Nuclear Instruments and Methods in Physics Research B 205 (2003) 15–19 www.elsevier.com/locate/nimb Determination of the electronÕs mass from g-factor experiments on 12C5þ and 16O7þ Thomas Beier a,*, Slobodan Djekic b, Hartmut H€ affner b, Paul Indelicato c, H.-J€ urgen Kluge a, Wolfgang Quint a, Vladimir M. Shabaev d, Jose Verd u b, Tristan Valenzuela b, G€ unther Werth b, Vladimir A. Yerokhin c,d a d Gesellschaft f€ur Schwerionenforschung (GSI), Atomic Physics, Planckstrasse 1, 64291 Darmstadt, Germany b Institut f€ur Physik, Universit€at Mainz, 55099 Mainz, Germany c Laboratoire Kastler-Brossel, UPMC, case 74, 4 place Jussieu, F-75252 Paris Cedex 05, France Department of Physics, St. Petersburg State University, Oulianovskaya 1, Petrodvorets, St. Petersburg 198504, Russia Abstract We present a derivation of the electronÕs mass from our experiment on the electronic g factor in 12 C5þ and 16 O7þ together with the most recent quantum electrodynamical predictions. The value obtained from 12 C5þ is me ¼ 0:0005485799093ð3Þ u, that from oxygen is me ¼ 0:0005485799092ð5Þ u. Both values agree with the currently accepted one within 1.5 standard deviations but are four respectively two-and-a-half times more precise. The contributions to the uncertainties of our values and perspectives for the determination of the fine-structure constant a by an experiment on the bound-electron g factor are discussed. Ó 2002 Elsevier Science B.V. All rights reserved. PACS: 14.60.Cd; 06.20.Jr; 31.30.Jv; 32.10.Fn Keywords: ElectronÕs mass; g factor; Highly charged ions; Quantum electrodynamics; Fine-structure constant 1. Introduction The mass of the electron is a fundamental physical constant. In atomic mass units, its currently accepted value is given by [1] me ¼ 0:0005485799110ð12Þ u: ð1Þ This value is mainly based on the measurement of Farnham and co-workers [2,3] who subsequently * Corresponding author. Tel.: +49-6159-712116; fax: +496159-712901. E-mail address: [email protected] (T. Beier). observed single 12 C6þ ions and clouds of several electrons in a Penning trap. By comparing the cyclotron frequencies for both the ion and the electrons, the ratio of the electron mass to that of 12 6þ C is obtained. Earlier determinations were performed using mainly the proton–electron mass ratio [4–9] or the antiproton–electron mass ratio [8]. Wineland et al. [10] measured the ground-state g factor of 9 Beþ in a Penning trap and compared their experimental result with the value predicted in [11] and derived the mass ratio mð9 Beþ Þ=me . We have developed a setup to measure the g factor of a single highly charged ion with a precision of about 109 using a Penning trap [12,13]. In 0168-583X/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0168-583X(02)01968-7 16 T. Beier et al. / Nucl. Instr. and Meth. in Phys. Res. B 205 (2003) 15–19 our evaluations of the electronic g factors in 12 C5þ [13] and 16 O7þ [14], the published uncertainty of the electron mass proved to be the limiting factor. It thus seemed reasonable to compare experiment and the precise theoretical predictions ([15] and references therein) to obtain an independent value for the mass of the electron. For carbon, this was already performed in [16], with a result of me ¼ 0:0005485799092ð4Þ u. Here we are going to focus on the progress made since the publication of [16]. 2. Derivation of the electrons mass The g factor of the electron bound in a hydrogenlike ion is measured by determining the corresponding Larmor frequency xL , given by e B: ð2Þ xL ¼ g 2me The Larmor frequency is the frequency related to a spin-flip transition between the different magnetic angular-momentum substates in the given magnetic field B. In order to determine the actual magnetic field strength, the cyclotron frequency of the ion is measured, q xC ¼ B; ð3Þ mion where q denotes the charge of the ion. From the value xL =xC , the g factor is obtained as xL q me g¼2 : ð4Þ xC e mion Details of this procedure are given in [12–14,16]. The comparison of theoretically evaluated and experimentally determined g factors for hydro- genlike carbon and oxygen is given in Table 1. The experimental uncertainties resulting from the xL =xC