values are improved substantially due to inclusion of contributions from the triple excitations. Using the above discussed E2 and M1 matrix elements and wavelength values from the experiments [64], we present lifetimes for the metastable and states of the considered ions in Table V. For completeness, we give the calculated transition probabilities from both the metastable states of each ion due to the E2 and M1 decay channels in the above table. Our analysis shows that, for all ions, the lifetimes are governed primarily by the E2 channel with an exception arises in the case of the state of Ba, where the M1 channel provides a significant additional contribution. It can be observed from the above table that our values agree well with some of the calculations and measurements [15, 17–22, 67], yet differ significantly from others [16, 23, 68]. Since we use the E2 and M1 matrix elements that accounts more physical effects through the triples, +Breit and +VP compared to the earlier calculations, our estimated values are considered to be very reliable.
| Method | \multicolumn{3}{c}{Ca} | \multicolumn{3}{c}{Sr} | \multicolumn{3}{c}{Ba} | ||||||
| \multicolumn{10}{c}{E2 amplitudes} | |||||||||
| DHF | 9.767 | 11.978 | 5.018 | 12.969 | 15.973 | 7.261 | 14.764 | 18.385 | 8.092 |
| RCCSD | 7.914 | 9.716 | 3.777 | 11.142 | 13.762 | 6.005 | 12.662 | 15.850 | 6.777 |
| RCCSDT | 7.959 | 9.770 | 3.803 | 11.159 | 13.779 | 6.008 | 12.700 | 15.890 | 6.797 |
| +Basis | −0.011 | −0.013 | −0.008 | −0.017 | −0.020 | −0.014 | −0.029 | −0.036 | −0.019 |
| +Breit | −0.006 | −0.009 | −0.005 | −0.007 | −0.014 | −0.008 | −0.009 | −0.020 | −0.010 |
| +VP | |||||||||
| Final | 7.942(7) | 9.748(11) | 3.790(5) | 11.135(15) | 13.745(20) | 5.986(7) | 12.662(20) | 15.834(25) | 6.768(9) |
| Theory [68] | 7.78(4) | 9.56(4) | 3.68(2) | 11.01(4) | 13.60(5) | 5.90(2) | 12.49(11) | 15.65(14) | 6.65(6) |
| Theory [67] | 7.94(4) | 9.750(47) | 11.13(39) | 13.745(29) | 12.63(11) | 15.800(79) | |||
| \multicolumn{10}{c}{M1 amplitudes} | |||||||||
| DHF | −4.1[−7] | 1.549 | 1.6[−6] | 1.549 | 5.5[−6] | 1.549 | |||
| RCCSD | 3.5[−4] | 1.555 | 3.0[−5] | 1.553 | 3.1[−4] | 1.554 | |||
| RCCSDT | 3.5[−4] | 1.555 | 3.2[−5] | 1.553 | 3.2[−4] | 1.554 | |||
| +Basis | |||||||||
| +Breit | −1.1[−5] | 1.1[−5] | −0.002 | 1.0[−5] | |||||
| +VP | |||||||||
| Final | 3.4(2)[−4] | 1.555(2) | 4.3(3)[−5] | 1.551(2) | 3.3(3)[−4] | 1.554(2) | |||
| Theory [68] | 1.5491 | 1.5492 | 0.0002 | 1.5493 |
Table IV. Reduced E2 and M1 amplitudes (in a.u.) of the clock transitions in Ca, Sr, and Ba at different levels of calculations and their comparisons with previously available precise calculations. Numbers appearing as mean .
| Atomic State | Decay State | Mode | in s | (in s), Theory | (in s), Expt. |
| Ca ion | |||||
| M1 | 1.98[−6] | 1.195(4)* | 1.20(1) [15] | ||
| E2 | 0.83690 | 1.194(11) [67] | 1.176(11) [16] | ||
| 1.24(1) [68] | |||||
| M1 | 2.43[−6] | 1.164(5)* | 1.168(7) [15] | ||
| E2 | 2.21[−13] | 1.163(11) [67] | 1.168(9) [16] | ||
| E2 | 0.85936 | 1.21(1) [68] | |||
| Sr ion | |||||
| M1 | 3.84[−8] | 0.441(2)* | 0.435(4) [17, 18] | ||
| E2 | 2.26822 | 0.437(14) [67] | 0.455(29) [18] | ||
| 0.451(3) [68] | |||||
| M1 | 0.24[−3] | 0.394(2)* | 0.408(22) [18] | ||
| E2 | 1.16[−9] | 0.3945(22) [67] | 0.3908(16) [19] | ||
| E2 | 2.53467 | 0.403(3) [68] | 0.372(25) [22] | ||
| Ba ion | |||||
| M1 | 0.08[−6] | 81.0(5)* | 79.8(4.6) [20] | ||
| E2 | 0.01234 | 81.4(1.4) [67] | 89(16) [21] | ||
| 83(1) [68] | |||||
| M1 | 0.00558 | 30.2(2)* | 32.0(4.6) [21] | ||
| E2 | 2.82[−7] | 30.34(48) [67] | 34.5(3.5) [23] | ||
| E2 | 0.02754 | 30.81(65) [68] |
Table V. Estimated lifetimes of clock states of the considered ions computed using the experimental wavelengths (shown with *), their comparisons with the literature values. Transition rates from the M1 and E2 modes are also given.
