1 INTRODUCTION

One of the main problems in designing a thermonuclear tokamak reactor or neutron source is solving the issue of the interaction of the plasma with materials of the vacuum chamber and divertor plates. The complexity of the problem is determined by the fact that these materials should pertain performance for several years under hot plasma flows with an energy of up to hundreds of electron volts, creating heat fluxes on the divertor plates of up to 10 MW/m2. In addition to constant flows, a significant number of rapid energy bursts of up to 100 MJ/m2 are also possible, leading to the evaporation of the surface layers of the material. No material can withstand such conditions, including the most refractory one, tungsten. It should be noted that in addition to sputtering, particle flows lead to the modification of the characteristics of the material on the surface, increasing its roughness, porosity, cracking and saturation of the material with hydrogen and helium. This can lead to additional erosion of materials. In this case, the sputtered material of the divertor and the wall leads to contamination of the plasma with impurities with a large nuclear charge, which cause large radiative energy losses and plasma cooling. Works on solving the problem of the interaction of the plasma with a wall began almost from the first experiments on tokamaks and has been going on for more than 60 years. However, the practical possibility of implementing long or permanent discharges arose only with the advent of tokamaks with superconducting magnetic systems in the 1980s. At the same time, the methods of non-inductive current maintenance began to be proposed. To date, it has become clear that it is not possible to achieve a constant operation mode even when providing stationary cooling of the wall. Moreover, the discharge duration decreases with increasing energy intensity of the discharge. In [1], this is explained by the gradual accumulation of erosion products of elements in contact with the plasma in the tokamak chamber. In this case, the discharge is terminated after reaching a critical amount of accumulated erosion products. Since the erosion rate increases with increasing energy intensity, critical conditions in this case are achieved faster and the discharge is shortened. This work shows that the maximum achievable discharge duration decreases with increasing energy flux density onto the wall. A similar dependence of the discharge duration decrease with the increase in the triple product value obtained in it (niTiτE) is given in [2], where ni is the ion density, Ti is the ion temperature and τE is the energy confinement time in the plasma. Since the maximum triple product values are achieved at high heating powers, the results of these works are in qualitative agreement. Another possible mechanism for limiting the discharge duration is overheating of surfaces in contact with the plasma discussed in [3]. In [3], a quantitative model for estimating the time limiting discharge based on the consideration of heat fluxes to the surface elements of the first wall is developed. This model is shown to be consistent with results from existing tokamaks.

Historically, two approaches to solving the wall problem in a tokamak have been identified. The first approach is to increase radiation losses, as a result of which the energy flows of particles from the plasma are almost completely converted into radiation, which does not lead to erosion of materials. However, the pure deuterium plasma has low emissivity, since radiation losses increase greatly with increasing ion charge. Therefore, for re-emission it is necessary to inject multiply charged impurities such as nitrogen, neon, argon, and krypton into the plasma. But their addition leads to plasma contamination, reducing the deuterium concentration and plasma cooling due to radiation losses from the central regions, since impurities with a large charge tend to accumulate in the center of the plasma. In addition, intense cooling of the periphery can lead to changes in discharge characteristics and degradation of energy confinement in the plasma. Another disadvantage of the radiative re-emission concept is that it can work under stationary conditions, but it does not prevent wall erosion during large pulse energy ejections when the periphery is unstable during Edge Localized Modes (ELMs) and disruptions.

