Hyperfine Interactions 127 (2000) 271–276 271 Testing atomic structure theories with high-accuracy mass measurements on highly charged ions H. Häffner a , N. Hermanspahn b , P. Indelicato c , H.-J. Kluge a , E. Lindroth d , V. Natarajan e , W. Quint a , S. Stahl a , J. Verdú b and G. Werth b a GSI Darmstadt, Planckstr. 1, D-64291 Darmstadt, Germany Institut für Physik, Universität Mainz, D-55099 Mainz, Germany c Laboratoire Kastler-Brossel, Unité Mixte de Recherche du CNRS no. C8552, École Normale Supérieure et Université Pierre et Marie Curie, Case 74, 4 place Jussieu, F-75252 Paris Cedex 05, France d Department of Atomic Physics, Stockholm University, S-10405 Stockholm, Sweden e Department of Physics, Indian Institute of Science, Bangalore 560012, India b The mass of a highly charged ion is the sum of the mass of the nucleus, the mass of the electrons and the electronic binding energies. High-accuracy mass measurements on highly charged ions in a sequence of different charge states yield informations on atomic binding energies, i.e., the ionisation potentials. In our contribution we discuss the possibility of determining atomic binding energies of highly charged ions to better than 20 eV via cyclotron frequency measurements in a Penning trap. At this level of accuracy different contributions to the binding energies, like relativistic corrections, Breit corrections and QED corrections, can be measured. Keywords: atomic structure, mass measurements, Penning trap 1. Motivation Many-body problems occur in different fields of physics. In the case of atomic physics theoretical calculations of the electronic configurations of few- and multielectron systems can be performed to high accuracy [1]. Theoretical evaluation methods like relativistic many-body perturbation theory can be tested by experimentally determining the binding energy of a given electronic configuration. The masses of atomic or molecular ions can be measured to extremely high accuracy using Penning trap mass spectrometers [2,3]. The presently achieved experimental accuracy is sufficient to determine electronic binding energies for highly charged ions. High-accuracy mass measurements on highly charged ions in a Penning trap are performed by the Seattle group [2] and by the SMILETRAP group in Stockholm [4]. In this contribution we report on the measurement of the cyclotron frequency of a highly charged ion (C5+ ) in our Penning trap mass spectrometer with a fractional resolution of a few parts in 10−10 . We plan to extend these measurements in the future to heavy highly charged ions. J.C. Baltzer AG, Science Publishers 272 2. H. Häffner et al. / High-accuracy mass measurements Atomic theory: electronic binding energies In recent theoretical studies on cesium (Z = 55) in several charge states the contributions to the total atomic binding energies were investigated with two different computational schemes. The total binding energy of the neutral cesium atom, e.g., is more than 200 keV. The main part of the binding energy (about 99.5%) can be obtained from a single-configuration Dirac–Fock (i.e., relativistic Hartree–Fock) calculation. Beyond the single-configuration Dirac–Fock approximation, several contributions to the atomic binding energies are to be included: correlation energies due to the Coulomb interaction between the electrons, the Breit interaction (magnetic interaction between the electrons and retardation effects of the electron–electron interaction) and the radiative QED effects (self-energy and vacuum polarisation). These three effects are of the same order of magnitude. The radiative effects play a significant role only for the innermost shells; about 90% come from the 1s-electrons and about 10% come from the 2s-electrons. The correlation energy is proportional to the number of electrons and is of particular importance for the outer shells. The Breit interaction is also a two-particle interaction and scales with the number of electrons, but it has, in addition, a strong Z-dependence and the largest contributions are shifted towards inner electrons compared to the Coulomb interaction. In the comparison between binding energies of intermediately charged ions the Coulomb correlation is thus the most important ingredient. Since the Coulomb correlation contributes with around 100 eV to the total binding energy of an element as Cs, experiments which would test the atomic binding energy calculations should be accurate to at least 10 to 20 eV, which means an experimental accuracy of the mass measurement of about 1 part in 10−10 . At this level of accuracy there are discrepancies between different computational schemes, and measurements on this level would constitute a real challenge to theory. For heavy and highly charged ions, i.e., ions with up to three electrons, the radiative effects are important contributions to the binding energy and, e.g., a measurement with an accuracy of 5 parts in 10−11 of hydrogen-like uranium (Z = 92) would compete in accuracy with the most accurate experimental data available today. 