In Table XI, we report our calculated magnetic dipole hyperfine structure constants () and ratio of and electric quadrupole hyperfine structure constants () of the , and ions from the DHF, RCCSD and RCCSDT methods including +Basis, +Breit, +VP and Bohr-Weisskopf effect (+BW) contributions. We find, our values are in good agreement with other calculations and available precise measurements [24–33, 40, 46, 69]. Details of correlation trends through MP and RPA results, and also the nuclear magnetic moments used for estimating by us are discussed in the Supplemental Materials [51]. From our values, we anticipate that the obtained are equally accurate. Combining our calculations with the precisely measured values from Refs. [28, 31, 33], we have extracted the values of , and . They are compared with the earlier extracted values [46–49] in the above table. We find differences of about 4–9% from previously reported values compared to ours. Since our calculations incorporate correlations and relativistic effects in a more rigorous manner, the inferred values in the present work are expected to be more accurate.
| Method | |||||||||
| DHF | |||||||||
| MP(2) | |||||||||
| MP(3) | |||||||||
| MP(2) | |||||||||
| MP(3) | |||||||||
| RPA | |||||||||
| RCCSD | |||||||||
| RCCSDT | |||||||||
| DHF | |||||||||
| MP(2) | |||||||||
| MP(3) | |||||||||
| MP(2) | |||||||||
| MP(3) | |||||||||
| RPA | |||||||||
| RCCSD | |||||||||
| RCCSDT |
Table XI. Calculated and values of the ground and metastable states of , , and using the DHF and many-body methods.
This work is supported by ANRF with grant no. CRG/2023/002558 and Department of Space, Government of India. All calculations were performed on the ParamVikram-1000 HPC cluster at the Physical Research Laboratory (PRL), Ahmedabad, Gujarat, India.
References
Supplemental Materials: Investigating Roles of Triple Excitations for High-precision Determination of Clock Properties of Alkaline Earth Metal Singly Charged Ions
For an atomic state , depends on and . Using the -dependent factors, it yields [1, 2]
Here and are called as the scalar and tensor polarizabilities, respectively. These independent quantities can be written in terms of reduced matrix elements as [1, 3]
and
where , with is the E1 operator. In the linear response (LR) approach, we express as
where and denote the unperturbed wave function of the atomic Hamiltonian () and first-order perturbed wave function due to the E1 interaction, respectively. The term refers to the effective dipole operators for scalar and tensor components, given by and .
To begin with, we consider at the Dirac-Coulomb (DC) approximation, given by (in a.u.)
Here, and are the Dirac matrices, is the single-particle momentum operator, represents the nuclear potential felt by an electron, and represents the Coulomb repulsion between two electrons. Since all the considered atomic states have closed-core and a valence orbital with different parity and angular momentum, we consider the potential formalism, where is the total number of electron of the considered singly charged ion, to produce the initial wave function, , using the DHF Hamiltonian, .
At the first step, we express the exact unperturbed wave function of the closed-core, , due to by
where is referred to as the wave operator, which accounts for the electron correlation effects arising from the residual interaction . In order to obtain the desired wave function of an intended state we append the required valence orbital, , to the closed-core configuration in the next step by defining the modified DHF wave function as . This follows the final unperturbed wave function in the wave operator formalism
where is responsible for accounting the correlation effects involving the electron from the valence orbital . Similarly, the corresponding first-order perturbed wave functions can be expressed as
and
where superscript (1) on wave operators stands for the first-order perturbation. In the DHF method contribution from is completely neglected; thus .
