2.1. Experimental setup

The experiments on ICRF plasma build-up by the W7-X-like antenna were carried out in the U-2M stellarator-type device (see Pankratov et al. Reference Pankratov, Beletskii, Berezhnyj, Burchenko, Chechkin, Grigor’eva, Hartmann, Koch, Konovalov and Kulaga2010; Moiseenko et al. Reference Moiseenko, Berezhnyj, Bondarenko, Burchenko, Castejón, Chechkin, Chernyshenko, Dreval, Garkusha and Glazunov2011; Voitsenya et al. Reference Voitsenya, Shapoval, Pavlichenko, Pankratov, Chechkin, Moiseenko, Lozin, Dreval, Grigor’eva and Konovalov2014; Moiseenko et al. Reference Moiseenko, Lozin, Shapoval, Dreval, Kulyk, Mironov, Romanov, Pashnev, Sorokovoy and Petrushenya2016, Reference Moiseenko, Lozin, Kozulia, Mironov, Romanov, Konovalov and Shapoval2017a). U-2M (see figure 2) was assembled in the early 1990s (Bykov et al. Reference Bykov, Georgievskij, Demchenko, Kuznetsov, Litvinenko, Longinov, Pavlichenko, Rudakov, Stepanov and Tolok1990; Pavlichenko Reference Pavlichenko1993), operated briefly and then conserved. It was renewed and put to operation again at the end of 2006 (see Moiseenko et al. Reference Moiseenko, Stadnik, Schvets, Stepanov, Volkov and Tereshin2007; Beletskii et al. Reference Beletskii, Berezhnyj, Burchenko, Chechkin, Chernyshenko, Grigor’eva, Gubarev, Konovalov, Kulaga and Kurilo2008).

Figure 2.

General view of Uragan-2M.

The U-2M scheme is shown in figure 3. The major radius of the device is R = 1.7 m, the average plasma radius is r_pl < 24 cm and the toroidal magnetic field at the toroidal axis at present is B ₀ < 0.6 T (B ₀ = B_tt + B_th, B_tt and B_th are the toroidal fields produced by the toroidal and helical coils, respectively, the maximum magnetic field with respect to the mechanical stresses is B ₀ = 2.4 T). The U-2M magnetic system consists of l = 2 helical coils with four periods (m = 4) in the toroidal direction. The minor radius of the helical coils is r_h = 0.44 m. The additional toroidal magnetic field is provided by a set of 16 circular toroidal coils evenly distributed along the torus. The poloidal coil system that consists of eight coils is mainly aimed at eliminating the vertical field induced by the helical coils. It includes four trim (correcting) coils for control of the residual vertical field and four coils for the magnetic island control. Two DC sources are used for power supply for the magnetic system. Each DC source consists of two generators P-22-33-17 K with a peak power of 2750 kW each. The generators are operated in a pulsed mode (see figure 4).

Figure 3.

Scheme of Uragan-2M: I, poloidal field coils; II, helical field coils; III, toroidal field coils numbered 1–16. Different toroidal cross-sections are shown by red lines and denoted by capital letters and numbers.

Figure 4.

Time evolution of currents in helical (a, 1), toroidal (a, 2) and poloidal (correcting) (b, 3) coils of a magnetic system, where t_s is start time of the RF pulse.

In the U-2M stellarator, the RF methods are used to produce and heat the plasma. For this purpose, two RF systems are used: ‘Kaskad-1’ (K-1) and ‘Kaskad-2’ (K-2) (Korovin & Kramskoy Reference Korovin and Kramskoy2012). Both systems operate independently. Each RF system includes the RF generator, the transmission line, the matching circuit, the antenna and the control and diagnostic systems. The RF generators are single-staged and have a push–pull design. They use the GI-26A powerful tubes. Pairs of tubes are connected in parallel at each generator push–pull shoulder with the purpose of doubling the power output without increasing the anode voltage. The generators operate at a constant frequency in the range of 3–15 MHz and provide a pulse duration up to 100 ms. The maximum output power of the generators is about 0.8 MW. The output of the generators is coaxial bipolar. High-frequency power from the generator is delivered to the antenna by the feeder line, which consists of four coaxial cables (the RF cable RK-50-11-13 is used): two in each generator shoulder.

The toroidal vacuum chamber with the minor radius r_c = 0.34 m has a volume V_c = 3.88 m³ (without vacuum ports) and the inner surface area of S = 22.8 m². There are 48 ports used for diagnostics, gas puffing, vacuum pumping, etc. The vacuum chamber is pumped by three turbo-molecular pumps TMN-500 (cross-sections V, Z1 and V1 in figure 2), each has a pumping speed of 0.5 m³ s⁻¹ and is equipped with a liquid nitrogen cryotrap aimed at trapping heavy gas molecules such as water and located between the pump and the vacuum chamber. Exhaust ends of the three turbo-molecular pumps are connected in parallel to the fore vacuum pump AVZ-63D (the pumping speed is 0.063 m³ s⁻¹) through the nitrogen cryotrap. The ultimate residual pressure in the vacuum chamber is 1.1 × 10⁻⁵ Pa. There is a mass spectrometer IPDO-2 of the omegatron type with the omegatron tube RMO-4S for measuring the residual gas partial pressures in the vacuum chamber. To measure the total pressure in the vacuum vessel, the thermocouple PMT-2 and the ionization PMI-2 and PMM-32-1 pressure gauges are used. The puffing of the working gas into the vacuum chamber is made using the SNA-2 stationary gas puff system.

2.2. Two-strap antenna

The U-2M has a history of ICRF plasma production and heating studies. The single and double frame, the crankshaft, the three-half-turn and the four-strap antennae (Pankratov et al. Reference Pankratov, Beletskii, Berezhnyj, Burchenko, Chechkin, Grigor’eva, Hartmann, Koch, Konovalov and Kulaga2010; Moiseenko et al. Reference Moiseenko, Berezhnyj, Bondarenko, Burchenko, Castejón, Chechkin, Chernyshenko, Dreval, Garkusha and Glazunov2011, Reference Moiseenko, Lozin, Shapoval, Dreval, Kulyk, Mironov, Romanov, Pashnev, Sorokovoy and Petrushenya2016, Reference Moiseenko, Lyssoivan, Wauters, Tripský, Lozin, Pavlichenko, Kozulya, Dreval, Mironov and Romanov2017a) were used.

For plasma production studies, the existing vertical strap antenna, the THT antenna (Moiseenko et al. Reference Moiseenko, Lozin, Kozulia, Korovin, Beletskii, Baron, Grigor’eva, Chechkin, Mironov and Romanov2019), could be used. However, it does not excite low k _|| waves, and this could be an obstacle for plasma production (Lyssoivan et al. Reference Lyssoivan, Douai, Koch, Ongena, Philipps, Schüller, VanEester, Wauters, Blackman and Bobkov2012, Reference Lyssoivan, Wauters, Tripsý, Bobkov, Crombé, Douai, Kreter, Nicolai, Noterdaeme and Rohde2014).

Recently a new two-strap antenna was manufactured and installed at the U-2M stellarator. This antenna is similar in shape to the ICRH antenna of the W-7X and the main difference is that it is smaller. The antenna consists of two parallel poloidal straps (see figure 5a,b) made of stainless steel, 2 mm thick. The strap width is 60 mm and the length is 600 mm. The antenna is installed in the cross-section R1 (see figure 3). Its shape is fitted to the last closed flux surface, and the straps are placed at 10 mm distance from it (the magnetic configuration with K _φ = 0.32 is used, where K _φ = B_th/(B_tt + B_th)). Two antenna straps are oriented perpendicular to the toroidal axis. Part of the plasma column comes too close to the antenna port, and at this place, the small limiters are inserted at both sides of the antenna. The main antenna elements and limiters are covered with titanium nitride film.

Figure 5.

(a) Sketch of two-strap antenna, (b) two-strap antenna view in U-2M vacuum chamber (1 denotes straps, 2 denotes antenna limiter) and (c) connection scheme (L_{A 1} and L_{A 2} represent the antenna strap electric responses, C_cir, C_com are capacitors banks).

The two-strap antenna has four feeding points. The antenna can operate in different phasing, monopole or dipole, and be grounded or floating. This, in contrast to the THT antenna, gives more possibilities to control antenna ${k_\parallel }$ spectrum. The water cooling of the strap feed-throughs allows one to use the antenna in long pulses, up to 0.5 s. In the experiments reported here, the antenna operated in monopole phasing.

