WorksPhysics and Astronomy
Fock-space relativistic coupled-cluster study of systematic shifts and fine-structure constant sensitivity in {\rm Cf}, {\rm Cm} and {\rm Bk} highly charged ions
Abstract
Highly charged ions (HCIs) are among the most promising candidates proposed for the development of next-generation optical clocks and the precision tests of fundamental physics. In this work, we investigate the clock-relevant properties of , and HCIs using Fock-space relativistic coupled-cluster (FSRCC) theory. The corrections from Breit interaction, QED effects and perturbative triples are incorporated to improve the accuracy of results further. We compute excitation energies, transition amplitudes, lifetimes, hyperfine structure constants, electric quadrupole moments, and electric dipole polarizabilities for low-lying states of these ions. These properties are further used to characterize important clock transitions and estimate key systematic frequency shifts. Our calculations reveal that the proposed clock transitions exhibit strongly suppressed systematic frequency shifts. We also compute the sensitivity coefficients, , and enhancement factors, , for the proposed clock transition as a probe to variation in the fine-structure constant, . Our calculations predict exceptionally large of and for transition in and transition in , respectively. These giant s place the proposed transitions among the most sensitive optical probes for the laboratory searches for variation. Further, a comparative analysis of proposed transitions in and shows that, transition possess smaller systematic shifts and is therefore attractive for precision frequency metrology, whereas transition exhibit significantly enhanced sensitivity to variation.
hci.pdf
Introduction—The extraordinary accuracy and stability of optical atomic clocks have enabled applications ranging from precision frequency metrology and relativistic geodesy [1] to the test of general relativity [3] [2] and the searches for physics beyond the Standard Model [4, 5]. Among the key applications, the search for possible temporal variation of fine-structure constant, , has attracted a considerable interest [6, 7, 8]. Any detectable variation in would provide evidence for physics beyond the Standard Model and can offer important insights into cosmology and the nature of dark matter. The sensitivity of a clock-comparison experiment is determined by the difference in the enhancement factors () of the two clock transitions. For example, comparisons between and optical clocks, for which , constrained the temporal variation of at the level of [9]. Further, recent measurements of the frequency ratio between E2 and E3 clock transitions in , characterized by , improved this limit to [10]. These remarkable achievements have motivated the search for atomic systems possessing substantially larger s.
Highly charged ions (HCIs) have emerged as promising candidates for probing [11, 12, 13]. Their compact electronic structure leads to reduced sensitivities to external perturbations. In addition, the large nuclear charge, strong relativistic effects, and high ionization potentials in HCIs enhance their sensitivity to variation. Among the proposed HCI based clocks, ions near the - level crossing are of particular interest [14]. One of the most prominent examples is Cf, which was identified by Berengut et al. [11] as a highly promising candidate for probing variation in . The optical transition in this ion combines a narrow linewidth with one of the largest enhancement factors predicted for any optical-clock transition. Moreover recently, Porsev et al. [15, 16] carried out high-precision coupled-cluster (CC) calculations and demonstrated that Cf possesses favorable clock characteristics together with extraordinary sensitivity to variation. The neighboring ions Cm and Bk have also been proposed as attractive clock candidates by Dzuba et al. [17]. The states , , and in these HCIs lie within the optical domain and are suitable for two potential clock transitions, and . The former one involves orbitals of markedly different relativistic character, whereas the latter corresponds to a fine-structure transition within the manifold. These contrasting electronic configurations in Cm and Bk make them interesting systems for investigating the interplay between clock accuracy and sensitivity to the variation in fundamental constants.
Despite their considerable promise, experimental information on the low-lying spectra and clock-relevant properties of these HCIs remains extremely limited. Consequently, the characterization of these systems presently relies almost entirely on theoretical predictions. However, the available theoretical studies are also limited. For Cf, only the pioneering work of Berengut et al. [11] and the subsequent CC studies of Porsev et al. [15, 16] are available, whereas for Cm and Bk the available data are largely restricted to the recent linearized CC calculations of Dzuba et al. [17]. Although these works established the exceptional potential of these ions, a comprehensive and uniform investigation of their clock-relevant properties is still lacking. In particular, while both and transitions were proposed as promising transitions in Cm and Bk, a quantitative comparison of their relative merits has not yet been carried out. Importantly, key properties such as transition amplitudes, lifetimes, hyperfine structure constants, electric quadrupole moments, dipole polarizabilities, and systematic frequency shifts have not been investigated consistently within a common high-precision many-body framework which systematically incorporates correlation and relativistic effects to the same level of accuracy. Moreover, the calculations reported for Cm and Bk employ a linearized CC treatment, missing the important correlation effects from nonlinear terms.
In this work, we perform a systematic Fock-space relativistic coupled-cluster (FSRCC) [18–20] investigation of the spectroscopic properties of low-lying states of Cf, Cm, and Bk HCIs. Electron-correlation effects are treated to all orders within the CC framework, while the corrections from the Breit interaction, QED effects and perturbative triples are incorporated to improve the accuracy of calculations further. We calculate the aforementioned clock’s relevant spectroscopic properties. Further, we estimate the dominant systematic shifts in clock transition frequencies, including blackbody-radiation (BBR), electric-quadrupole (EQ), quadratic Zeeman, and micromotion-induced (MI). In addition, relativistic sensitivity coefficients and enhancement factors are determined as a probe to variation. Our work also enables a direct comparison of the competing clock transitions in Cm and Bk, providing a quantitative assessment of their suitability for high-accuracy clocks and as a probe for variation.
Basis Set, Excitation Energy, Lifetime and Hyperfine Constants—We have used Gaussian type orbitals (GTOs) as basis function for FSRCC calculations [21]. The self-energy (SE) correction to orbitals’ energies is incorporated through the model Lamb-shift operator [22], whereas the correction from the vacuum polarization (VP) is accounted using Uehling potential [23]. The GTO parameters are optimized to reproduce the self-consistent field (SCF) and single-particle energies obtained from B-spline [24]. Tables ?? and ?? in Supplemental Material (SM) lists the optimized GTO parameters and comparison of GTO energies with B-spline data. We observe an excellent agreement for both single-particle as well as SCF energies with B-spline results. Further, Fig. ?? in SM demonstrates the convergence of computed properties with basis size.