The second approach is to implement the concept of a “renewable” wall, i.e., to create a constantly renewable layer on the surface of the wall. This layer absorbs the flow of energy and particles, protecting the base material from erosion. The practical implementation of this concept was carried out in several ways. The first way consisted of spraying various coatings onto the wall. In this case, carbon coatings were used, carbonization [4], silicon coatings, siliconization [5], and boron coatings, boronization [6]. However, this method is applicable for pulsed discharges, since the coatings are applied before the discharge and the sprayed layer is sputtered during the discharge. It has not been shown at present that such coatings can be renewed and the spent layer removed during the continuous discharge required for the reactor. The ultimate way to implement the renewable wall concept is to use liquid metals as a surface interacting with the plasma. In this case, almost all problems are solved. Thus, there is no surface degradation due to the convection of liquid lithium. Problems of material erosion and removal of sorbed sputtering products and working gas are solved by the possibility of pumping liquid metal. The use of lithium in a reactor concept was first proposed in [7]. A direct way to use liquid metals was to use a jet of droplets of liquid gallium [8] or a tray with liquid lithium in the case of a single-zero lower divertor [9] as a limiter. However, the implementation of a jet of droplets of liquid metal is technically difficult, and the use of the surface of liquid metal leads to its splashing due to the interaction of induced currents with the magnetic field. To eliminate metal splashing, it was proposed to cover the surface of the liquid metal with a porous structure [10]. In this case, the liquid metal is kept from splashing and forms a thin protective layer due to capillary forces. The concept of using capillary-porous structures was proposed in [10] and was used in experiments with liquid lithium [1113]. Recently, experiments have also been carried out using liquid tin [14].

Experiments on T-10 with lithium capillary-porous structures have been carried out for more than 10 years. The effect of lithium on the properties of plasma was studied and experience in working with such structures was obtained. A detailed analysis of changes in plasma characteristics, its composition, structure of lithium, impurity and deuterium flows in experiments with a lithium diaphragm on T-10 is presented in [15]. This work focuses on the technical aspects of the use of lithium capillary-porous structures, the study of the specific properties of these structures when interacting with plasma and the mechanisms of the impact of lithium on impurity flows and recycling.

Section 2 presents the results of experiments with lithium deposition prior to discharges in a T-10 chamber with graphite rail and ring diaphragms. Section 3 describes experiments using a capillary-porous movable lithium diaphragm when introduced into the Scrape of Layer (SOL) region and also into the main plasma under conditions of tungsten ring and rail diaphragms. In Section 4, the results of experiments with lithium are discussed, conclusions and recommendations are made.

2 EXPERIMENTS USING LITHIUM GETTERING OF THE T-10 TOKAMAK WALL

2.1 Experimental Conditions

The T-10 facility is a limiter tokamak with a circular cross section of a toroidal plasma column. The major and minor radii of the plasma were 1.5 and 0.3 m, respectively. Typical toroidal magnetic field and plasma current values are 2.4 T and 220 kA, respectively. The plasma column was limited by a circular poloidal diaphragm with a radius of 0.33 m and a rail diaphragm with a minor radius of 0.3 m, with a poloidal size of 0.35 m. The vacuum chamber was made of stainless bellows with a radius of 0.4 m. The first experiments with lithium on the T-10 began in 2006 and continued until 2011 [16]. During this time, the T‑10 operated with circular and rail graphite diaphragms. In this series, lithium was used only for sputtering on the surface of the chamber and diaphragms before experiments. The lithium cell was manufactured by the Krasnaya Zvezda Corporation on the basis of a capillary porous (10–100 μm) structure (CPS), which has already been successfully used in experiments on the T-11M [17] and FTU [18]. Lithium was located in a special container and spread due to capillary pressure along the CPS located on the outer front surface of the element. The element was equipped with a heater and thermocouples to measure temperature. It was introduced into the center of the T-10 tokamak chamber through a 30° port before the start of discharges in the same section where the rail and circular diaphragms were located. The experimental design is shown in Fig. 1. The lithium cell was mounted on a telescopic input system on the atmosphere. After fabrication, the surface of the capillary structure was coated with stainless foil, which was removed immediately before mounting. After mounting, the element was inserted into the input device and it was connected to the facility. Then the valve was closed and local pumping of the input device began. As a result, the lithium cell was exposed to the atmosphere for 10–20 min. After reaching a vacuum in the input device, the element was prepared by long-term heating of the lithium element at a temperature of up to 200°C for degassing after the atmosphere so as not to introduce impurities into the chamber during deposition. To sputter lithium in the chamber, a lithium element was introduced into the middle of the ch-amber and kept at a temperature of 450°C for 20 min. The typical amount of lithium deposited in the chamber was about 1 g. It was controlled by changing the duration of the temperature plateau during cooling of the element. This technique was proposed in [19]. Figure 2 shows a comparison of the time course of temperature during cooling of the element before the start of a series of experiments and after five deposition processes. It can be seen that the duration of the temperature plateau, corresponding to the time of the phase transition of lithium from the liquid to the solid state, decreased by 71 s. Since it was known that 16 g of lithium were initially charged, the loss of lithium mass per lithiation was estimated from the relative reduction in the duration of the temperature plateau as δm = (16/5) × (226 – 155)/226 = 1 g.