measurement and from the electron mass given in [1] are presented separately. Experiment and theory agree reasonably within their error margins thus forming one of the most stringent confirmations of the bound-state quantum electrodynamical (QED) calculations carried out by Yerokhin et al. [17,18]. We stress that the new calculations for the QED corrections of order ða=pÞ, presented in [17,18], are in agreement with the older ones for the same contribution ([19] and references therein) but about one order of magnitude more precise for the self-energy correction. The major limitation to the experimental precision is the error margin of the published value for the mass of the electron [1]. If we employ the theoretically determined value gtheo for the g factor and solve Eq. (4) for me instead, we obtain the values for the electron mass presented in Table 1. The good agreement between our two determinations is visible. We are not going to perform any averaging, however, since both in theory and experiment too many correlated effects might be present that are not yet fully investigated. Not considering them might lead to a too small uncertainty of the averaged value if employing standard error propagation. Care has to be taken about the masses of the ions since the mass of the electron enters also in their values. The mass of a positive ion of charge state N is given by mðA X N þ Þ ¼ mðA X Þ N me þ EB =c2 ; ð5Þ where EB refers to the total binding energy of the missing electrons and c is the speed of light. Taking this into consideration, Table 1 Experimental and theoretical values for the g factors of 12 C5þ and 16 O7þ , the derived values for the mass of the electron, and the mass ratio of proton and electron, based on a proton mass of mp ¼ 1:00727646688ð13Þ u [1] 12 g factor from experiment g factor from theory Derived me mp =me C5þ 2.0010415963(10)(44) 2.0010415901(3) 0.0005485799093(3) u 1836.1526731(10) 16 O7þ 2.0000470268(16)(44) 2.0000470202(6) 0.0005485799092(5) u 1836.1526735(16) For comparison: in [1] a value of mp =me ¼ 1836:1526675ð39Þ is given. The experimental uncertainties resulting from the xL =xC measurement and from the electron mass given in [1] are given separately. T. Beier et al. / Nucl. Instr. and Meth. in Phys. Res. B 205 (2003) 15–19 e me ¼ ðmðA X Þ þ EB =c2 Þ q 2 xL þ1 : gtheo xC ð6Þ In both the numerator and denominator the second term is small compared to the first one. In particular, binding energies which are hardly better known than to a precision of 106 affect the relative precision of the final result only on a level DEB =ðmðA X Þc2 Þ. A detailed discussion on the binding energies is presented in [20]. 3. Uncertainties and the value of a In the determination of the electron mass, our experimental error margins are thought to include all experimental sources of error. A comprehensive discussion of these uncertainties is provided elsewhere [12,14]. On the theoretical side, the uncertainties quoted in [17,18] comprise all error margins currently known for the theory. The value given for the total uncertainty is largely dominated by an estimate. For the g factor of an electron bound in the 1s1=2 state of a hydrogenlike ion, the QED contributions 2 of orders ða=pÞ and higher are known only up to the first term in a (Za) expansion to be DgðQED; order ða=pÞn Þ ! 2 a n ðZaÞ þ ; 1þ ¼ 2 Að2nÞ p 6 ð7Þ where 2 Að2nÞ refers to the QED contribution of the same order to the g factor of the free electron, i.e. Að2Þ ¼ 1=2, Að4Þ ¼ 0:328 . . . [21,22], etc. An explicit derivation of Eq. (7) can be found in [23]. For precise bound-state QED calculations, an expansion in powers of ðZaÞ is hardly sufficient (cf. the reviews [24] and [25] and references therein). For the order ða=pÞ, a comparison between formula (7) and exact calculations is provided in [19]. For the order ða=pÞ2 and higher orders, however, calculations non-perturbatively in ðZaÞ were not yet performed and to get any estimate for these contributions, Eq. (7) has to be employed. In order to estimate the error due to this approximation, the difference of the non-perturbative calculation and formula (7) for the order ða=pÞ is taken, 17 multiplied by ðAð4Þ =Að2Þ Þ ða=pÞ. To be on the safe side, in [17,18] these values were multiplied by 1.5 to obtain a reasonable value for the error which at present also defines the total theoretical error of the g-factor prediction. The choice of this factor of 1.5 is somewhat arbitrary and a less optimistic value of 2 was employed for the same factor in [16]. Increasing the theoretical error in this way does not influence the total precision of our values for the electronÕs mass which at present are dominated by the experimental error margins. For the numerical uncertainties caused by the calculation of the terms of order ða=pÞ, we refer to [17,18]. The present value of the fine-structure constant a is given by a ¼ 1=137:03599976ð50Þ [1]. In the calculation of the g factor, this leads to an uncertainty of 2 1011 for 12 C5þ and of 3 1011 for 16 7þ O , respectively. This insensitivity on the finestructure constant is a feature of the particular Z region investigated so far, as the bound-state correction to the Dirac value for theffig-factor, given by qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi gDirac ¼ ð2=3Þ 1 þ 2 1 ðZaÞ 2 and the leading QED correction, given by ða=pÞ, almost cancel here (and thus the value of the g factor is rather close to 2). This is different, however, for even moderately higher Z. If we consider only the two mentioned contributions as crucial for the determination of a, it is quite possible to obtain also a value for a from the g-factor measurements from aðfrom gÞ 2 3 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 3 41 1 8 2 ¼ 2 Z ðgexpt 2 sÞ5; 4Z p p 3 ð8Þ where the ‘‘’’ in front of the square root has to be taken for Z 6 5. Here, s is a small number, given by the theoretical value of g with the crucial leading terms (in a) subtracted: a 2 2 s ¼ gtheo 2 þ ðZaÞ : p 3 ð9Þ The remaining dependence sðaÞ is negligible in the region Z 20 and below if a ‘‘reasonable’’ a (e.g. the value from [1]) is employed in the bound-state QED calculations. However, care should be taken when going to higher Z. In order not to perform 18 T. Beier et al. / Nucl. Instr. and Meth. in Phys. Res. B 205 (2003) 15–19 systematical errors, it has to be checked whether more terms have to be subtracted separately. The uncertainty da ðfrom gÞ from the determination Eq. (8) is given by " #1=2 2 1 8 2 da ðfrom gÞ ¼ Z gexpt 2 s p 3 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 : ð10Þ dgexpt þ dgtheo perspectives, we therefore include also a curve for da=a with a theoretical precision for the g factor of 109 over the whole range. At about Z ¼ 70, the precision in determining a from a g-factor experiment would match that of the value currently published in [1]. Ions of these charge numbers are planned to be investigated in the future HITRAP facility at GSI [28]. If we take 40 Ca19þ as an example and consider an experimental precision of 1 109 and the current theoretical one of 1:4 108 [18] for this ion, we thus obtain an error margin of 5:4 107 for a which is only slightly less precise than those values obtained by several other methods as listed in [1]. In Fig. 1, we display the current total theoretical uncertainties for gtheo [18] and the resulting uncertainties for a according to Eq. (10), assuming an experimental precision of 1 109 , over a large range of Z. The theoretical uncertainty increases 2 due to the ða=pÞ -bound-state contributions (cf. the discussion above) and also because of nuclear size and structure effects [26,27]. These latter are thought to be partly overcome by investigating also g factors of lithiumlike ions [26]. For future 4. Summary fractional uncertainty 10 10 10 10 10 We have presented our way of determining the mass of the electron from a high-precision g-factor experiment on hydrogenlike carbon and oxygen together with precise theoretical QED predictions. The obtained value for the mass of the electron from carbon is me ¼ 0:0005485799093ð3Þ u which is four times more precise than that of the best experiment up to now [2]. The value obtained from oxygen (me ¼ 0:0005485799092ð5Þ u) still is twoand-a-half times more precise than that from [2]. Both values match each other and agree also with the older one