In order to understand roles correlation effects in different properties of our interest, we start calculations considering as perturbation in the second-order and third-order perturbation theory (denoted as MP(2) and MP(3), respectively), then include to all-orders through the random phase approximation (RPA) and relativistic coupled-cluster (RCC) theory. In the MP method, amplitudes of the wave operators are obtained using the Bloch equation [4]
and
where ‘l’ means that only the linked diagrams will contribute to the wave operator. It should be noted that energy of the state is given by
In the Rayleigh-Schrödinger MP approach (MP method), the wave function and energies are evaluated order-by-order explicitly [4, 5]. On the other hand, in the Brillouin-Wigner MP approach (MP method), amplitudes of the wave operators are obtained by rewriting Eq. (10) as [4, 5]
Similarly, the first-order perturbed wave operators in the MP method can be derived by extending the
and
In the RPA, are equivalent to the DHF method and the first-order perturbed wave operators are obtained by
and
where , and the RPA potentials are defined as
and
In the RCC method, we express the exact wave function of a one-valence atomic system (with valence orbital ) as , where is an wave operator that accounts for excitation amplitudes to all-orders of the residual interactions that are neglected at producing the mean-field wave function . In the expanded form, it yields with total number of electrons in the system and the subscript denotes level of excitation of electrons from core orbitals of to its virtuals; i.e. in the RCCSD method and in the RCCSDT method . We adopt the Fock-space approach to express , where is the excitation operator responsible for exciting only the core electrons and takes care of excitation of the valence electron coupling with the core-electrons. Due to presence of only one valence electron in the considered systems, the exact form of wave function yields . The amplitudes of the and operators are obtained by solving
where subscript means it retains only linked terms and denotes for core and valence excitation determinants. In the ab initio approach, the energy is calculated as . Since both amplitude and energy solving equations are interdependent, they are solved simultaneously. In the LR-RCC method, the and operators are expanded up to the zeroth- and first-order due to for a small value of ; i.e. and .
In the RCC theory ansatz, the unperturbed wave operators are defined as
and
Extending these definitions to the first-order perturbed wave functions, we can define the corresponding wave operators as
and
Following the second-order energy expression , we arrive at the LR-RCC expressions
and
where , .
In order to evaluate the M1 and E2 transition matrix elements, electric quadrupole moment (), magnetic dipole hyperfine structure constants () and electric quadrupole hyperfine structure constants (), we use the following general expression
where with the respective operator . In the evaluation of , and expressions, we set .
To account for the Bohr-Weisskopf (BW) effect [7], we define the nuclear magnetization function () in the Fermi nuclear charge distribution approximation as
for and
for , where
and
In Eq. (27) when Woods-Saxon potential correction is not taken into account, else it is estimated using following expressions after neglecting the spin-orbit interaction within the nucleus [8, 9]
for and
for .
Our calculated energies and values for the clock states of the alkaline-earth-metal ions, obtained using DHF and different many-body methods, are summarized in Table VII. To analyze correlation effects in the energies, we present contributions from the DC Hamiltonian at the DHF and MP(2) levels, together with results from the RCC methods. The data show that core-polarization (CP) effects, which first appear at the MP(2) level, contribute as much as 5–12% to the total energies. In contrast, the difference between the RCCSD and MP(2) results is only about 0.3–1.6%, indicating all-order CP and pair-correlation (PC) effects play a relatively minor role in energy calculation. To illustrate the role of correlation effects on , we include perturbative results from both the and perturbation schemes. It is evident that the approach provides values closer to the final results. This is expected, since unlike the scheme, the method accounts for correlation-corrected energies in the property evaluation, leading to improved accuracy. A comparison of DHF results with MP(2) and MP(2) shows that the lower-order CP effects generally reduce the values. However, the correlation trend from MP(2) to MP(3) differs significantly between the two perturbation schemes. For the method, the PC contribution, which appears at the MP(3) level, increases . In contrast, the method shows an increase at MP(3) only for the ground state; for the states, MP(3) values are smaller than MP(2), except in the case of Ba, where the difference is negligible. We also find that the RPA results remain much closer to the DHF values compared to those from MP(2). Since RPA includes CP contributions from MP(2) to all orders, this close agreement suggests that there are cancellations between the lowest-order and higher-order CP effects in determining . Finally, we compare these results with those obtained using the RCCSD method. The RCCSD approach accounts for CP and PC effects to all orders and incorporates correlation in the state energies, thereby providing more reliable values.