Fig. 1.
figure 1

Scheme of an experiment with lithium deposition at T-10.

Fig. 2.
figure 2

Determination of the lithium amount by the duration of the temperature plateau. (a) Before the experiment; (b) after five lithiations.

2.2 Effect of Lithium Sputtering on Deuterium Recycling

The main consequences of lithium gettering were an increase in deuterium sorption by the chamber and diaphragms and the level of impurities in the plasma. An increase in deuterium sorption was manifested in the rapid decay of the average density after switching off the gas supply. Figure 3 shows the time evolution of the average plasma density after the valve is switched off in a series of discharges after lithiation. In all discharges, a program was set to maintain the average density at the level of 2.5 × 1019 particles/cm3. However, in the first discharge, it was not possible to reach this value because of strong deuterium sorption, despite the full opening of the gas inlet valve. But already in the second pulse, the valve performance was enough to reach the specified level of the average density, but in this case, the valve was also completely open. Figure 3 shows a rapid density decay after the valve is turned off. However, the effect of lithium gettering decreased from discharge to discharge. Figure 4 shows the change in the decay time of the average density depending on the discharge number after lithiation. The decay time was found from the derivative of the decrease in the average density in the initial phase of the decay as: τ = –N/(dN/dt). The dotted line shows the decay time before lithiation. It can be seen from the figure that the density decay time doubles after five pulses, but the effect of lithiation remains noticeable even after 25 pulses. Recycling was determined by the decay time of the total number of particles in the column τ* = 1/NΣ(dNΣ/dt) after turning off the valve from the particle balance equation dNΣ/dt = Γlim+ Γst + ΓvalNΣ/τ, where NΣ is the total number of particles in the plasma column, τ is the lifetime of deuterium particles in the plasma τ = NΣ/(Γlim+ Γst) under stationary plasma conditions with recycling close to 100%, and Γlim, Γst, Γval are, respectively, deuterium inflows from the diaphragm, wall, and gas inlet. The amount of recycling is determined as R = (Γlim + Γst)/(NΣ/τ), the ratio of the inflow from the diaphragm and wall to the total flow from the plasma. Then we get R = 1 – (τ/τ*) from the balance equation. Figure 5 shows the time evolution of the total number of particles in the plasma for the second pulse after lithiation. The total number of particles was found by summing the chord data of a 16-channel interferometer. It can be seen that the total number of particles begins to decrease at a rate of 3 × 1020 particles/s but the decay rate increases to 5.7 × 1020 particles/s after 25 ms, which may indicate deterioration in the plasma confinement. Deterioration of confinement under low (0.3) recycling conditions with pellet injection feeding was found at the ASDEX tokamak [20]. Deterioration of plasma confinement is also evidenced by a significant increase in the level of density fluctuations detected in this discharge at the density decay stage [21]. The decrease rate in the number of particles of 3 × 1020 particles/s at NΣ = 7.2 × 1019 corresponds to τ* = 0.24 s. This decay rate of the total number of particles can be an estimate of the inflow of particles from the valve (Γval) in the stationary stage. A close value of 2 × 1020 particles/s was obtained from an estimate of the growth rate of the number of particles when the valve is fully turned on in a discharge with recycling close to 1 [22]. In this work, it was also found that the ratio of the integral emission of the Hα line from the diaphragms to the valve was 4.8. Then taking the inflow from the valve of 3 × 1020 particles/s, the inflow from the diaphragm can be estimated as 1.44 × 1021 particles/s. According to [23], the inflow from the diaphragms is 65% of the total deuterium inflow. Accordingly, the total inflow into the plasma can be estimated as 2.2 × 1021 particles/s. This estimate of the total deuterium inflow into the plasma is close to 2.2 × 1021 particles/s measured in [15, 23], which gives the value τ = 0.033 s. Thus, it can be estimated that the recycling value in the second pulse after lithiation was R = 1 – (τ/τ*) = 0.86.