Table I lists the excitation energies (EEs) for low-lying excited states of considered HCIs. As to be expected, Dirac-Fock (DF) constitutes the dominant part of EE in all three ions. For example, it accounts for approximately 53%, 118%, and 155% of total EE for first excited state in all three HCIs. The second largest contribution comes from the electron correlation due to residual Coulomb interaction. This exhibits a pronounced state dependence both in magnitude as well as sign due to different strengths of core-valence correlation effects. The largest cumulative Breit+QED contribution is observed to be in state of Cf (Fig. ?? in SM). This is a significant contribution and essential for high accuracy clocks. To the best of our knowledge, there are no reported values of experimental EEs for HCIs considered. In the case of Cf, for , there is a noticeable difference among previous results. The lowest and highest reported values differ by approximately 13%. Though Refs. [15, 16] employ the same CCSDT method, their results differ by about 4%. Our EE for is lower than all previous values. The
| States | DF | FSRCC | Breit+ QED | Total | Others |
| 8450 | 16288 | 16076 | , , | ||
| 233476 | 240772 | 240724 | , , | ||
| 20448 | 21516 | 21509 | , , | ||
| 51258 | 43302 | 100 | 43402 | ||
| 67020 | 60101 | 102 | 60203 | ||
| 183280 | 184398 | 77 | 184475 | ||
| 22176 | 14139 | 127 | 14266 | ||
| 40208 | 33221 | 117 | 33338 | ||
| 203445 | 203201 | 106 | 203307 |
Table I. TABLE I. Excitation energies (cm) of low-lying states of Cf, Cm and Bk HCIs using FSRCC theory. For a quantitative assessment of electron correlation, separate contributions from different correlation terms are also provided. Our final values are compared with other available theoretical results.
Ref. [11][CI + MBPT], Ref. [15][CCSDT], Ref. [16][CCSDT], Ref. [17][LCCSD].
lower value, compared to Ref. [11], could be attributed to the differences in the computational methodologies employed in the two works, whereas the difference with Ref. [16] may partly be attributed to different implementations of Breit interaction. For and states, our results are consistent with previous results and, as to be expected, more closer to the CCSDT results from Refs. [15, 16]. The observed differences among previous results and ours can be attributed to the varying treatment of electron-correlation and relativistic effects. For Cm and Bk, we observed large differences in the EEs for first two excited states compared to linearized coupled-cluster theory with single and double approximation (LCCSD) results of Ref. [17]. For example, for Cm, our results are larger by and 10%, respectively. For Bk, the difference is even more. The reason could be attributed to the different treatment of electron correlations in the two works. In Ref. [17], core-core and core-valence correlations are treated using a linearized CC theory, whereas the present work uses a non-linear CC theory. Additional differences could also arise because of different implementation of Breit and QED effects.
Table II presents our results on transition rates (A) and lifetimes () for some low-lying excited states of HCIs considered. For a quantitative assessment, we have explicitly given the relevant decay channels and corresponding contributions. To the best of our knowledge, there are no experimental results on for comparison for any of the considered HCIs. Our calculated for state of Cf is observed to be approximately three times larger than that of only available result [15]. The reason for this could be attributed to the difference in EEs in the two works. In contrast, the EEs of and states are in closer agreement with Ref. [15], resulting in good agreement between the corresponding s. A similar trend is also observed for Cm and Bk ions, where we observe a noticeable difference for first excited state compared to LCCSD results of Ref. [17]. This difference could be attributed to the more accurate treatment of core-core and core-valence correlations in our calculations. The contributions from Breit and QED corrections are observed to be not very significant in all three ions (Fig ?? in SM). The largest Breit+QED contribution is observed to be in the case of state of Cf. Whereas, perturbative triples are observed to have non negligible contributions, with a largest contribution of in the case of state of Cf.

Figure 1. Energy level diagrams of low-lying states of Cf, Cm and Bk HCIs. Energies, lifetimes and wavelengths shown in the figures are obtained from present calculations. The proposed clock transitions are indicated by double-headed arrows.
Table III presents our results on magnetic-dipole () and electric-quadrupole () hyperfine structure (HFS) constants for low-lying states of HCIs. For , our calculations show that the contributions from Breit, SE, and perturbative triples corrections are not very significant in all three ions (Fig. ?? in SM). Whereas, VP has a noticeable contribution of approximately to state in all three ions. For , we get negligible contributions from all higher-order relativistic and correlation corrections. To the best of our knowledge, no experimental measurements of or are available for any of the states considered. Theoretical calculations are too very few. Our calculated for state of Cf is approximately smaller than Ref. [15], whereas for state, the two calculations are closer. Since Ref. [15] does not provide separate contributions from higher-order corrections, a detailed term-by-term comparison is not feasible. It should, however, be noted that the uncertainty related to HFS constants in Ref. [15] is estimated to be 20-30%. Comparing our results for Cm and Bk, there are differences with only available calculation [17] using LCCSD theory. And, as mentioned earlier, the reason for this could be the missing non-linear correlation effects in Ref. [17], along with small differences Breit and QED corrections. For , our calculations provide the first theoretical data for the HCIs considered. Since both and s serve as a probe for electronic wavefunction in nucleus region, they constitute a stringent test for the accuracy of many-body wavefunctions in HCIs.
| Upper state | Lower state | Decay channel | (s) | (s) This work | (s) Other cal. |
| E2 | 5.115[−2] | 19.607 | 6.0 | ||
| E2 | 2.825[4] | 6.7[−6] | 7.0[−6] | ||
| M1 | 9.746[4] | ||||
| E2 | 4.715[3] | ||||
| M1 | 4.396[−1] | ||||
| E2 | 1.894[4] | ||||
| E2 | 7.427[−3] | 0.0084 | 0.0095 | ||
| M1 | 1.188[?] | ||||
| E2 | 3.158 | 0.318 | 0.576 | ||
| E2 | 2.936[−3] | 0.018 | 0.02 | ||
| M1 | 5.653[?] | ||||
| E2 | 1.320[4] | 1.5[−5] | |||
| M1 | 5.352[4] | ||||
| E2 | 4.476[?] | ||||
| M1 | 9.561[−2] | ||||
| E2 | 1.506[3] | ||||
| E2 | 9.874[−3] | 101.388 | 787 | ||
| E2 | 4.731[−3] | 0.012 | 0.013 | ||
| M1 | 8.284[?] | ||||
| E2 | 1.948[4] | 9.9[−6] | |||
| M1 | 7.271[4] | ||||
| E2 | 1.679[3] | ||||
| M1 | 2.237[−1] | ||||
| E2 | 6.310[3] |
Table II. FSRCC results for transition rates () and lifetimes () for low lying states of Cf, Cm and Bk HCIs. The values given for from this work also include the contribution from Breit, QED and perturbative triples corrections.