| Method | Ca 4s | Ca 3d | Ca 3d | Sr 5s | Sr 4d | Sr 4d | Ba 6s | Ba 5d | Ba 5d |
| Method | \multicolumn{9}{c}{Energies in cm} | ||||||||
| DHF | 91440.02 | 72618.65 | 72594.55 | 84041.96 | 67384.80 | 67241.97 | 75339.35 | 68139.75 | 67665.52 |
| MP(2) | 96176.40 | 82729.13 | 82650.14 | 89577.17 | 75061.22 | 74741.76 | 81794.94 | 76934.16 | 76014.21 |
| RCCSD | 95906.31 | 82293.80 | 82212.33 | 89094.93 | 74404.68 | 74101.58 | 80883.80 | 75738.94 | 74907.46 |
| RCCSDT | 95681.86 | 81912.20 | 81836.51 | 88859.83 | 74239.34 | 73953.86 | 80482.23 | 75420.56 | 74629.95 |
| Method | \multicolumn{9}{c}{ values in a.u.} | ||||||||
| DHF | 98.90 | 104.92 | 102.24 | 127.03 | 148.90 | 139.42 | 185.52 | 91.29 | 88.78 |
| MP(2) | 86.49 | 93.14 | 90.85 | 106.46 | 126.86 | 120.02 | 146.377 | 71.38 | 70.49 |
| MP(3) | 90.91 | 104.88 | 100.19 | 113.17 | 163.90 | 145.56 | 162.94 | 76.30 | 72.89 |
| MP(2) | 72.56 | 36.64 | 36.55 | 86.21 | 61.29 | 61.09 | 112.04 | 49.52 | 50.59 |
| MP(3) | 75.31 | 30.33 | 30.20 | 90.17 | 60.20 | 59.41 | 121.09 | 49.95 | 50.50 |
| RPA | 93.28 | 99.78 | 97.27 | 117.80 | 139.15 | 130.53 | 168.37 | 82.07 | 80.17 |
| RCCSD | 76.33 | 32.01 | 31.87 | 92.05 | 64.53 | 63.32 | 125.37 | 52.45 | 52.56 |
Table VII. Calculated energies (in cm) and values (in a.u.) of the clock states of Ca, Sr, and Ba using the DHF and various many-body methods.
In Table VIII, we present the values of and obtained from DHF and various many-body methods. The overall correlation behavior is generally consistent with that observed for . Owing to the inclusion of correlation effects in the energy values, the results from the scheme are more accurate than those from the perturbation scheme. Similar to the case of , the lowest-order CP contribution reduces the magnitude of the tensor polarizability. However, in the scheme, the value of increases when going from MP(2) to MP(3), whereas in the scheme, the inclusion of PC effects produces the opposite trend, further decreasing the magnitude. Similar to the case of , we find that the lowest-order CP effect reduces the values. However, due to cancellations between lower-order and higher-order CP contributions, the RPA results remain much closer to DHF values than to those from MP(2) or MP(2). Unlike , where the correlation trends in and perturbation theories differ, in the case of , both schemes show a consistent behavior; the inclusion of PC effects at the MP(3) level leads to a further reduction in the values.
| Method | Ca 3 | Ca 3 | Ca 3 | Ca 3 | Sr 4 | Sr 4 | Sr 4 | Sr 4 | Ba 5 | Ba 5 | Ba 5 | Ba 5 |
| DHF | −66.39 | 1.712 | −90.68 | 2.451 | −93.83 | 2.469 | −118.32 | 3.559 | −45.50 | 2.733 | −58.63 | 3.994 |
| MP(2) | −58.79 | 1.666 | −80.41 | 2.386 | −79.50 | 2.390 | −100.93 | 3.457 | −34.97 | 2.595 | −45.97 | 3.809 |
| MP(3) | −70.27 | 1.152 | −93.04 | 1.653 | −112.12 | 1.916 | −130.32 | 2.780 | −40.08 | 2.153 | −48.84 | 3.185 |
| MP(2) | −19.24 | 1.666 | −27.21 | 2.386 | −31.99 | 2.396 | −44.33 | 3.458 | −20.15 | 2.597 | −28.12 | 3.811 |
| MP(3) | −16.29 | 1.265 | −22.89 | 1.814 | −33.03 | 2.008 | −44.72 | 2.908 | −20.88 | 2.249 | −28.22 | 3.315 |
| RPA | −63.15 | 1.687 | −86.29 | 2.416 | −87.71 | 2.426 | −110.87 | 3.501 | −41.22 | 2.640 | −53.43 | 3.874 |
| RCCSD | −17.05 | 1.288 | −23.99 | 1.846 | −35.72 | 2.038 | −48.02 | 2.948 | −22.45 | 2.278 | −30.14 | 3.353 |
| RCCSDT | - | 1.296 | - | 1.857 | - | 2.037 | - | 2.984 | - | 2.279 | - | 3.354 |
Table VIII. and values (both are in a.u.) of the metastable states of Ca, Sr, and Ba using the DHF and different many-body methods.