Fig. 3.
figure 3

Time evolution of the average plasma density after valve shutdown in a series of discharges after lithiation. (1) First discharge after lithiation (61 390); (2) second discharge after lithiation (61 391); (3) tenth discharge after lithiation (61399); (4) twenty-third discharge after lithiation (61412).

Fig. 4.
figure 4

Change in the density decay time in a series of discharges after lithiation.

Fig. 5.
figure 5

Time evolution of the total number of particles in the column after turning off the gas influx in the second pulse after lithium deposition.

2.3 Change in the Plasma Impurity Composition

In addition to reducing recycling, lithium sputtering also led to a decrease in the concentration of carbon, oxygen impurities, and heavy impurities, which determined a decrease in radiation losses from the plasma. Figure 6 shows the radial distributions of the local radiative loss density measured by the bolometer and AXUV semiconductor detectors. In both cases, radial power losses were obtained by abelianization of multichord measurement data. It can be clearly seen that both diagnostics show that radiation losses are significantly reduced. Moreover, if the reduction in radiation losses at the periphery, for which light impurities are responsible, is less than two times, then losses in the central regions caused by heavy impurities are reduced by more than half. This may be due to both a decrease in the inflow of heavy impurities into the discharge and a decrease in the neoclassical accumulation of heavy impurities in purer discharges after lithiation [15].

Fig. 6.
figure 6

Radial loss distributions recorded by a pyroelectric bolometer and AXUV sensors in discharges before and after lithium deposition. (solid black curve) A pyroelectric bolometer before lithiation; (red dashed curve) after lithiation; (black dash-dotted curve) with AXUV points before lithiation; (red dotted curve) AXUV after lithiation.

Figure 7 shows the change in plasma parameters during the experimental campaign with lithium in 2008. Plasma parameters were measured in the first pulses of the working day. The top left plot shows the names of the procedures performed before the start of the experiment of each day. Red bars indicate lithiation procedures, and their height corresponds to sputtering duration of 10 and 20 min. Lithium was not sputtered before the remaining working days, but various chamber cleaning procedures were carried out marked in the figure. The work was carried out in a constant discharge conditions with a plasma current Jp = 200 kA, toroidal magnetic field Bt = 2.4 T and average density 〈Ne〉 = 2.5 × 1019 m–3. Since recycling and the impurity composition of the discharge varied during the day, Fig. 7 shows the data corresponding to the first pulse in which the density reached a given level. This was almost achieved in the second pulse. In the first two days of the experiment, the discharge conditions were measured before the start of lithium deposition. It can be seen that each lithiation is accompanied by a strong decrease in recycling and inflow of carbon and oxygen impurities measured near the diaphragm. A decrease in the inflow of impurities leads to a decrease in the effective charge of the plasma Zeff, a decrease in radiation losses measured by the pyroelectric bolometer and AXUV semiconductor detectors, as well as the intensity of soft X-ray (SXR) radiation from the plasma. However, when lithiation was not carried out, the level of impurities began to increase. Deterioration was slight on day 4. But after two days without lithiation (7 and 8), the discharge characteristics almost returned to those before lithiation. It should be noted that the intensity of the LiII line, which characterizes lithium flows, decreased only by half during these days. This may indicate the reduced effectiveness of lithium in reducing recycling and carbon and oxygen levels because of its poisoning by impurities. Data in Fig. 7 show a high correlation of changes in such integral discharge characteristics as loop voltage and radiation losses with impurity flows from the diaphragm, while the carbon flow far from the diaphragm changes little and gradually. This suggests that the main source of impurities occurs from the diaphragm and sputtering lithium effectively reduces the level of impurities in the discharge in this area. It should also be noted that a glow discharge in helium does not lead to the recovery of the sorption properties of lithium.

Fig. 7.
figure 7

Changes in some discharge characteristics in the course of an experimental campaign with lithium deposition. In two lower right-hand panels CIIIA, OIIA and CIIIC, designation A means measurements in the cross section of the diaphragm, and designation C means those in the cross section opposite the diaphragm.