Systematic shifts– Now we present and discuss our results on leading frequency shifts for proposed clock transitions and assess their suitability for high-precision optical clock appliRef.[15][CCSDT], Ref.[17][LCCSD].
cations. Within the static approximation, the BBR-induced frequency shift for a clock transition between and states at temperature is given by
where is the differential static scalar electric dipole polarizability between and states. is the mean-squared electric field associated with blackbody radiation and is the Planck’s constant. As it is evident from Eq. (1), the accuracy of is governed by the accuracy of .
Table ?? in SM lists for relevant states of considered HCIs. The dominant contribution is observed to originate from the closed-shell core. For example, in the case of Cf, the core contributes approximately 82%, 94%, and 94% of total for , , and states, respectively. Similar trends are also observed for Cm and Bk ions. Among the valence-electrons, state provides the largest contribution of approximately 17-18% to . For and , the dominant valence contributions arise mainly from and states and contribute only a few percent to total . The Briet and QED corrections are found to have a negligible influence on . In contrast, perturbative triples are observed to have more significant contributions, with the largest one of 0.3%.
To the best of our knowledge, no experimental measurements of are available for any of the states considered. Theoretical data are also too scarce. Our calculated for state of Cf is approximately 18% larger than Ref.
[15]. The for state from two works are closer with a small difference of about 2%. For Cm and Bk, the only available theoretical results are those in Ref. [17]. Relative to this, the present calculations differ by approximately 13%, 4%, and 4% for , , and states, respectively. These differences could be attributed to more complete treatment of electron correlation in our work through nonlinear contributions in CC.
Using our calculated , we estimate the BBR shifts associated with proposed clock transitions in considered HCIs (Table IV). For clock transition in Cf, our estimated is approximately an order of magnitude larger than the value in Ref. [15]. This difference arises primarily due to the differences in used in the two calculations. The same trend is observed for clock transition in Cm with respect to Ref. [17]. In contrast, for transition in Cm is of the same order as that in Ref. [17]. For Bk, our estimated for transition is an order smaller than in Ref. [17]. Overall, our analysis indicates that transition exhibit smaller than transition in Cm and Bk. This is due to extremely small associated with fine-structure transition.
| Ion | Transition | (a.u.) | Others | |
| Cf | 0.1418 | [a] | ||
| Cm | [b] | |||
| Cm | [b] | |||
| Bk | [b] | |||
| Bk |
Table IV. Relative frequency shift due to BBR in Cf, Cm and Bk HCIs.
Ref.[15][CCSDT], Ref.[17][LCCSD].
Next, we estimate the EQ shift which originates due to an interaction between atomic quadrupole moment and an external electric-field gradient. In trapped-ion optical clocks, electric-field gradients generated by the trapping potential can perturb states with non-vanishing quadrupole moments, leading to shifts in the clock frequency. The interaction of electric quadrupole moment () with an external electric-field gradient gives rise to a frequency shift
where is the zero component of the electric quadrupole operator, denotes the difference in the expectation values of between and states, and .
The from our calculations for low-lying states are listed in Table V. For Cf, our calculated for state is in excellent agreement with the value reported in Ref. [15], except the sign difference which could arise due to different conventions used in the definition of single-electron quadrupole opRef. [15]-CCSDT, Ref. [17]-LCCSD,
| States | Cf Present | Cf Others | Cm Present | Cm Others | Bk Present | Bk Others |
Table V. The electric quadrupole moment () for low lying states of Cf, Cm and Bk HCIs.
erator. For Cm and Bk, our results slightly differ from those reported in Ref. [17] due to the reasons discussed earlier. Since EQ shift is directly proportional to the differential , provides a measure of the intrinsic sensitivity of a clock transition to the electric-field gradients. The differential are calculated to be for the transition in Cf and and for transitions in Cm and Bk, respectively. In contrast, for transition, since the two states have very close , differential is very small; , for Cm and Bk, respectively. These values are approximately five times smaller than those associated with transitions. Therefore, from the perspective of EQ systematic effects, the transition is expected to provide a more robust clock reference than transition in both Cm and Bk. However, the EQ shift can be eliminated for clock transition in all three systems by choosing the , hyperfine sublevels, for which the angular factor vanishes, irrespective of the magnitude of the electronic quadrupole moments.
We now assess the micromotion-induced frequency shifts. In radio-frequency (RF) Paul trap, the oscillating trapping field induces a driven motion of confined ion, commonly referred to as the micromotion. This gives rise to two competing contributions to the clock frequency: the second-order Doppler shift associated with the ion velocity and, an AC Stark shift caused by oscillating electric field. These two effects can be made to cancel each other at a particular RF drive frequency, known as the magic trap drive frequency. Operation of trap at this frequency suppresses the net micromotion-induced frequency shift and hence improves the attainable clock accuracy. The magic trap drive frequency can be expressed as
where is the ion mass ( is the mass of proton) and is the differential polarizability of the clock states. As given in Table IV, are obtained as a.u. for the transition in Cf and a.u. and a.u. for the transition in Cm and Bk, respectively. For transition, is significantly smaller, a.u. and a.u., respectively in Cm and Bk. It should be noted that, our computed for transition in Cf is of opposite sign compared to that in Ref. [15] due to a larger for state in our work. Our computed for state is, however, in close agreement with Ref. [15]. Since is positive for this clock transition, a magic trap drive frequency does not exist for Cf ion.