In order to illustrate how the value of varies with the size of the basis functions, we present in Table IX the results obtained using the MP(3) and MP(3) methods. For this purpose, we have employed three different basis sets characterized by maximum angular quantum numbers , 6, and 9. As can be seen from Table IX, in the case of the scheme, the difference between the results obtained with and is about 2–3%, while the corresponding change between and reduces to only 0.4%. In contrast, for the scheme, the variations are even smaller; the difference between the MP(3) results with and still exceeds 1%, but the change between and is less than 0.3%. This trend indicates that the basis with already incorporates the dominant part of the correlation contributions. Taking into account both this observation and the practical limitations of our computational resources, we have therefore carried out the RCC calculations with . Nevertheless, in order to ensure that the contributions from higher angular momentum orbitals are not neglected, their contributions have been explicitly included in the manuscript as the +Basis correction, evaluated within the scheme, since this approach accounts for correlation effects rigorously in energies and matrix elements, in contrast to the scheme.
| Basis | Ca | Ca | Sr | Sr | Ba | Ba |
| 1.186 | 1.701 | 1.961 | 2.844 | 2.203 | 3.256 | |
| 1.152 | 1.653 | 1.916 | 2.780 | 2.153 | 3.185 | |
| 1.147 | 1.647 | 1.909 | 2.770 | 2.144 | 3.171 | |
| MP(3), | 1.272 | 1.845 | 2.040 | 2.954 | 2.284 | 3.365 |
| 1.265 | 1.814 | 2.008 | 2.908 | 2.249 | 3.315 | |
| 1.262 | 1.810 | 2.003 | 2.901 | 2.243 | 3.306 |
Table IX. Convergence of the MP(3) and MP(3) values for the estimated (in a.u.) of the metastable states of Ca, Sr, and Ba with different sizes of basis functions.
We present our calculated E2 and M1 matrix elements in Table X. For the E2 matrix elements, we observe a correlation trend similar to that found for the quadrupole moment , which is expected. In this case as well, the PC effects are far more significant than the CP contributions, bringing the MP(3) results much closer to our recommended RCC values. As shown in Table X, the M1 transition amplitudes are significantly smaller in magnitude compared to the corresponding E2 transitions in all the considered ions. Interestingly, the two cases we consider, namely the and transitions, display distinct correlation trends. For the transition, both CP and PC effects increase the M1 amplitude relative to the DHF value, often by one or even two orders of magnitude, with the PC contribution being more prominent than the CP effect. By contrast, in the transition, a comparison between MP(2) and MP(3) results reveals an opposite interplay; while CP effects tend to reduce the M1 amplitude, the PC contributions act in the opposite direction and increase it. Despite these competing trends, there is strong cancellation among the different correlation effects, which results in only small net changes in the M1 amplitudes across the many-body methods.