2.4 Stability of Lithium in a Capillary-Porous Structure

In a series of first experiments, the introduction of a lithium element into the plasma was not planned, since it was mounted on a long console, which could not withstand the Ampere forces when the current flows from the plasma to the element in a magnetic field. Therefore, after deposition, the lithium element was cooled and removed from the port into the container. However, in one series, the lithium cell was removed from the chamber, but remained in a hot state inside the tokamak port. In this case, multiple splashes of lithium were observed throughout the port when the chamber was opened to the atmosphere. This was due to the fact that when the chamber vibrated during the discharge, the resulting forces were greater than the holding capillary forces. A similar phenomenon of lithium splashing was observed in experiments on the Italian FTU tokamak [24] when the lithium element was poorly fixed. Therefore, to prevent splashing, the lithium cell should be securely fastened.

2.5 Comparison of the Results of Experiments with Lithium Deposition with Similar Ones at Other Facilities

Similar experiments with lithium deposition on the chamber walls before discharges were carried out at several toroidal plasma facilities. At the American divertor spherical tokamak facility NSTX [25], lithiation was carried out using lithium evaporators located on top of the torus. Before the discharges, lithium was sprayed onto the divertor region located below. The authors note that spraying leads to a decrease in density, mainly at the periphery of the column, associated with a decrease in recycling. However, stabilization of the peripheral Edge Localized Mode instability as a result of lithium sputtering leads to improved confinement and accumulation of carbon impurities in the central regions of the column. It should be especially noted that the lithium concentration itself in the center remains less than 0.1%. Lithium sputtering before discharges was also carried out at the Chinese divertor tokamak EAST [26]. There is a strong decrease in the level of light impurities of carbon and oxygen. The effective plasma charge Zeff was reduced to less than 2. Deuterium recycling after deposition was reduced to 0.89 immediately after deposition, allowing for better density control. Recycling increased to 0.96 after 100 discharges. Lithium sputtering was also carried out at the Italian RFX facility [27]. Sputtering was carried out, just like at T-10, using the same capillary-porous structure located at the bottom of the chamber. As a result, a decrease in the main carbon impurity and recycling R to 0.95–0.99 is observed. Better density control is also noted.

Thus, we may conclude that lithium sputtering before discharges on several facilities leads to results close to those obtained on T-10, namely, significant reduction in impurity levels and recycling.

3 EXPERIMENTS WITH A LITHIUM DIAPHRAGM BASED ON A CAPILLARY-POROUS STRUCTURE

Long-term operation of the T-10 with graphite diaphragms led to the formation of hydrocarbon films on the chamber and an increase in carbon impurities in the plasma. In order to reduce light impurities and explore the possibility of working with tungsten chamber elements, the rail and circular graphite diaphragms were replaced with tungsten ones in 2015. In 2016, in addition to them, a movable lithium diaphragm based on a capillary structure was installed. Thus, in addition to previous experiments with graphite diaphragms, the opportunity was implemented to study the effect of lithium under tungsten diaphragm conditions, including the possibility of using a lithium element as the main limiting diaphragm.

3.1 Experimental Conditions

The arrangement of tungsten and lithium diaphragms in the T-10 chamber is shown schematically in Fig. 8. The rail (bottom) and circular tungsten diaphragms are indicated in black. They were manufactured at the JSC Efremov NIIEFA. Tungsten of Polema brand was used. The radius of the circular and rail diaphragms was 33 and 30 cm, respectively. A lithium diaphragm that moves from pulse to pulse was mounted in the upper port. It was manufactured by Krasnaya Zvezda Corporation based on a capillary porous structure (CPS) [28]. The diaphragm could move from a radius of 43 to 25 cm from pulse to pulse. The design of the lithium diaphragm is shown in Fig. 9. It consisted of a molybdenum cylinder surrounded which a molybdenum mesh. This mesh was connected with a side lithium tank. Using a heater located inside the molybdenum cylinder, lithium was heated to melting and, because of capillary forces, lithium impregnated the molybdenum mesh. To prevent the possibility of lithium leakage due to gravity, the diaphragm was tilted so that the lithium container was located below the cylindrical capillary structure. In the central region, the molybdenum mesh was protected from the heat flux by a layer of porous tungsten felt. The diaphragm temperature was controlled by three thermocouples as shown in Fig. 10. Thermocouple T2 was located on the back side in the lithium layer opposite the point where the diaphragm touched the magnetic surface, where the maximum interaction with the plasma occurred. In the T-10 geometry, this location was shifted 50 mm inward along a major radius from the center of the rail diaphragm. Thermocouple T1 was additionally shifted inward by 10 cm. Thermocouple T3 controlled the temperature of the lithium container. Before the lithium container was mounted into the T-10 chamber, about 50 g of lithium was filled into it, and its consumption was controlled by the duration of the plateau at the temperature during the transition of lithium from liquid to solid state. The lithium diaphragm was mounted on a movable support isolated from the chamber. In the experiments, the support was connected to the chamber through a 1 Ω resistance to measure the current to the diaphragm.