Further, we evaluated for the proposed clock transitions in Cm and Bk. To the best of our knowledge, the present work provides the first theoretical estimates of for Cm and Bk ions. For transition, is obtained to be MHz and MHz for Cm and Bk, respectively. In contrast, substantially larger values, MHz and MHz, respectively are obtained for transition in these ions. This considerably larger could be attributed to their extremely small . Although these small reflect a strong cancellation of AC Stark contribution arising from trapping field, they simultaneously require much higher RF drive frequencies to achieve a complete cancellation of the micromotion-induced shift. So, the present analysis shows that the micromotion-induced frequency shift can be effectively suppressed in the proposed clock transitions in Cm and Bk through operation at corresponding magic trap drive frequencies.
The magnetic-field induced frequency shifts are important sources of systematic uncertainty in the optical atomic clocks. In the weak-field regime, the Zeeman splitting to energy of a state is given by , where ( is the Bohr magneton and is the Lande g-factor) is the first-order Zeeman coefficient. The first-order Zeeman shift can be suppressed by choosing clock transitions with states or by averaging it with equal and opposite first-order shifts. The quadratic Zeeman shift then leads to the dominant frequency shift. The coefficient can be evaluated, within second-order perturbation theory [25],
where only levels with contribute. The expressions used in our calculations and the calculated hyperfine splitting energies and coefficients are provided in the SM.
For Cf, since the nuclear spin () is , the hyperfine structure is determined solely by the magnetic-dipole HFS constant . Using our calculated values (Table III) and the nuclear magnetic moment [26], the smallest differential coefficient is obtained to be kHz/(mT) for with or clock transitions. For a clock frequency of Hz and the magnetic field of T, this corresponds to a fractional frequency shift (FFS) of . Our computed FFS in good agreement with Ref. [15].
| States | FSRCC | Breit+QED +Triples | Total | Others |
| Cf | ||||
| 1764(3295) | −4(9) | 1760(3304) | 1900 | |
| 198766(0) | 791(0) | 199557(0) | 195000 | |
| 10450(26362) | −8(17) | 19442(26379) | ||
| 1231(3857) | 3(11) | 1234(3868) | ||
| Cm | ||||
| 157685(0) | 633(0) | 158314(0) | 154000 | |
| 1392(2381) | 9(6) | 1401(2387) | 1550 | |
| 1011(3324) | 2(12) | 1013(3336) | 689 | |
| 8821(24296) | −7(17) | 8814(22120) | ||
| Bk | ||||
| 178846(0) | 712(0) | 179558(0) | 174000 | |
| 1568(3138) | 8(6) | 1576(3144) | 1740 | |
| 1120(3673) | 0(10) | 1120(3683) | 772 | |
| 9644(24296) | −6(18) | 9638(24314) |
Table III. in the units of MHz/(MHz/eb) for low lying states of considered HCIs. The cumulative Breit, QED and perturbative triples corrections are listed separately. Values from previous works correspond to .
For Cm and Bk, since the data on the nuclear quadrupole moments are not available, to evaluate the hyperfine splitting energies we use the dominant magnetic-dipole interaction. We used the nuclear magnetic moments [27] and [27], respectively, to obtain the smallest for and transitions. For Ref.[11]-CI + MBPT, Ref.[17]-LCCSD.
Cm, we get as 103.575 and 255.65 kHz/(mT), with corresponding fractional shifts obtained as and at and Hz clock frequencies, respectively. For Bk, the corresponding s are 21.151 and 32.655 kHz/(mT), leading to fractional shifts of and at clock frequencies of and Hz, respectively. These results may be considered as estimates because the electric-quadrupole correction may modify the hyperfine splittings and, hence, . More precise predictions will require improved knowledge of the magnetic-field strength and nuclear magnetic moments.
Sensitivity to variation in the fine-structure constant
We now present and discuss our results on sensitivity coefficient () and enhancement factor () for the proposed clock transitions. The sensitivity coefficient is extracted by computing the gradient
where is the atomic transition frequency and is a perturbation arising due to change in from its current value . The enhancement factor can be obtained from the sensitivity coefficient using the relation .
To extract , transition energies were calculated for several values of in the vicinity of and is extracted from fitting the vs. data. In all HCIs, our calculated energy exhibit an excellent linearity (Fig. ?? in SM), demonstrating that higher-order terms in the expansion of are negligible over the investigated range. And hence, can be reliably identified as the slope of vs. curve. Since the fitting procedure utilizes multiple calculated points, it is less susceptible to numerical fluctuations than the two-points finite-difference estimates. Among the higher-order corrections in (discussed in SM), VP has the largest contribution, about 0.3% for state in Cm. Although small, these are important for obtaining reliable at accuracy commensurate with precision frequency standards.
For the HCIs considered here, a very few theoretical results on are available in the literature. Our computed is consistent with previously reported values for all HCIs, with minor differences due to more accurate treatment of electron correlation in our work. For example, in the case of Cf, our computed differ by , , and for corresponding states from CI+MBPT calculation [11]. Similarly, for Cm, our calculated for and states differ by and , respectively, from LCCSD calculation [17]. Whereas, for Bk, the corresponding differences are 0.8% and 0.6%, respectively. These deviations could be attributed to the inclusion of non-linear CC terms in present calculations, which are absent in the LCCSD method adopted in Ref. [17].