One of the electric connection schemes for monopole phasing is presented in figure 5(c). To match the impedance of the feeder line to the antenna, a matching device with a bi-polar L-circuit was used (Hayward Reference Hayward1994). For the dipole phasing, the serial connection of the straps is a possibility.

The RF power is calculated using the amplitudes of the forward and reflected waves in the feeder line measured by the directional coupler, according to the formula

(2.1)\begin{equation}{P_{RF}} = \frac{1}{2} \cdot \frac{{|{U_f}{|^2}}}{{{Z_0}}}(1 - |\varGamma|^2),\end{equation}

where Z ₀ is the impedance of the transmission line, Z ₀ = 25 Ω, $|\varGamma|$ is the reflection coefficient modulus, $|\varGamma| = |{U_r}|/|{U_f}|$, $|{U_f}|$ and $|{U_r}|$ are the voltage modules of the forward and reflected waves, respectively.

2.3. Diagnostic methods

As for plasma diagnostics, the time dependence of the linear electron density averaged over the line of sight was measured using a microwave interferometer (Pavlichenko, Zamanov & Kulaga Reference Pavlichenko, Zamanov and Kulaga2017, Reference Pavlichenko, Zamanov and Kulaga2018). The operational frequency of this super-heterodyne interferometer is equal to 140 GHz. The critical electron density for this frequency is 2.43 × 10¹⁴ cm⁻³. The interferometer launches the ordinary polarized wave through the horizontal equatorial port along the minor axis of the elliptically shaped plasma in the cross-section R (see figure 3).

U-2M is equipped with the time-resolved optical emission spectroscopy (OES). OES was carried out using a monochromator-spectrograph SOLAR TII by SOL instruments Ltd, model MS7501i with Cherny–Turner optical scheme and a photomultiplier tube (PMT). The signals from the photomultiplier are amplified by a broadband amplifier and transmitted to the data acquisition system. The OES system has four independent channels. The database (Kramida et al. Reference Kramida, Ralchenko and Reader2019) was used to identify the optical emission lines of atoms and ions. In addition to the integral emission registration, the scanning was made with 41 chords (see figure 6) in cross-section P1 (see figure 3). The chord distributions are measured using a pulse-by-pulse technique.

Figure 6.

Sketch of chord measurements.

Signals from the microwave interferometer and OES system are transmitted via the fibreoptic transmission line to the data acquisition system. The data acquisition system of U-2M is based on the high-speed ADC modules L-CARD E-20-10 with maximum acquisition frequency 10 MHz and module L-CARD L-783 that is used only for the acquisition of the data from the RF directional coupler. The magnetic system, the RF system, the diagnostic equipment and the data acquisition system are synchronized with the U-2M timing facility (Lyashchenko et al. Reference Lyashchenko, Ozherel’ev, Reva and Tsibliev2009).

3.1. Discharge characteristics

Helium was chosen as the working gas and hydrogen was always present during our experiments because the U-2M vacuum chamber inner surfaces were pre-loaded with hydrogen. Hydrogen ions are also produced by dissociative ionization of the residual water vapor and desorbed hydrogen-containing impurities from the wall.

The experiments were carried out at a helium pressure p = 0.145 Pa. The magnetic field value was B ₀ = 0.347 T, K _φ = 0.32, the current in the poloidal correcting coil was I_corr = 470 A. The RF generator frequency was f = 5.15 MHz, whereas the hydrogen cyclotron frequency was 5.27 MHz at the axis of the torus. In this case ω ≈ ω_ci (H⁺) ≈ 2ω_ci (He²⁺) ≈ 4ω_ci (He⁺). K-1 generator anode voltage was up to 7.5 kV. The timing was 10 ms (start), 11 ms (step 1, the first stepwise power increase), 12 ms (step 2, the final power increase) and 24 ms (shutdown). The RF generator anode voltage was varied stepwise as follows: U_{a 1} ≈ 0.4⋅U_a at the start, U_{a 2} ≈ 0.6⋅U_a at step 1 and maximal anode voltage U_a was set at step 2.