Fig. 8.
figure 8

Scheme of the location of tungsten and lithium diaphragms in the T-10.

Fig. 9.
figure 9

Lithium diaphragm design. (1) Lithium layer impregnating the molybdenum mesh; (2) molybdenum mesh; (3) container with lithium; (4) layer of tungsten felt; (5) molybdenum tube with heater; (6) mounting support.

Fig. 10.
figure 10

Thermocouple location on the lithium diaphragm.

To observe the lines of impurities from all three diaphragms, the upper, lower and equatorial ports of the T-10 were equipped with optical windows through which the brightness of the lines in the visible region was measured. The radial distribution Zeff was measured with the continuum in the visible region. The spatial distributions of the spectral lines of deuterium, lithium and impurities were also measured. The glow distribution in the diaphragm port was observed tangentially using a MotionPro Y4-S1 color video camera. The impurity concentration and ion temperature were measured with CXRS diagnostics in port away from the diaphragm port in the toroidal direction. The electron temperature profile was measured using electron cyclotron radiation, and the absolute value was determined by soft X-ray radiation using a semiconductor proportional detector. Radial radiation loss profiles were measured using multichannel pyroelectric and semiconductor AXUV detectors. Density profiles were measured with a 16-channel interferometer.

3.2 Preparing a Lithium Diaphragm for Operation after Mounting in the T-10

After fabrication, the lithium capillary structure of the diaphragm was coated with stainless foil from the interaction with the atmosphere. This foil was removed immediately before installing the CPS into the tokamak. Then the T-10 chamber was pumped to a high vacuum. As a result, lithium was exposed to the atmosphere for several hours. Therefore, to clean the lithium surface, long heating was carried out in a vacuum at a temperature of 200–300°C. The heating duration was determined by the pressure in the chamber reaching the level before the start of heating. After heating in vacuum, to clean the chamber, a Taylor discharge in deuterium was switched on with a frequency of 50 Hz, a current of 5 kA, and a magnetic field of 0.05 T. The diaphragm was introduced into the discharge and heated to 550°C by the discharge. Typically, in order to achieve this temperature, it was necessary to introduce the diaphragm into the discharge to a radius of 27 cm, however, the temperature began to rise after 5–10 min, and it was necessary to reduce the depth of insertion of the diaphragm. This was probably due to surface purification, which resulted in a decrease in the degree of blackness and, accordingly, heat loss through radiation. At the same time, a color video camera recorded the growth of the red glow in the diaphragm region, determined by the 6105 Å line of the LiI atom. After this, the diaphragm was removed from the discharge and was ready for use in the experiment.