Our reported further highlights the suitability of the proposed clock transitions for probing possible variation in . Transitions with larger provide stronger experimental signature for probing variation. As evident from Table VI, in particular, the established clock transition in Cf exhibits a remarkably large of , making it one of the most sensitive optical clock transitions proposed to date for the test of -variation. Our obtained is in good agreement with the value, , reported in Ref. [11]. For Cm, transition possesses a of 16.8, whereas the has a comparatively smaller value, 1.61. This is in good agreement with the values, 15 and 2, reported in Ref. [17], respectively. In the case of Bk, our associated with transition increases significantly to 57.5, while transition exhibits a value of 1.62. The obtained in present work are substantially larger than those currently used in the state-of-the-art optical-clock comparisons for probing the variation in . For example, the comparison between E2 and E3 transitions in , one of the most sensitive laboratory probes of variation [10], is characterized by . In contrast, transition in Cf and transition in Bk exhibit exceeding 50. Such large values highlight the exceptional potential of these HCIs for future optical-clock experiments aimed at probing possible temporal variation of fundamental constants.
| States | Breit+QED | Others | |||
| 0 | 0 | 0 | 0 | 0 | |
| 0 | 1.6 | ||||
| 0 | 0 | 0 | 0 | 0 | |
| 692 | 16.8 | ||||
| 642 | 12.6 | ||||
| 536 | 2.9 | ||||
| 0 | 0 | 0 | 0 | 0 | |
| 770 | 57.5 | ||||
| 674 | 25.4 | ||||
| 840 | 2.9 |
Table VI. Relativistic sensitivity coefficients () (in ) and enhancement factors () for the proposed clock transitions in Cf, Cm and Bk HCIs with respect to their ground state. The uncertainties quoted for correspond to the asymptotic standard errors of the fit.
Conclusion
We have carried out a comprehensive FS-RCC investigation of the spectroscopic properties, systematic shifts, and sensitivity to variation in Cf, Cm, and Bk HCIs. To improve the accuracy of computed properties further, corrections from Breit interaction, QED effects and perturbative triples are incorporated. Further, we have estimated the systematic frequency shifts for the proposed clock transitions. Our obtained strongly suppressed frequency shifts indicate that the proposed clock transitions exhibit a remarkable robustness against external perturbations. These characteristics satisfy a key requirement for the realization of ultrahigh-accuracy optical clocks based on HCIs.
The relativistic sensitivity coefficients and enhancement factors determined in our work further reveal an exceptional sensitivity of considered HCIs to possible variation. In particular, clock transition in exhibits an enhancement factor of , placing it among the most sensitive optical-clock transitions proposed to date. Similarly, transition in possesses an enhancement factor of , demonstrating a comparable potential. Further, a comparison of the two proposed clock transitions in and reveals a clear distinction in their optimal applications. The transition is observed to exhibit a giant enhancement factor and is therefore particularly attractive for laboratory searches for the variation in . In contrast, transition displays reduced systematic shifts, making it a more favorable transition for the realization of an ultra-precise optical frequency standard. The upper bound to theoretical uncertainty in E1, E2, M1 and HFS matrix elements is 1%, whereas it is 2% in the case of .
**Acknowledgments—**We thank Suraj Pandey for useful discussions. PG acknowledges funding support from UGC (BININ04154142), Govt. of India. BKM, acknowledges the funding support from SERB, DST (Grant No. CRG/2022/003845). Calculations are performed using High Performance Computing cluster Tejas at Indian Institute of Technology Delhi and PARAM Rudra at Inter-University Accelerator Centre, New Delhi.
Supplemental Material: Fock-space relativistic coupled-cluster study of systematic shifts and fine-structure constant sensitivity in Cf, Cm and Bk highly charged ions
Palki Gakkhar; D. Angom; B. K. Mani
supplementary.pdf
Supplemental Material: Fock-space relativistic coupled-cluster study of systematic shifts and fine-structure constant sensitivity in Cf, Cm and Bk highly charged ions
Methodology
We have employed a FSRCC theory to calculate the many-body wavefunctions which are used to calculate the relevant clock’s properties and systematic shifts. For a one-valence atomic system, the many-body wavefunction is obtained by solving the eigen value equation
where is the exact atomic state of a one-valence system and is the corresponding energy. The Hamiltonian is the Dirac-Coulomb Hamiltonian [1]. In FSRCC, is expressed as
where and represent the unperturbed coupled-cluster (CC) operators for closed-shell and one-valence sectors, respectively. These are obtained by solving the nonlinear CC equations discussed in our previous works: Refs. [3] [2] for closed-shell and Ref. [3, 4] for one-valence systems. is the Dirac-Fock reference state for one-valence systems and is obtained by adding an electron to the closed-shell reference state, as . We have used CCSD approximation where and CC operators are truncated to have only one− and two−body operators as these alone provide a good description of the atomic properties. However, to improve the accuracy of our calculations further, we include the contributions from triples excitations perturbatively discussed in our previous works [4] [2].
Basis set and convergence of properties
To get reliable properties results using FSRCC theory, it is essential to use accurate single-electron basis. In this work, we have employed Gaussian-type orbitals (GTOs) [5] as the single-electron basis. GTOs form a finite basis set where single-particle wavefunctions are expressed as linear combinations of Gaussian-type functions (GTFs). The large component of the wavefunction is expressed as
where refers to the GTO index and is the total number of GTFs. The exponent is further expressed as , where and are two independent parameters. Parameters and are optimized to reproduce the self-consistent field (SCF) and single-particle energies obtained from B-spline [6] basis. Table S1 lists the optimized Gaussian parameters and S2 compares GTOs single-particle and SCF energies with B-spline data. As evident from the table, GTOs energies are in good agreement with B-spline energies. This confirms that a good quality basis is used in our FSRCC calculations.
| Atom | ||||||||
| Cf | 0.00400 | 1.825 | 0.00428 | 1.812 | 0.00555 | 1.875 | 0.00435 | 1.885 |
| Cm | 0.00340 | 1.783 | 0.00440 | 1.823 | 0.00555 | 1.875 | 0.00435 | 1.885 |
| Bk | 0.00300 | 1.825 | 0.00438 | 1.855 | 0.00355 | 1.875 | 0.00335 | 1.885 |
Table S1. The and parameters of the even tempered GTO basis used in our calculations.
Since GTO forms a mathematically incomplete basis, it is essential to check the convergence of the computed properties with basis size. Figure S1 shows the convergence of excitation energy with augmentation of orbitals in , , , , , , , and symmetries. As we observed from the figure, our excitation energy is well converged with respect to orbitals from each symmetries and the final converged basis contains orbitals. However, from our calculations, we find that properties converge even faster with a basis set having orbitals. This finalized basis is then used for the calculation of other properties. Corrections arising from the Breit interaction, QED effects and perturbative triple excitations are also evaluated and incorporated into the final results to improve their accuracy.