The RF power, the average density, and the spectral line intensities are shown in figure 7. Three stages could be separated in this discharge. The first stage is the plasma production that starts at 10 and ends at 15 ms. Figure 7 shows that the power radiated by the antenna varied significantly: P_RF is ~21 kW in the time interval between 10 ms (start) and 11 ms (step 1), P_RF is ~65 kW in the time interval between 11 ms (step 1) and 12 ms (step 2). The maximal power ~290 kW is registered after 12 ms (step 2) and later decreases down to ~160 kW. The initial plasma production and the plasma density growth occur at the first stage (the plasma production initial stage is discussed in detail in § 3.2). At the second stage that lasts from 15 to 18 ms, plasma density grows up to 2.4 × 10¹² cm⁻³. The atomic H* (H_Δ), He* (He I) spectral line intensities decrease and excited ions He^+* (He II), O^+* (O II) spectral line intensities increase until certain values. The power fed to the antenna is in the range of 160–190 kW.

Figure 7.

Time evolutions of (a) RF power and (b) average plasma density, and optical emission intensities of (c) He I, 447.15 nm, (d) He II, 468.6 nm, (e) H_Δ, 410.1 nm and (f) O II, 441.5 nm, where the K-1 generator anode voltage is U_a = 7.5 kV.

The average plasma density changes in the range of (2.2–2.6) × 10¹² cm⁻³ at the third stage that lasts from 18 to 24 ms while power fed to antenna decreases from 184 to 140 kW. Plasma decay begins after RF heating switch-off (24 ms). The average plasma density slowly ramps down with the characteristic time of about ~4 ms. He I, H_Δ and O II spectral lines intensities increase during the certain time after the RF heating switch-off, and He I afterglow is the most pronounced. It can be explained by electron temperature decrease and recombination of He⁺, H⁺ and O²⁺ ions with electrons. The absence of He II spectral line afterglow is likely a result of the small He²⁺ ion concentration in plasma. The ionization energy threshold for He⁺ ion is 54.4 eV, for O⁺ is 35.1 eV, for He is 24.6 eV and for H is 13.6 eV (Kramida et al. Reference Kramida, Ralchenko and Reader2019). Thus, effective He²⁺ generation by electron impact ionization requires a higher temperature than O²⁺ generation. Note that the spectral lines of the ions O⁴⁺ (OV, the ionization energy threshold is 113.9 eV) and C⁴⁺ (CV, the ionization energy threshold is 392.1 eV) were not detected in the experiments. This means that the electron mean energy is much lower than the lowest previously cited threshold.

Figure 8 shows the dependency of the maximal average plasma density on the K-1 generator anode voltage. Plasma density does not exceed 1.2 × 10¹² cm⁻³ at voltages lower than 6.5 kV, whereas it is about 2 × 10¹² cm⁻³ at higher voltage. Further increase of the voltage does not result in increasing plasma density. Note that the average plasma density was of the same order as in other experiments for RF plasma production (see, e.g., Esser et al. Reference Esser, Lyssoivan, Freisinger, Koch, Van Oost, Weschenfelder and Winter1997; Brakel et al. Reference Brakel, Hartmann and Grigull2001).

Figure 8.

Maximum average plasma density as a function of the anode voltage on the generator K-1.

3.2. Start-up details

Research on the initial plasma production stage of RF discharges in toroidal devices were carried out in experimental (Zaleskij et al. Reference Zaleskij, Kurilko, Nazarov, Plyusnin and Shvets1989; Carter et al. Reference Carter, Lysojvan, Moiseenko, Nazarov, Shvets and Stepanov1990; Lyssoivan et al. Reference Lyssoivan, Hartmann, Noterdaeme, Koch, Bobkov, Blackman, Braun, Cox, de Vries and Esser2005; Moiseenko et al. Reference Moiseenko, Lyssoivan, Wauters, Tripský, Lozin, Pavlichenko, Kozulya, Dreval, Mironov and Romanov2017b, Reference Moiseenko, Lozin, Kozulia, Korovin, Beletskii, Baron, Grigor’eva, Chechkin, Mironov and Romanov2019) and theoretical (Lyssoivan et al. Reference Lyssoivan, Moiseenko, Shvets and Stepanov1992, Reference Lyssoivan, Douai, Koch, Ongena, Philipps, Schüller, VanEester, Wauters, Blackman and Bobkov2012; Moiseenko et al. Reference Moiseenko, Stadnik, Lysoivan and Korovin2013; Tripský et al. Reference Tripský, Wauters, Lyssoivan, Bobkov, Schneider, Stepanov, Douai, Van Eester, Noterdaeme and Van Schoor2017; Wauters et al. Reference Wauters, Borodin, Brakel, Brezinsek, Brunner, Buermans, Coda, Dinklage, Douai and Ford2020) studies. Following these studies, separate plasma production stages can be identified that are characterized by different physical pictures.