3.3 Experiments with the Location of a Lithium Diaphragm in the SOL Region

Experiments with purely tungsten diaphragms showed a high level of light and heavy impurities in the discharge [29]. In this regard, the possibility of plasma purification under the conditions of tungsten diaphragms, similar to the experiments with graphite diaphragms described in Subsection 2.3, was studied. The same as in previous experiments, the diaphragm could be used to deposit lithium before the discharge. However, in new experiments, the increased reliability of the fastening made it possible to conduct pulses with a diaphragm introduced into the discharge. In this case, lithium was heated above the melting point (typically 250–350°C) and the diaphragm was introduced into the near-wall region of the Scrape Off Layer (SOL) at radii from 33 to 31 cm. Thus, the lithium diaphragm could affect the discharge conditions in two ways. The first way is a fairly long exposure of the heated diaphragm before the discharge, which led to the deposition of a thin lithium film on the chamber elements due to the evaporation of lithium in the region where the diaphragm is located. The flow of lithium sputtered from the capillary structure during the discharge also led to covering of the walls of the chamber and tungsten diaphragms with the lithium. This lithium gettering method corresponded to the conditions of previous experiments at T-10 [16]. The second way could be connected with direct capture of impurities by the diaphragm introduced into the SOL. Unfortunately, these two methods worked together in real experimental conditions. In this case, the amount of sputtered lithium on the tungsten diaphragms grew from discharge to discharge and from day to day, and the influx of lithium into the discharge began to occur not only from the lithium diaphragm, but also from the tungsten diaphragms. Thus, the impact of the lithium diaphragm on the discharge characteristics was determined by three parameters: the heating temperature and the set radius, as well as the integral amount of lithium already deposited on the walls of the chamber and tungsten diaphragms. The evolution of the glow distribution in the diaphragm region during the experiments, observed tangentially with a video camera, is presented in Fig. 11. Three frames are shown. The frame on the left is at the beginning of a series of experiments (67 887). The middle frame (67 910) is in the middle, and the frame on the right (67 952) is at the end of the experiments. The red color corresponds to the emission of the Dα 6561 Å and LiI 6105 Å lines. Green color corresponds to the emission of the 5485 Å LiII line. All frames were taken with the same exposure. A lithium diaphragm was located at a radius of 32 cm in the frame region. A tungsten rail diaphragm at a radius of 30 cm was located in the lower region. In the left-hand part of the first two frames, the glow of a circular tungsten diaphragm at a radius of 33 cm is visible, which is caused by the fact that the column was displaced in these discharges 2 cm inward along the major radius, in contrast to the third frame, where the offset was zero. We can conclude from the first frame that the glow in the red region is determined by the intensity of the 6105 Å line of lithium and not of deuterium, since the main deuterium flow goes to the lower rail diaphragm, which is almost invisible in the first frame. It can be seen that at first the main glow in the red region comes from the top of the lithium diaphragm and from the inside of the circular diaphragm. In the middle frame, the tungsten rail diaphragm at the bottom begins to glow. In the right-hand panel, the LiI emission from it becomes predominant. In addition, the glow of the lithium diaphragm in the third frame decreases due to the depletion of the lithium supply. This evolution of the lithium glow distribution indicated a long-term lithium accumulation on tungsten diaphragms. The first image also clearly shows the green luminescence filament of the LiII line stretched along the magnetic field lines.

Fig. 11.
figure 11

Glow of the lithium diaphragm. On the left: at the beginning of the campaign; middle: in the middle; on the right: at the end of the campaign.

Experiments were begun in discharges with tungsten diaphragms and high concentrations of light and heavy impurities [29]. Under these conditions, introducing a lithium diaphragm into the SOL resulted in significant plasma purification and increased lithium flows. It should be noted that the level of lithium lines intensity also depended on the temperature of the lithium diaphragm. When the temperature of the lithium diaphragm before the discharge increased from 280 to 370°C, the emission of the LiI line in the equatorial plane of the diaphragm pipe increased by 2.5 times at a constant input radius that indicated an increase of the lithium flow. This fact is associated with an increase in the lithium sputtering coefficient with temperature [30]. The temperature dependence of lithium flows was studied by the intensity of the diaphragm glow in the red region of the spectrum recorded by a color video camera in a Taylor discharge. Figure 12 shows the dependence of the glow in the red region on the temperature of the diaphragm. This glow could be caused by the intensity of the Dα 6561 Å and LiI 6105 Å lines. However, when describing Fig. 11, it was shown that the brightness of the lithium line dominates. In addition, in the experiment, a cold lithium diaphragm was immediately introduced to a constant radius of 27 cm and its temperature increased in time due to heating by the plasma. Thus, the deuterium flow to the diaphragm and the Dα 6561 Å emission were constant in time, and if this line could make a contribution at low temperatures, then with an increase in the lithium glow by orders of magnitude, this contribution can be neglected. Since the plasma flow to the diaphragm was constant, the only changing parameter was the temperature of the diaphragm. It can be seen that the glow grows exponentially with an increment of 23°C, which is close to the theoretical and experimental values of 22.6°C for this temperature range [31].