Figure S1. Panels illustrate the basis-set convergence for excitation energies of first excited states, in Cf and in Cm and Bk HCIs. Orbitals in , , , , , , , and symmetries are augmented until the excitation energy converges.
| Orbitals | Cf | Cm | Bk | |||
| GTO | B-spline | GTO | B-spline | GTO | B-spline | |
| 5011.01288 | 5011.03256 | 4759.46911 | 4759.40577 | 4883.95074 | 4884.00019 | |
| 977.86177 | 977.86519 | 920.08034 | 920.06781 | 948.60129 | 948.60991 | |
| 944.60465 | 944.61029 | 887.98170 | 887.98441 | 915.91749 | 915.92779 | |
| 750.91545 | 750.91510 | 713.64752 | 713.64746 | 732.14492 | 732.14424 | |
| 265.60734 | 265.60811 | 247.24488 | 247.24190 | 256.30067 | 256.30269 | |
| 250.41982 | 250.42134 | 232.62204 | 232.62284 | 241.39478 | 241.39757 | |
| 204.62434 | 204.62420 | 191.60805 | 191.60809 | 198.05151 | 198.05127 | |
| 181.10269 | 181.10257 | 168.83682 | 168.83686 | 174.90626 | 174.90604 | |
| 172.23562 | 172.23551 | 160.78358 | 160.78362 | 166.45368 | 166.45347 | |
| 82.07445 | 82.07466 | 74.63659 | 74.63576 | 78.29937 | 78.29990 | |
| 75.03788 | 75.03831 | 67.90586 | 67.90608 | 71.41601 | 71.41676 | |
| 62.52833 | 62.52828 | 56.81944 | 56.81944 | 59.63760 | 59.63751 | |
| 51.44848 | 51.44844 | 46.17107 | 46.17107 | 48.77410 | 48.77402 | |
| 49.18689 | 49.18685 | 44.14779 | 44.14779 | 46.63425 | 46.63418 | |
| 34.56576 | 34.56571 | 30.09211 | 30.09209 | 32.29597 | 32.29589 | |
| 33.95263 | 33.95259 | 29.55219 | 29.55218 | 31.72030 | 31.72023 | |
| 29.52684 | 29.52689 | 25.83993 | 25.83970 | 27.65303 | 27.65317 | |
| 26.49859 | 26.49871 | 22.99086 | 22.99102 | 24.71475 | 24.71494 | |
| 23.16480 | 23.16481 | 20.12674 | 20.12675 | 21.62302 | 21.62299 | |
| 18.49807 | 18.49809 | 15.73861 | 15.73864 | 17.09570 | 17.09569 | |
| 17.95858 | 17.95859 | 15.27849 | 15.27852 | 16.59694 | 16.59692 | |
| 13.57324 | 13.57327 | 11.46675 | 11.46668 | 12.50055 | 12.50059 | |
| Virtuals | GTO | B-spline | GTO | B-spline | GTO | B-spline |
| 11.80981 | 11.83081 | 9.84562 | 9.86589 | 10.80834 | 10.82901 | |
| 10.78452 | 10.79915 | 9.01054 | 9.02490 | 9.88144 | 9.89593 | |
| 11.84832 | 11.91451 | 9.61207 | 9.67432 | 10.70736 | 10.77163 | |
| 11.75515 | 11.81812 | 9.54026 | 9.59955 | 10.62519 | 10.68638 | |
| -32963.45287 | -32963.52204 | -31253.75038 | -31253.78154 | -32100.84541 | -32100.97833 |
Table S2. Comparison of single-particle and SCF energies (in Hartree) from GTO with B-Spline results.
Breit, QED and Perturbative Triples Contributions to Excitation energy, Lifetime and HFS constants
To further improve the accuracy of the calculated properties, corrections arising from the Breit interaction and QED effects, such as vacuum polarization and self-energy are incorporated. In this section, we provide some quantitative assessment of their contributions to these properties.
The dominant contributions from these higher-order relativistic effects to excitation energy are shown in Figure S2. Among the low-lying states, the largest Breit and QED corrections are observed for the first excited states. The Breit interaction contributes up to 0.4%, −0.1%, and −0.3% to the first excited states of Cf, Cm, and Bk, respectively. Whereas, the corresponding QED corrections amount to be −1.6%, 0.4%, and 1.2%. Although these corrections are considerably smaller than the dominant residual Coulomb interaction, their inclusion is important for achieving high-precision spectroscopic predictions.
Figure S2. Dominant contributions arising from Breit interaction, self-energy, and vacuum-polarization effects to the excitation energies of low-lying states of Cf, Cm, and Bk respectively.
Figure S3 shows the corresponding dominant contributions for lifetime. As it is discernible from the figure, contributions from these corrections are generally negligible for and states in all three HCIs. The largest Breit contribution is found to be −0.009% for state of Cf. Similarly, the largest QED contribution is approximately −0.01% for the first excited states, in Cf and in both Cm and Bk. Among all the considered states, the largest perturbative triples correction to the lifetime is found for state of Cf, amounting to approximately 0.3%.
Figure S3. Dominant contributions arising from Breit interaction, self-energy, and vacuum-polarization effects to the lifetime of low-lying states of Cf, Cm, and Bk respectively.
The dominant contributions from Breit interaction, self-energy correction, and vacuum polarization to magnetic dipole HFS constants are shown in Figure S4. As can be observed from the figure, contributions from Breit and self-energy are negligible. In contrast, the vacuum-polarization correction exhibits a noticeable contribution of approximately 0.4% to in all three systems. This behavior can be attributed to the significant electron density of state in nuclear region, which enhances its sensitivity to short-range QED effects.
Figure S4. Largest contributions arising from the Breit interaction, self-energy and vacuum-polarization effects to the magnetic dipole hyperfine structure constant in Cf, Cm and Bk respectively.