The first stage (non-wave stage) is the initial stage of plasma production. Here, the electromagnetic waves do not propagate in toroidal vacuum vessels of fusion devices because of the low charged particle density ω_pe < ω (where ω_pe is the electron plasma frequency). The neutral gas is ionized by the electrons accelerated in the electric field of the RF antenna. The resulting low-density plasma (${N_e} \ll {N_0}$, where N ₀ is the density of neutral particles) propagates along the magnetic field lines and practically has no influence on the electromagnetic field structure.

Numerical simulations show transition to a second stage when the wave propagation conditions are met ω_pe $\gt$ ω (Tripský et al. Reference Tripský, Wauters, Lyssoivan, Bobkov, Schneider, Stepanov, Douai, Van Eester, Noterdaeme and Van Schoor2017), that cause an increase of the electron multiplication rate. Here, the plasma produced electrostatic field amplitude is of the same order as the vacuum E _|| field of the antenna. The plasma is mostly created via electron impact ionization. These collisions are accompanied by a significant electron power loss, which needs to be compensated for via the RF power supply.

At higher densities, the conditions given by the lower hybrid resonance ${\omega ^2} = \omega _{pi}^2 + \omega _{ci}^2$, the ICR ω = ω_ci and the fast wave cut-off density $\omega _{pi}^2 = (N_\parallel ^2 - 1){\omega _{ci}}(\omega + {\omega _{ci}})$ determine the final radial extend of the discharge as well as the coupling efficiency of the RF power (Lyssoivan et al. Reference Lyssoivan, Douai, Koch, Ongena, Philipps, Schüller, VanEester, Wauters, Blackman and Bobkov2012, Reference Lyssoivan, Wauters, Tripsý, Bobkov, Crombé, Douai, Kreter, Nicolai, Noterdaeme and Rohde2014). Both electrons and ions are heated through collisional absorption of the electromagnetic waves, with the additional possibility for resonant absorption on ions in the case of ICR.

The time of plasma production can be associated with the time of appearance of excited atoms and ions in the plasma t_app. This is the time between RF pulse start-up and the start of the emission signal. Figure 9 shows t_app values for excited atoms and ions of helium, hydrogen and oxygen. As can be seen in figure 9, t_app is different for different optical lines and decreases with RF generator anode voltage increase. The optical emission by excited atoms have weaker dependence of t_{app .} on U_a than that of the excited ions. The minimal values of t_app are observed for the hydrogen atom optical line H_Δ and helium line He I (figure 9a). H_Δ has a value of t_app between 0.13 and 0.2 ms, and the He I line has t_app ≈ 0.23–0.35 ms accordingly. The t_app difference between H_Δ and He I line is likely a result of different energy thresholds and cross-section value of excitation by the electron impact. For H_Δ 410.1 nm and He I 447.15 nm lines, the corresponding thresholds are 13.2 and 23.7 eV accordingly (Kramida et al. Reference Kramida, Ralchenko and Reader2019). Note that atomic hydrogen in ground or excited state appears in the discharge from electron impact with H₂ and other hydrogen-containing molecules in the processes of dissociation, dissociative ionization and dissociative attachment (Itikawa Reference Itikawa2001).

Figure 9.

Time delay of spectral line emission appearing for (a) helium atoms He I 447.15 nm, H_Δ 410.1 nm and (b) ions He II 468.6 nm, O II 441.5 nm as a function of anode voltage of K-1 generator. Dashed lines show time moments of stepwise K-1 anode voltage increase.