| Method | \multicolumn{3}{c}{Ca} | \multicolumn{3}{c}{Sr} | \multicolumn{3}{c}{Ba} | ||||||
| \multicolumn{10}{c}{E2 amplitudes} | |||||||||
| DHF | 9.767 | 11.978 | 5.018 | 12.969 | 15.973 | 7.261 | 14.764 | 18.385 | 8.092 |
| MP(2) | 9.608 | 11.784 | 4.884 | 12.700 | 15.651 | 7.052 | 14.300 | 17.844 | 7.712 |
| MP(3) | 7.449 | 9.150 | 3.380 | 10.648 | 13.172 | 5.654 | 11.939 | 15.011 | 6.423 |
| MP(2) | 9.609 | 11.785 | 4.885 | 12.701 | 15.653 | 7.053 | 14.304 | 17.848 | 7.717 |
| MP(3) | 7.884 | 9.679 | 3.712 | 11.001 | 13.591 | 5.921 | 12.416 | 15.563 | 6.698 |
| RPA | 9.712 | 11.912 | 4.945 | 12.854 | 15.841 | 7.142 | 14.539 | 18.142 | 7.849 |
| RCCSD | 7.914 | 9.716 | 3.777 | 11.142 | 13.762 | 6.005 | 12.662 | 15.850 | 6.777 |
| RCCSDT | 7.959 | 9.770 | 3.803 | 11.159 | 13.779 | 6.008 | 12.700 | 15.890 | 6.797 |
| \multicolumn{10}{c}{M1 amplitudes} | |||||||||
| DHF | 1.549 | 1.549 | 1.549 | ||||||
| MP(2) | 1.528 | 1.528 | 1.520 | ||||||
| MP(3) | 1.544 | 1.545 | 1.544 | ||||||
| MP(2) | 1.528 | 1.528 | 1.520 | ||||||
| MP(3) | 1.544 | 1.544 | 1.542 | ||||||
| RPA | 1.549 | 1.549 | 1.550 | ||||||
| RCCSD | 1.555 | 1.553 | 1.554 | ||||||
| RCCSDT | 1.555 | 1.553 | 1.554 |
Table X. Reduced E2 and M1 amplitudes (in a.u.) of the clock transitions in Ca, Sr, and Ba ions with various many-body methods. Numbers appearing as mean .
In Table XI, we present our results for the and values. For calculation, we have used for , for , and for [10]. Similar to the E2 and M1 amplitudes, the differences between the and results for and at a given perturbative level are found to be small. For the ground state, our analysis shows that both CP and PC effects contribute in the same direction, increasing the magnitude of value. However, for , value is found to be very sensitive to correlation effects. As can be seen from the MP(2) results in Table XI, the CP effects contribute up to 600-1500% in states. The states also exhibit a striking sensitivity to PC effects, with contributions reaching as high as 200%. Furthermore, we observe that triple excitations contribute only marginally to in the and states, but play a dominant role in the states, where they account for more than 50% of the total value. This unusually large contribution also hints at non-negligible contributions from higher-order excitations, such as quadrupole excitations. Interestingly, the correlation behavior in is quite different. In this case, the sensitivity to correlation effects is far less pronounced. Contributions from CP, PC, and even triple excitations remain relatively small across all states, including the levels. This contrast between and demonstrates the property-dependent nature of correlation and emphasizes the need for a case-by-case treatment when evaluating the hyperfine structure constants.
References
References
- [1]N. L. Manakov, V. D. Ovsiannikov, and L. P. Rapoport, Atoms in a laser field, Phys. Rep. 141, 320 (1986).DOI
- [2]J. E. Stalnaker, D. Budker, S. J. Freedman, J. S. Guzman, S. M. Rochester, and V. V. Yashchuk, Dynamic Stark effect and forbidden-transition spectral line shapes, Phys. Rev. A 73, 043416 (2006).DOI
- [3]M. G. Kozlov and S. G. Porsev, Polarizabilities and hyperfine structure constants of the low-lying levels of barium, Eur. Phys. J. D 5, 59 (1999).DOI
- [4]I. Lindgren and J. Morrison, Atomic many-body theory, Springer-Verlag Berlin Heidelberg (1986).
- [5]I. Shavitt, R. J. Bartlett, Many-body methods in chemistry and physics: MBPT and coupled-cluster theory, Cambridge University Press (2009).
- [6]A. Chakraborty and B. K. Sahoo, Deciphering core, valence, and double-core-polarization contributions to parity violating amplitudes in 133Cs using different many-body methods, J. Phys. Chem. A 127, 7518 (2023).DOI
- [7]A. Bohr and V.F. Weisskopf, The influence of nuclear structure on the hyperfine structure of heavy elements, Phys. Rev. 77, 94 (1950).DOI
- [8]V. M. Shabaev, M. Tomaselli, T. Kühl, A. N. Artemyev, and V. A. Yerokhin, Ground-state hyperfine splitting of high-Z hydrogenlike ions, Phys. Rev. A 56, 252 (1997).DOI
- [9]A. V. Volotka, D. A. Glazov, I. I. Tupitsyn, N. S. Oreshkina, G. Plunien, and V. M. Shabaev, Ground-state hyperfine structure of H-, Li-, and B-like ions in the intermediate-Z region, Phys. Rev. A 78, 062507 (2008).DOI
- [10]N. J. Stone, Table of nuclear magnetic dipole and electric quadrupole moments, At. Data Nucl. Data Tables 90, 75 (2005).