Electric dipole polarizability
In Table S3, we provide our results on static electric dipole polarizability () for clock relevant states using FSRCC theory. The corrections from Breit, QED and perturbative triples are listed separately for quantitative assessment of correlation effects.
| Cf | |||||||||
| 6p | 5f | 5f | |||||||
| E1 transition | E1amp | E1 transition | E1amp | E1 transition | E1amp | ||||
| 7s 6p | 0.3520 | 0.0099 | 6d 5f | 0.7937 | 0.0247 | 6d 5f | 0.8842 | 0.0225 | |
| 8s 6p | 0.0238 | 0.0 | 6d 5f | 0.1737 | 0.0011 | 5g 5f | 0.1963 | 0.0007 | |
| 6d 6p | 1.2494 | 0.1886 | 5g 5f | 1.1016 | 0.0259 | 5g 5f | 1.2505 | 0.0300 | |
| Core | 0.8907 | 0.8907 | 0.8907 | ||||||
| CCSD | 1.0892 | 0.9424 | 0.9439 | ||||||
| CCSD(T) | 1.0870 | 0.9443 | 0.9458 | ||||||
| CCSD(T)+Breit | 1.0870 | 0.9443 | 0.9458 | ||||||
| CCSD(T)+Breit+QED | 1.0859 | 0.9441 | 0.9454 | ||||||
| Other theory | 0.919 | 0.961 | |||||||
| Cm | |||||||||
| 6p | 5f | 5f | |||||||
| E1 transition | E1amp | E1 transition | E1amp | E1 transition | E1amp | ||||
| 7s 6p | 0.3779 | 0.0135 | 6d 5f | 0.9958 | 0.0497 | 6d 5f | 1.0668 | 0.0422 | |
| 6d 6p | 1.3311 | 0.2448 | 6d 5f | 0.1913 | 0.0017 | 5g 5f | 0.2335 | 0.0012 | |
| 5g 5f | 1.2033 | 0.0416 | 5g 5f | 1.4480 | 0.0459 | ||||
| Core | 1.1045 | 1.1045 | 1.1045 | ||||||
| CCSD | 1.3628 | 1.1975 | 1.1938 | ||||||
| CCSD(T) | 1.3587 | 1.1987 | 1.1950 | ||||||
| CCSD(T)+Breit | 1.3587 | 1.1987 | 1.1950 | ||||||
| CCSD(T)+Breit+QED | 1.3577 | 1.1982 | 1.1937 | ||||||
| Other theory | 1.1998 | 1.2460 | 1.2433 | ||||||
| Bk | |||||||||
| 6p | 5f | 5f | |||||||
| E1 transition | E1amp | E1 transition | E1amp | E1 transition | E1amp | ||||
| 7s 6p | 0.3642 | 0.0115 | 6d 5f | 0.8740 | 0.0337 | 6d 5f | 0.9583 | 0.0299 | |
| 6d 6p | 1.2954 | 0.2167 | 6d 5f | 0.1824 | 0.0014 | 5g 5f | 0.2144 | 0.0009 | |
| 5g 5f | 1.1084 | 0.0329 | 5g 5f | 1.3474 | 0.0371 | ||||
| Core | 0.9901 | 0.9901 | 0.9901 | ||||||
| CCSD | 1.2183 | 1.0581 | 1.0580 | ||||||
| CCSD(T) | 1.2146 | 1.0592 | 1.0591 | ||||||
| CCSD(T)+Breit | 1.2146 | 1.0592 | 1.0591 | ||||||
| CCSD(T)+Breit+QED | 1.2135 | 1.0590 | 1.0585 | ||||||
| Other theory | 1.0685 | 1.1091 | 1.1069 |
Table S3. Electric dipole polarizability (a.u.) for 6p, 5f and 5f states of Cf, Cm and Bk using sum-over-states approach. The dominant contributing E1 matrix elements are also provided to quantify the nature of electron correlations.
We have also provided the dominant contributing E1 transitions contributing to .
Zeeman Shift
The second-order Zeeman shift arising from the coupling between hyperfine-structure levels can be evaluated using second-order perturbation theory:
For the operator, only the terms satisfying contribute to the second-order Zeeman coefficient. The corresponding matrix elements are expressed in terms of angular-momentum coupling coefficients as [9]
The hyperfine energies are calculated as [10]
where and are the magnetic-dipole and electric-quadrupole hyperfine-structure constants, respectively, and
Using our calculated value of magnetic dipole hyperfine constants (), we determine the hyperfine-level energies , which are subsequently employed in the evaluation of the second-order Zeeman coefficients. Table S4, S5, S6 lists the calculated hyperfine-level energies and coefficients for the hyperfine levels of the proposed clock states of Cf, Cm and , respectively.
| 6p | 5f | ||||
| 0 | 2 | ||||
| 1 | 3 |
Table S4. Hyperfine-structure and second-order Zeeman coefficients () for 6p and 5f states with in Cf. The values of and are given in terms of MHz and , respectively.
| 4 | 2 | 1 | ||||||
| 5 | 3 | 2 | ||||||
| 4 | 3 | |||||||
| 5 | 4 | |||||||
| 6 | 5 | |||||||
| 7 | 6 | |||||||
| 7 | ||||||||
| 8 |
Table S5. Hyperfine-structure and second-order Zeeman coefficients () for , and states with in . The values of and are given in terms of MHz and , respectively.
| 3 | 1 | 0 | ||||||
| 4 | 2 | 1 | ||||||
| 3 | 2 | |||||||
| 4 | 3 | |||||||
| 5 | 4 | |||||||
| 6 | 5 | |||||||
| 6 | ||||||||
| 7 |
Table S6. Hyperfine-structure and second-order Zeeman coefficients () for , and states with in . The values of and are given in terms of MHz and , respectively.
Sensitivity Coefficient
Figure S5 shows the linear fit of vs. data for considered HCIs, which is used to extract the sensitivity coefficients for the proposed clock transitions. We further account for the contributions arising from the Breit interaction, vacuum-polarization effect, and self-energy correction to obtain the final recommended sensitivity coefficients, . These corrections are shown in the Figure S6. The self-energy contribution is negligible for all the states considered, while the vacuum-polarization correction remains below 0.3% across the three ions. Its largest contribution, approximately 0.3%, occurs for the state in Bk. In Cf, the vacuum-polarization correction is approximately 0.2% for the state, while in Cm, it is approximately 0.2% for all the states considered. The Breit contribution is even smaller, with the largest corrections being , , and for the state in Cf, the state in Cm, and the state in Bk, respectively. Although these contributions are small relative to the dominant Coulomb-correlation effects, their consistent inclusion provides a more complete assessment of the sensitivity coefficients and strengthens the reliability of the recommended values for precision clock applications.