3. Experimental principle In the Penning trap mass spectrometer the cyclotron frequency ωc of a single highly charged ion Q B (1) M with mass M and charge Q is measured in a homogeneous magnetic field B. In the comparison of the cyclotron frequencies ωc,1 and ωc,2 of two different ions the magnetic field B cancels and the ratio of the cyclotron frequencies ωc = Q1 M2 ωc,1 = ωc,2 Q2 M1 (2) H. Häffner et al. / High-accuracy mass measurements 273 gives directly the mass ratio. The mass Mi of an atomic ion with i electrons is composed of the mass of the bare nucleus Mnuc , the mass of the electrons i · me and the total (negative) electronic binding energy EBtot Mi c2 = Mnuc c2 + i · me c2 + EBtot . (3) The total electronic binding energy EBtot is the sum of the ionisation potentials of the different charge states where i is the number of electrons. The ionisation potential EBi of a highly charged ion, i.e., the binding energy of the outer electronic shell, is experimentally determined in the comparison of the cyclotron frequencies ωc,i and ωc,i−1 of the ion in two neighbouring charge states with i and i − 1 electrons, given by EBi Qi Mi c2 − me c2 − EBi ωc,i = . ωc,i−1 Qi−1 Mi c2 Solving eq. (4) for the ionisation potential EBi yields ωc,i Qi−1 i EB = − 1 Mi c2 + me c2 . ωc,i−1 Qi {z } | {z } | {z } | (i) (ii) (4) (5) (iii) With eq. (5) the ionisation potential EBi can be determined from the ratio of measured cyclotron frequencies ωc,i /ωc,i−1 , the known charge ratio Qi−1 /Qi , the mass Mi of the highly charged ion and the mass me of the electron. In the following we discuss the three expressions (i)–(iii) in eq. (5) and their contributions to the uncertainty in the determination of the ionisation potential EBi . (i) The cyclotron frequency ωc of a single ion in a Penning trap can be measured with very high accuracy. The principle of the Penning trap as a mass spectrometer is described in chapter 4. In measurements on C4+ - and C5+ -ions, e.g., an accuracy of 1.3 · 10−11 has been achieved [2]. With the assumption that the cyclotron frequencies ωc,i and ωc,i−1 can be measured with an accuracy of 10−10 the ionisation potential EBi of a highly charged ion with mass A = 200, for example, can be determined with an accuracy of about 20 eV. (ii) Mass units have to be converted to energy units, i.e., the mass Mi c2 of the highly charged ion has to be known in units of electron Volts (eV), in order to compare the experimentally measured masses with the theoretically calculated binding energy EBi . A systematic error arises from this conversion because the relation between the atomic mass unit Mu and the energy unit eV is known only with an uncertainty of 4 · 10−8 (Mu c2 = 931 494 013 (37) eV) [5]. However, the contribution of this uncertainty to the total systematic error in the determination of the binding energy EBi is in general negligible (<0.1 eV). The mass Mi of the highly charged ion can be expressed in atomic mass units Mu by comparing its cyclotron frequency to the cyclotron frequency of a carbon ion 274 H. Häffner et al. / High-accuracy mass measurements (e.g., C6+ or C5+ ) since the atomic mass unit is defined as 1/12 of the mass of the neutral carbon atom 12 C. For this comparison of cyclotron frequencies an accuracy of 10−6 is sufficient to reach an uncertainty of <1 eV in the determination of the binding energy EBi because the expression (i) in eq. (5) is a small quantity of the order of 10−5 . (iii) The mass me c2 of the electron in units of eV is known with an uncertainty of 0.021 eV (me c2 = 510 998.902 (21) eV). The contribution of this uncertainty to the total systematic error is negligible. 4. Penning trap mass spectrometer In a Penning trap a highly charged ion is stored in the combination of a homogeneous magnetic field and an electrostatic quadrupole field. The ion oscillates along the magnetic field lines with axial frequency ωz due to the electric harmonic binding force. In the plane perpendicular to the magnetic field lines the electric field transforms the free cyclotron motion into two circular motions: the perturbed cyclotron motion (with frequency ω+ < ωc ) and the magnetron motion (with frequency ω− ωc ) which is a ~ ×B ~ drift motion around the trap axis. The free cyclotron frequency ωc = (Q/M )B E is, therefore, not an eigenfrequency of the ion in the Penning trap. However, the frequency ωc can be derived from the three oscillation frequencies ω− , ωz and ω+ of the stored ion by [6] 2 2 ωc2 = ω− + ωz2 + ω+ . (6) Due to the hierarchy of the eigenfrequencies (ω− ωz ω+ ) only the reduced cyclotron frequency ω+ and the axial frequency