within 1.5 standard deviations. The precision of our experiment is such that even a determination of a competitive with other methods seems possible if we investigate heavier ions. -5 -6 (δα/α) from g, δgexp = 10 −9 (δgtheo/gtheo ) -7 (δα/α) if δgtheo = 10 -8 -9 Acknowledgements −9 (δα/α) from CODATA 1998 20 30 40 50 60 70 nuclear charge Z 80 90 Fig. 1. Current uncertainty of the theoretical predictions for g, based on the values for 32 S15þ , 40 Ar17þ , 40 Ca19þ , 52 Cr23þ , 74 Ge31þ , 132 Xe53þ , 208 Pb81þ and 238 U91þ published in [18], and derived values for da=a, according to Eq. (10) and assuming an experimental uncertainty of 1 109 . In addition the calculated values for da=a are shown, assuming also a theoretical uncertainty of 1 109 over the whole range of Z. For Z P 70, this uncertainty becomes smaller than that given in the recent compilation of fundamental constants [1]. Wiggles in the curves are due to spline interpolation between the calculated values. We thank R.N. Faustov, A.V. Nefiodov, A.S. Yelkhovsky and S.G. Karshenboim for stimulating and helpful discussions. Our work was supported by the European Community through the TMR network ‘‘EUROTRAPS’’ (contract number ERB FMRX CT 97-0144) and through the RTD network ‘‘HITRAP’’ (contract number HPRI-CT2001-50036), by the Russian Foundation for Basic Research (project no. 01-02-17248), and by the programme ‘‘Russian Universities’’ (project no. UR.01.01.072). References [1] P.J. Mohr, B.N. Taylor, Rev. Mod. Phys. 72 (2000) 351. [2] D.L. Farnham, R.S. Van Dyck Jr., P.B. Schwinberg, Phys. Rev. Lett. 75 (1995) 3598. T. Beier et al. / Nucl. Instr. and Meth. in Phys. Res. B 205 (2003) 15–19 [3] R.S. Van Dyck Jr., D.L. Farnham, P.B. Schwinberg, Physica Scripta T 59 (1995) 134. [4] G. G€ artner, E. Klempt, Z. Phys. A 287 (1978) 1. [5] G. Gr€ aff, H. Kalinowsky, J. Traut, Z. Phys. A 297 (1980) 35. [6] R.S. Van Dyck Jr., P.B. Schwinberg, Phys. Rev. Lett. 47 (1981) 395. [7] R.S. Van Dyck Jr., F.L. Moore, D.L. Farnham, P.B. Schwinberg, Int. J. Mass Spectrom. Ion Proc. 66 (1985) 327. [8] G. Gabrielse, X. Fei, L.A. Orozco, R.L. Tjoelker, J. Haas, H. Kalinowski, T.A. Trainor, W. Kells, Phys. Rev. Lett. 65 (1990) 1317. [9] B. de Beauvoir, F. Nez, L. Julien, B. Cagnac, F. Biraben, D. Touahri, L. Hilico, O. Acef, A. Clairon, J.J. Zondy, Phys. Rev. Lett. 78 (1997) 440. [10] D.J. Wineland, J.J. Bollinger, W.M. Itano, Phys. Rev. Lett. 50 (1983) 628 (1983) 1333. [11] L. Veseth, J. Phys. B 16 (1983) 2891. [12] H. H€ affner, T. Beier, S. Djekic, N. Hermanspahn, H.-J. Kluge, W. Quint, S. Stahl, J. Verdu, T. Valenzuela, G. Werth, Eur. Phys. J. D, in press. [13] H. H€ affner, T. Beier, N. Hermanspahn, H.-J. Kluge, W. Quint, S. Stahl, J. Verd u, G. Werth, Phys. Rev. Lett. 85 (2000) 5308. [14] J.L. Verd u, T. Beier, S. Djekic, H. H€affner, H.-J. Kluge, W. Quint, T. Valenzuela, G. Werth, Can. J. Phys. 80 (2002) 1233. 19 [15] V.M. Shabaev, V.A. Yerokhin, Phys. Rev. Lett. 88 (2002) 091801. [16] T. Beier, H. H€affner, N. Hermanspahn, S.G. Karshenboim, H.-J. Kluge, W. Quint, S. Stahl, J. Verd u, G. Werth, Phys. Rev. Lett. 88 (2002) 011603. [17] V.A. Yerokhin, P. Indelicato, V.M. Shabaev, Phys. Rev. Lett. 89 (2002) 143001. [18] V.A. Yerokhin, P. Indelicato, V.M. Shabaev, Can. J. Phys. 80 (2002) 1249. [19] T. Beier, Phys. Rep. 339 (2000) 79. [20] T. Beier, S. Djekic, H. H€affner, N. Hermanspahn, H.-J. Kluge, W. Quint, S. Stahl, T. Valenzuela, J. Verd u, G. Werth, Can. J. Phys. 80 (2002) 1241. [21] A. Petermann, Helv. Phys. Acta 30 (1957) 407. [22] C.M. Sommerfield, Ann. Phys. (New York) 5 (1958) 26. [23] A. Czarnecki, K. Melnikov, A. Yelkhovsky, Phys. Rev. A 63 (2001) 012509. [24] P.J. Mohr, G. Plunien, G. Soff, Phys. Rep. 293 (1998) 227. [25] V.M. Shabaev, Phys. Rep. 356 (2002) 119. [26] V.M. Shabaev, D.A. Glazov, M.B. Shabaeva, V.A. Yerokhin, G. Plunien, G. Soff, Phys. Rev. A 65 (2002) 062104. [27] A.V. Nefiodov, G. Plunien, G. Soff, Phys. Rev. Lett. 89 (2002) 081802. [28] W. Quint, J. Dilling, S. Djekic, H. H€affner, N. Hermanspahn, H.-J. Kluge, G. Marx, R. Moore, D. Rodriguez, J. Sch€ onfelder, G. Sikler, T. Valenzuela, J. Verdu, C. Weber, G. Werth, Hyp. Int. 132 (2001) 457.
© Copyright 2026 Paperzz