In the experiments, H and He I optical line emissions appears at the RF generator step 1 (low-power operation) at time interval from 10 ms (start) to 11 ms (step 1). The RF plasma production start-up occurs in this time interval. Then the second (wave) discharge stage begins. The plasma density increases, and the excited ion lines appear, first oxygen and then helium (figure 9b). Here t_app ≈ 1.4–2.3 ms for O II and t_app ≈ 1.9–2.6 ms for He II. The times t_app cover different RF generator operational intervals. At low anode voltage, these lines appear at step 3 of K-1 operation. With increasing anode voltage, their start times migrate to step 2 (figure 9b). The difference of t_app for excited ions and atoms by an order of magnitude may be explained by a cascade mechanism of ion excitation (Kovtun, Shapoval & Siusko Reference Kovtun, Shapoval and Siusko2019). The corresponding thresholds for the lines O II 441.5 nm and He II 468.6 nm are 26.3 and 51.0 eV, accordingly (Kramida et al. Reference Kramida, Ralchenko and Reader2019). The oxygen atoms and ions appear in the discharge from electron impact excitation of O₂ and other oxide molecules (e.g. water) during dissociative electron impact processes (Itikawa Reference Itikawa2001).

Some time after all the registered lines started to emit, the interferometer signal appears. Figure 10 shows the dependences of time interval needed for the average density to reach a certain value on the RF generator anode voltage. The dependences are ramp-down. They are weaker for low-density values and stronger for high-density values. This is an indirect indication that the electron temperature decreases with the plasma density increase.

Figure 10.

Time intervals to reach average plasma density (a) 5 × 10¹⁰ and 1 × 10¹¹ cm⁻³ and (b) 5 × 10¹¹ and 1 × 10¹² cm⁻³ as a function of the anode voltage on the generator K-1. Dashed lines show time moments of K-1 anode voltage increase.

As can be seen from the data obtained, the production time of the average plasma density of 1 × 10¹² cm⁻³ is less than 10 ms.

The times of appearance the optical line emissions and the times of reaching certain plasma density levels have a tendency to decrease with the anode voltage increase that is typical for the above shots. One can suggest that the electron temperature is higher for the higher anode voltage at the same other conditions. Thus, the inelastic collision rates increase with the electron temperature that happens when the trend of the corresponding rate coefficients is positive. This occurs when the electron energy is not so high, below 200 eV, the energy for the maximum electron impact ionization cross-section for helium (see Sorokin et al. Reference Sorokin, Beigman, Bobashev, Richter and Vainshtein2004).

3.3. Chord measurements

The chord measurements in a shot-by-shot regime for the optical lines He I and He II was carried out at the He pressure p $=$ 0.145 Pa; magnetic field at the device axis B ₀ $=$ 0.347 T, K_φ $=$ 0.318, poloidal correctional coil current I_corr $=$ 447 A and K-1 generator frequency f $=$ 5.13 MHz. The K-1 generator operated as follows: start at 10 ms, the first power increase at 11 ms (step 1), the second power increase at 12 ms (step 2) and switch-off at 20 ms; the anode voltage was U_a $=$ 5 kV. The time dependences of RF power, average density and H_Δ and O II spectral line intensities for these conditions are shown in figure 11.

Figure 11.

Time evolutions of (a) RF power and (b) average plasma density, and optical emission intensities of (c) H_Δ, 410.1 nm and (d) O II, 441.5 nm. The K-1 anode voltage is U_a = 6 kV.

For this experimental series, the location of the chords with reference to the impact parameter is shown in figure 6. The measurement results for the optical lines He I and He II are presented in figure 12. Drawings are of a similar shape despite the fact that in the first case, light is emitted by neutral helium atoms, and in the second case, by ions. As can be seen in figure 12(a,b), the radiation sources have the same dimensions as the plasma column (|P| $\lt$ 18 cm), with the ion luminescence region being somewhat smaller than that in the case of atoms. It can also be seen that the emission of the He I and He II lines along the chords changes in time, and the dependence of the line intensities on the impact parameter is not symmetric. Note that the average plasma density also changes in time (see figure 11b). Because the plasma production is transient and could be accompanied by plasma non-uniformity along the flux surface, conclusions on the profile shapes are hard to draw in this particular case.

Figure 12.

Time evolution of (a) He I and (b) He II chord distribution. He I wavelength is 447.15 nm, He II wavelength is 468.6 nm and P is the impact parameter (see figure 5).