Laser spectroscopy of the fine-structure splitting in the 2PJ levels of 4He
X. Zheng,Y. R. Sun,J.-J. Chen,W. Jiang,K. Pachucki,S.-M. Hu
2017-01-01
Abstract:The fine structure splitting in the 2PJ (J = 0, 1, 2) levels of He is of great interest for tests of quantum electrodynamics and for the determination of the fine structure constant α. The 2P0-2P2 and 2P1-2P2 intervals are measured by laser spectroscopy of the 2PJ-2S1 transitions at 1083 nm in an atomic beam, and are determined to be 31 908 130.98 ± 0.13 kHz and 2 291 177.56 ± 0.19 kHz, respectively. Compared with the calculations, which include the terms up to αRy, the deviation for the α–sensitive interval 2P0-2P2 is only 0.22 kHz. It opens the window for further improvement of theoretical predictions and an independent determination of the fine structure constant α with a precision of 2×10. * Corresponding authors. E-mail: robert@mail.ustc.edu.cn (Y. R. Sun); smhu@ustc.edu.cn (S.-M. Hu) As the simplest multi-electron system, atomic helium has played an important role in the history of quantum-electrodynamics (QED) through ever-refined comparisons between experimental data and theoretical predictions. The fine structure of the 2PJ (J = 0, 1, 2) levels of He, with a large splitting interval of 31.9 GHz, was recognized [1] as the best atomic system for the determination of the fine structure constant α. It is also the best system for studying exotic spin dependent interactions between electrons [2], as it is presently more sensitive than the positronium atom. The reasons include the relatively long life-time of the 2PJ levels in the helium atom as compared to those in the hydrogen atom, and the weak dependence on the finite nuclear size which usually limits the accuracy of theoretical predictions. In the past 50 years, the helium fine structure has been calculated with unprecedented precisions by gradually including terms up to the order of αRy [3-5]. Although, at present it is calculated with the highest precision among all the multi-electron atoms, the determination of α is less accurate than those from the electron magnetic moment anomaly [6] and atomic recoil [7]. Considerable efforts have been made by both theorists and experimentalists in the last few decades, but there are still disagreements among the reported results. The latest theoretical work by Pachucki et al. [5] gives an uncertainty of 1.7 kHz for both splitting intervals 2P0-2P2 (ν02) and 2P1-2P2 (ν12). As shown in Fig. 1, the experimental results for the ν12 interval from different groups [8-11] used to show obvious discrepancies even within the stated uncertainties. These have recently been resolved by taking into account the quantum interference corrections [12] which had not been considered in previous measurements. As for the ν02 interval of 31.9 GHz, it has been measured by several groups employing a variety of methods, including direct microwave spectroscopy with a thermal atomic beam [13], saturation absorption spectroscopy in a discharge cell [10], and laser spectroscopy with a thermal atomic beam [14]. As shown in Fig. 1, there are apparent disagreements among the experimental and theoretical values. Prior to this work, the most precise experimental measurement was given by Smiciklas and Shiner [14] as 31 908 131.25 kHz with an uncertainty of 0.3 kHz. However, the microwave and saturation absorption spectroscopy values differ from this value by 2.7 kHz (1.5σ) and 4.5 kHz (4.5σ), respectively. Note that this discrepancy cannot be accounted for by the quantum interference correction, which is pronounced for the ν12 interval but much less so for the ν02 interval. Therefore more independent measurements are required. In this work we present the most accurate experimental determination of the ν02 and ν12 intervals up to date using laser spectroscopy of He atoms. By using an intense atomic beam transversely cooled by a resonant laser field, the signal-to-noise ratio of the recorded spectrum is improved and the uncertainty of the determined ν02 interval is reduced to 0.13 kHz. This would allow an independent determination of the fine structure constant α with a precision of about 2×10, provided theoretical predictions reach a similar level of accuracy. FIG. 1. Comparison of the ν02 and ν12 values from experimental and theoretical studies. The results for the ν12 values are shown with (red) and without (blue) quantum interference corrections. The correction is much less for ν02 and not shown here. The corresponding methods are listed on the right side. Laser: laser spectroscopy with atomic beam. SAS: saturation absorption spectroscopy in discharge cell. MS: microwave spectroscopy with atomic beam. FIG. 2. (a) The configuration of the Experimental setup. (b) Optical layout. Abbreviations: AOM: acoustic-optic modulator; ECDL: external-cavity diode laser; EOM: electro-optic modulator; PD: photodiode; ULE: Fabry-Pérot interferometer made of ultra-low-expansion glass. (c) Diagram of the frequency intervals between the master laser and two ECDL lasers. (d) Transitions