Visible transitions of highly charged tungsten ions observed with a compact electron beam ion trap

Tungsten is a major candidate for the divertor material of the next-generation fusion reactor ITER. Spectroscopic data of tungsten ions are thus necessary to diagnose and control the high-temperature plasma in ITER. In particular, there is strong demand in the diagnostics of the edge plasmas for emission lines in the visible range [1]. By using two electron beam ion traps (EBITs), the Tokyo-EBIT [2] and CoBIT [3], we have systematically measured visible spectra of highly charged tungsten ions.

In the previous study [4] with the Tokyo-EBIT, we observed visible spectra for the 340–400 nm range with electron energies of 1–1.7 keV. Several previously unreported lines from highly charged tungsten were observed, and the charge state of the ion responsible for those lines was studied from the appearance energy by observing electron energy dependence. However, for several lines observed at an electron energy of 1 keV, the appearance energy could not be identified because it was difficult to have an electron energy lower than 1 keV with the Tokyo-EBIT. In this study, the low-energy device CoBIT is used to study the appearance energy for those lines. The energy dependence of the spectra is studied in detail for the electron energy below 1 keV.

The experimental setup for this study is the same as that used in previous studies [5, 6]. To study tungsten ions, a vapor of W(CO)₆ was injected into CoBIT. Highly charged tungsten ions produced through successive electron impact ionization were trapped by the combination of the electrostatic well potential applied to the drift tube and the space charge potential of the magnetically compressed electron beam. Emissions from the trapped ions were observed with a Czerny–Turner spectrometer (Jobin Yvon HR320) with a 1200 gr mm⁻¹ grating blazed at 400 nm. A biconvex lens was used to focus the emissions on the entrance slit of the spectrometer. The diffracted light was detected by a liquid nitrogen-cooled-CCD (Princeton Instruments LN/CCD-1100-PB/VISAR). The wavelength was calibrated using emission lines from several standard lamps placed outside CoBIT.

Figure 1 shows the visible spectra obtained with electron energies of 780–940 eV.

Zoom In Zoom Out Reset image size

The electron beam current was 10 mA and the central magnetic field was 0.08 T for all the spectra. The data acquisition time was 30 min for each spectrum. The electron energy value was determined only from the output of the power supplies measured with a commercial digital multimeter. The values can thus be different from the real energy due to the space charge potential of the electron beam. In addition, the trap potential (30 V in this study) was not a squared well but rather like a 'u-shaped' trap, that could result in higher energy components in the trap.

The spectrum obtained at 940 eV well reproduced the spectrum previously observed with the Tokyo-EBIT at 1 keV [4]. The three lines (c, d and e) in the spectrum are labeled following the notations used in the previous paper [4]. Line d at 389 nm corresponds to the line identified as the ³H₅–³H₄ magnetic dipole (M1) transition in the ground state of W²⁶⁺ [4, 5]. The present observation also confirms the identification because the line was not observed at 780 eV, which is just below the ionization energy (784 eV) of W²⁵⁺, and appeared just after the energy was increased to 800 eV. When the energy was further increased from 800 eV, no additional line was observed until 920 eV, and lines c and e were observed only at 940 eV. Given the ionization potentials of W²⁶⁺, W²⁷⁺ and W²⁸⁺ (which are 833, 881 and 1132 eV, respectively), the ion responsible for lines c and e is considered to be W²⁸⁺.

Visible transitions of highly charged ions are mostly attributed to M1 transitions between fine structure levels in the same electronic configuration because transitions between different electronic orbitals should have a much shorter wavelength. However, since the ground state of Pd-like W²⁸⁺ has a closed shell structure of 4d¹⁰, there is no fine structure in the ground state. Thus, visible lines of W²⁸⁺ should be M1 transitions between fine structure levels of an excited state. For such transitions to have meaningful intensity, the excited state must have a long lifetime against decay to the ground state. Thus, our present observation implies non-negligible population of metastable states, such as 4d⁹ 4f with a large J, 4d⁹ 5s, etc in the EBIT.

However, if we accept the existence of such metastable states, we must consider the possibility of Rh-like W²⁹⁺ for the ion responsible for lines c and e because indirect ionization through the metastable Pd-like W²⁸⁺ is possible even when the electron energy is below the ionization energy of the ground state Pd-like W²⁸⁺ (1132 eV). A similar discussion was outlined in our previous paper [6], where the lines of Rh-like Ba and Xe ions were identified in the spectrum obtained with electron energies below the ionization threshold of the ground state Pd-like ions. The experimental result shown in figure 1 seems to support W²⁹⁺ rather than W²⁸⁺ because lines c and e appeared just above the threshold energy of the indirect ionization (about 930 eV).

To clarify which ion (W²⁸⁺ or W²⁹⁺) is responsible for lines c and e, observation was done with a longer exposure time (120 min) at 900 eV, which is just above the ionization energy of W²⁷⁺ but just below the threshold of the indirect ionization of W²⁸⁺. The result is shown in figure 2. As seen in the figure, lines c and e were definitely confirmed although the intensity is very weak. Consequently, we identify lines c and e as M1 transitions between fine structure levels in the metastable states (such as 4d⁹ 4f with a large J, 4d⁹ 5s) of W²⁸⁺. It appears that those lines were so weak that they were buried in noises for the 900 eV spectrum obtained with a 30 min exposure (figure 1). However, the possibility of M1 transitions between the fine structure levels (in the 4d⁹ ground state or metastable states such as 4d⁸ 5s) of W²⁹⁺ cannot be entirely excluded due to (i) possible errors in the ionization potential values taken from [7] and (ii) higher energy components that might arise from the trap potential. In order to make the identification more explicit, comparisons with theoretical simulation would be needed.

Zoom In Zoom Out Reset image size

This work was performed with the support and under the auspices of the NIFS Collaboration Research program (NIFS09KOAJ003) and JSPS KAKENHI grant number 23246165, and partly supported by the JSPS-NRF-NSFC A3 Foresight Program in the field of plasma physics.