Figure S5. Fit of the transition energies vs. data for Cf, Cm, and Bk HCIs. Bullet points represent our computed data, whereas linear fit is represented by solid lines.
Figure S6. Dominant contributions arising from Breit interaction, self-energy, and vacuum-polarization to the sensitivity coefficient in the case of Cf, Cm, and Bk.
Theoretical uncertainty
Based on the approximations used in our calculations, we identify four primary sources which contribute to the uncertainties of calculated one-body matrix elements. The first source of uncertainty arises from the truncation of basis set. Our computed properties are observed to converge well with a negligible change on further augmentation in the basis. Hence, we neglect the uncertainty associated with this source. The second contribution originates from truncating the dressed Hamiltonian to second order in in the evaluation of atomic properties [11, 12]. In our earlier investigations [4], iterative inclusion of higher-order terms revealed that the contributions from third- and higher-order terms in are below 0.1%. We, therefore, adopt 0.1% as an upper bound associated with this source. The third contribution arises from the partial treatment of triple excitations in FSRCC theory. Since perturbative triples account for dominant effects of triple excitations, the remaining higher-order contributions are expected to be relatively small. Among all transition properties computed in this work, the largest contribution of 0.3% is observed in the case of lifetime of state in Cf. Accordingly, we adopt 0.3% as an upper bound arising from this source. The fourth contribution stems from the omission of frequency-dependent Breit interaction. In our previous study of Ra [13], a series of GRASP2K calculations indicated that this effect contributes less than 0.13% to the calculated properties. Therefore, we take 0.13% as an upper bound from this source. Additional sources of uncertainty, such as higher-order couplings involving vacuum-polarization and self-energy corrections, quadruply excited cluster amplitudes, and other neglected higher-order many-body effects, are expected to contribute at a significantly lower level. Their cumulative contribution is estimated to be below 0.1%. Combining all these upper bounds, the overall theoretical uncertainty in our calculated E1, E2, and M1 transition matrix elements is estimated to be less than 1%.
Since the BBR shifts and magic rf-drive frequencies are derived from the calculated polarizabilities, uncertainties in these systematic shifts are governed primarily by the uncertainty in polarizability. To estimate this uncertainty, we refer to our previous work [14], in which we used a sum-over-states approach to calculate the scalar polarizabilities, , of the ground and excited clock states in Ca and Sr. And, we estimated an uncertainty of approximately 2% in . We therefore adopt 2% as a conservative upper-bound estimate for the uncertainties in the calculated polarizabilities in the present work.
For the electric-quadrupole shift, the relevant quantity is the electric quadrupole moment of the clock state. Although a separate convergence analysis of the quadrupole moments is not presented, we have analyzed the convergence of E2 transition matrix elements. Since both the quantities are determined by same rank-2 electric-quadrupole operator [10], the observed convergence of E2 matrix elements provides strong evidence for the reliability of our calculated quadrupole moments.
References
References
- [3]B. Mani, S. Chattopadhyay, and D. Angom, RCCPAC: A parallel relativistic coupled-cluster program for closed-shell and one-valence atoms and ions in fortran, Computer Physics Communications 213, 136 (2017).DOI
- [4]B. K. Mani and D. Angom, Atomic properties calculated by relativistic coupled-cluster theory without truncation: Hyperfine constants of Mg⁺, Ca⁺, Sr⁺, and Ba⁺, Phys. Rev. A 81, 042514 (2010).DOI
- [5]A. K. Mohanty, F. A. Parpia, and E. Clementi, Kinetically balanced geometric gaussian basis set calculations for relativistic many-electron atoms, in Modern Techniques in Computational Chemistry: MOTECC-91, edited by E. Clementi (ESCOM, 1991).
- [6]O. Zatsarinny and C. F. Fischer, DBSR-HF: A B-spline Dirac-Hartree-Fock program, Computer Physics Communications 202, 287 (2016).DOI
- [7]S. G. Porsev, U. I. Safronova, M. S. Safronova, P. O. Schmidt, A. I. Bondarev, M. G. Kozlov, I. I. Tupitsyn, and C. Cheung, Optical clocks based on the Cf¹⁵⁺ and Cf¹⁷⁺ ions, Phys. Rev. A 102, 012802 (2020).
- [8]V. A. Dzuba and V. V. Flambaum, Cm¹⁵⁺ and Bk¹⁶⁺ ion clocks with enhanced sensitivity to new physics, Phys. Rev. A 109, L021101 (2024).DOI
- [9]I. I. Sobelman, Atomic Spectra and Radiative Transitions, Springer Series in Chemical Physics, Vol. 1 (Springer-Verlag, Berlin, 1979).
- [10]W. R. Johnson, Atomic structure theory. Lectures on atomic physics (Springer, Berlin, 2007).DOI
- [11]B. K. Mani and D. Angom, Fock-space relativistic coupled-cluster calculations of two-valence atoms, Phys. Rev. A 83, 012501 (2011).DOI
- [12]R. Kumar, S. Chattopadhyay, D. Angom, and B. K. Mani, Relativistic coupled-cluster calculation of the electric dipole polarizability and correlation energy of Cn, Nh⁺, and Og: Correlation effects from lighter to superheavy elements, Phys. Rev. A 103, 062803 (2021).DOI
- [13]S. Chattopadhyay, B. K. Mani, and D. Angom, Electric dipole polarizability of alkaline-earth-metal atoms from perturbed relativistic coupled-cluster theory with triples, Phys. Rev. A 89, 022506 (2014).DOI
- [14]P. Gakkhar, D. Angom, and B. K. Mani, Fock-space relativistic coupled-cluster calculations of clock transition properties and systematic shifts in Ca⁺ and Sr⁺ optical clocks, Journal of Physics B: Atomic, Molecular and Optical Physics 59, 185001 (2026).DOI