excited by the pump laser. (e) Transitions excited by the probe laser. The experimental setup consists of two parts: an atomic beamline and an optical bench, as shown in Fig. 2. The configuration of the atomic beamline is similar to that reported in our previous work [8]. Helium atoms at the 2S1 metastable state are first produced by radio-frequency discharge and then collimated by transverse-cooling with a laser on resonance with the 2S1-2P2 transition. A two-dimensional magnetooptical trap is used to slightly focus the atomic beam and a second transverse cooling field deflects the atoms at the triplet state (2S1) from the original atomic beam by an angle of 0.1° to eliminate the background of unwanted particles such as singlet state (2S0) atoms and UV photons [15-17]. A circularly polarized laser beam is applied to pump the atoms from the 2S1 (m=0) level to 2P1. After several cycles of excitation and spontaneous decay, over 99% of the atoms at the 2S1 (m=0) state are transferred to the 2S1 (m=+1) state (or m=-1, depending on the pump laser’s polarization). The atoms then enter the spectral probe region where the 2S1 (m=0) state is repopulated by the probe laser scanning through the resonance of the 2PJ-2S1 (J=0, 2) transition. This region is shielded with three layers of cylinder-shaped μ-metal, inside which a homogeneous magnetic field is generated by a cos(θ) coil. A Stern-Gerlach magnet is used to deflect the atoms at m=±1 levels so that only atoms in the m=0 level can reach the detector at the end of the beamline. Owing to an enhanced atomic beam intensity, the signal-to-noise ratio has been improved by a factor of 5 compared with our previous study [8]. A full beat-lock laser chain and a sideband-free probing system are used for spectroscopy in this work, which is quite different from that used in our previous study [8]. A narrow-band fiber laser (NKT Photonics Koheras BOOSTIK Y10, referred to as the master laser), with a linewidth of 10 kHz, is locked to a longitudinal mode of a temperature-stabilized Fabry-Pérot cavity made of ultra-low expansion glass [18, 19]. An external-cavity diode laser (referred to as the cooling laser) with its frequency tuned to the 2P2-2S1 transition is used for transverse cooling. A distributed feedback laser (DFB, referred to as the pump laser) with its frequency tuned on resonance with the 2P1-2S1 transition is used for optical pumping. A fiber EOM is used to produce ±1 sidebands of up to 16 GHz on the master laser frequency. The fiber EOM is driven by an RF synthesizer (R&S SMB100A) referenced to a rubidium clock (Spectratime GPS Reference-2000). Two external-cavity diode lasers (ECDL1 and ECDL2, referred to as probe lasers) are phase locked to ±1 sidebands respectively (see Fig. 2c), whose frequency can be fine-tuned near the transition resonances (2P2-2S1 and 2P0-2S1). Both beams from ECDL1 and ECDL2 are coupled into one single-mode optical fiber after a beam splitter. Two mechanical shutters (EOPC CH-60) are used for time-sequence control of the two probe lasers so that during each data acquisition cycle there is only one probe laser interacting with the atomic beam. A frequency scan covers the 2P2-2S1 and 2P0-2S1 transitions successively by switching between the two probe lasers, with 22 frequency points around each peak. The scan sequence is purposely randomized to avoid possible systematic shifts due to the frequency drift of the master laser. The laser system in our previous study [8] was similar to that used by Smiciklas and Shiner [14] in that the carrier laser and both sidebands interacted with the atomic beam in the probe region. Although only one sideband scanned though the resonance, there was still concern that the presence of the carrier and the other sideband could cause a systematic deviation. In this study we avoided the potential influence from sidebands by selecting only one beam (with no sidebands) into the probe region. Another advantage of present configuration is that the probe laser power can be stabilized with considerably better precision, which also eliminates the uncertainty due to variations of the fiber EOM sideband power. In total, we carried out about 7000 scans of the transitions 2P2-2S1 and 2P0-2S1. The spectra were used to derive the frequency interval ν02 and its statistical uncertainty is 0.06 kHz. Various systematic effects have been investigated and they are described in the following: Power dependence: There is a dependence of the measured frequency interval on the power of the probe laser. It is a result of recoil induced Doppler shift. Under each experimental condition, a series of measurements were carried out with different probe laser powers in the range of 1/20 to 1/4 of the transition saturation intensity (167 μW/cm). The frequency interval obtained is found to be linear with the laser power within this range. Measurements both with and without retro-reflected probe laser beam have been carried out. We also repeated measurements using both initial states (m=+1 and m=-1). As shown in Fig. 3, the sa