ωz have to be measured with high accuracy. For a measurement of the cyclotron frequency of an ion in a Penning trap with an accuracy of better than 10−9 , it is essential to perform the experiment on a single ion because the Coulomb interaction between simultaneously trapped ions can substantially shift their oscillation frequencies [7]. Also deviations of the electrostatic trapping field from the ideal quadrupole geometry (anharmonicities) and inhomogeneities of the magnetic field shift the eigenfrequencies. The anharmonicities of the electric field can be minimised by adjusting the voltage applied to special compensation electrodes. The magnetic field homogeneity in the Penning trap is optimised by using only nonmagnetic material close to the trap center and by tuning the currents in special shim coils of the magnet which provides the magnetic field. With state-of-the-art superconducting magnets temporal stabilities of better than ∆B/B < 10−9 in one hour can be reached, which is sufficient to compare the masses of two ions with an accuracy of better than 10−9 . Our present experimental set-up is designed for the determination of the g-factor of the bound electron in light hydrogen-like systems [9,10]. A single stored C5+ -ion is nondestructively detected by picking up the image currents induced in the trap H. Häffner et al. / High-accuracy mass measurements 275 Figure 1. Cyclotron-frequency spectrum of a single trapped 12 C5+ -ion. The line width of the signal (∆ν/ν = 8 · 10−10 ) is due to magnetic field inhomogeneities. We, therefore, expect the frequency resolution to increase with improved magnetic field homogeneity. electrodes by the periodic ion motion. The image currents are transformed into a voltage signal by electronic resonant circuits attached to the trap electrodes. During the detection process the ion’s motional amplitudes are reduced, i.e., the ion is cooled, due to a dissipative force produced by the interaction with the attached circuit. After a few seconds the ion reaches thermal equilibrium with the resonant circuit. A demonstration of a cyclotron frequency measurement on a single trapped C5+ -ion is shown in figure 1. Using eq. (6) the cyclotron frequency of the single ion is determined in this specific example with a precision of 8 · 10−10 . The ion moves (with motional amplitudes of axial 50 µm, cyclotron 40 µm, magnetron 10 µm) in an inhomogeneous magnetic field (linear term 60 µT/mm, quadratic term 8.2 µT/mm2 ) resulting in frequency shifts and line broadening. The inhomogeneity is mainly due to a ferromagnetic ring which is needed to perform the g-factor measurement and which is situated close to the trap. We expect to improve the precision by more than one order of magnitude by removing the ferromagnetic ring and by tuning the currents in the shim coils, which will allow us to reach accuracies of better than 10−10 . 5. Conclusions and outlook High-accuracy mass measurements on highly charged ions in different charge states in a Penning trap make it possible to determine experimentally atomic binding energies, i.e., the ionisation potentials, of highly charged ions with an accuracy of better than 20 eV. Penning trap mass spectrometers are therefore an interesting tool for the investigation of the atomic structure of highly charged heavy ions. 276 H. Häffner et al. / High-accuracy mass measurements Acknowledgements We are grateful to Gerrit Marx and Johannes Schönfelder for helpful discussions. V.N. acknowledges travel support by a DST-DAAD fellowship. J.V. acknowledges financial support by the European Community (TMR Network CT-97-0144). References [1] J. Sapirstein, Rev. Modern. Phys. 70 (1998) 55. [2] R.S. VanDyck, D.L. Farnham, S.L. Zafonte and P. Schwinberg, in: Trapped Charged Particles and Fundamental Physics, eds. D.H.E. Dubin and D. Schneider, AIP Conference Proceedings, Vol. 457 (Woodbury, New York, 1999) p. 101. [3] F. DiFilippo, V. Natarajan, M. Bradley, F. Palmer and D.E. Pritchard, Phys. Scripta 59 (1995) 144. [4] C. Carlberg, H. Borgenstrand, T. Johansson, R. Schuch, I. Bergström, G. Rouleau, J. Stein and U. Surkau, Phys. Scripta 73 (1997) 347. [5] CODATA 98, NIST physical reference data, http://physics.nist.gov/PhysRefData/ contents.html (1999). [6] L.S. Brown and G. Gabrielse, Phys. Rev. A 25 (1982) 2423. [7] R.S. Van Dyck et al., Phys. Rev. A 40 (1989) 6308. [8] L.S. Brown and G. Gabrielse, Rev. Modern Phys. 58, 233 (1986). [9] W. Quint, Phys. Scripta 59 (1995) 203. [10] N. Hermanspahn, W. Quint, S. Stahl, M. Tönges, G. Bollen, H.-J. Kluge, R. Ley, R. Mann and G. Werth, Acta Phys. Polon. 27 (1996) 357. [11] D.J. Wineland and H.G. Dehmelt, J. Appl. Phys. 46 (1975) 919. [12] M. Diederich, H. Häffner, N. Hermanspahn, M. Immel, H.-J. Kluge, R. Ley, R. Mann, W. Quint, S. Stahl and G. Werth, Hyp. Interact. 115 (1998) 185.