Fig. 12.
figure 12

Experimental dependence of the glow of a diaphragm in a Taylor discharge on its temperature; (solid line) exponential with an increment of 23°C.

Introducing the diaphragm into the discharge led to a decrease in the level of impurities during the experimental day. During the experimental campaign, as lithium accumulated in the chamber, the level of impurities with the diaphragm pulled out corresponded to the increasingly pure plasma. In this case, introducing a diaphragm led to only a slight improvement in the discharge. The highest plasma purification effect was achieved immediately after finishing the work with lithium with the diaphragm pulled out. Further, the level of impurities gradually increased, but a significant lithiation effect persisted for two weeks before opening the chamber to the atmosphere. Figure 13 shows changes in the main discharge characteristics with increasing lithium flow. The lithium flow was estimated by the emission of the LiI line in the equatorial plane of the diaphragm port. The data were obtained in the ohmic discharges with a current Ip = 220 kA, a magnetic field Bt = 2.4 T and an average density of (2–2.5) × 1019 m–3. It can be seen from Fig. 13 that there is good correlation of all the given parameters with the intensity of the LiI line. In Fig. 14, the data on the intensity of the OII line shown in the bottom plot of Fig. 13 were re-plotted in the form of reciprocal glow values. The direct proportionality of the decrease in the oxygen line brightness with increasing lithium flows is clearly visible. The intensity of oxygen and carbon impurity lines decreases by 10–15 times, which leads to a decrease in the effective plasma charge shown in Fig. 15. At the same time, the bolometric losses shown in Fig. 13 at the periphery of the plasma, associated with light impurities and charge exchange losses, drop only in four times. It should be noted that according to CXRS diagnostic measurements, the lithium concentration in the center of the plasma always remained less than 1%. Crucially, lithium flows have the strongest impact on the tungsten content in the inner regions of the column, which determines the central radiative loss measured by the AXUV semiconductor detectors. Figure 16 shows the dependences of the signals of the central AXUV chord and the intensity of the WI line in the equatorial plane of the cross section of the diaphragm location on the intensity of the LiI line. Despite the fact that the radiation losses in the center associated with the W concentration in the plasma [32] decreased by a factor of 30, the intensity of the WI line, which characterizes the tungsten inflow, decreased only by a factor of 3. This small decrease may be due to less W sputtering in the pure plasma compared to the case of plasma heavily contaminated with carbon and oxygen. An additional 10-fold decrease in the W concentration in the center is explained by a decrease in the neoclassical W accumulation in the pure plasma [32], i.e., the lithium flow during the discharge forms no protective film on the W diaphragm. However, lithium gettering before the discharge slightly reduces the tungsten inflow in the first discharge, as shown in Fig. 17. But after five discharges the inflow returns to its previous level. The question of the possibility of blocking the tungsten inflow with a lithium film was one of the most important ones during experiments with lithium. Experimentally, it was not possible to reduce the tungsten line and losses from the center to zero despite the increase in lithium deposition on the diaphragms. This is probably due to the fact that tungsten sputtering occurs at the very top of the diaphragm, from which the lithium film is cleaned off by plasma. The glow of lithium is associated with the lithium deposition on parts of the diaphragm more distant from the plasma. It is not possible to achieve a significant reduction in the tungsten inflow even in the case of the first discharges after lithium gettering in the region of tungsten diaphragms. It is possible that the lithium layer is cleared off already in the first phase of the discharge. Summarizing the experimental results using a lithium diaphragm in the SOL of the plasma, it can be stated that its use under these conditions leads to a significant reduction in light and heavy impurities in the plasma. Moreover, this decrease increases with increasing lithium flows in the plasma. In this case, the effective charge of the plasma approaches unity. It is extremely important that, despite the strong increase in lithium flows as a result of the introduction of the diaphragm and the long-term accumulation of lithium in the chamber, the lithium concentration in the central regions, measured by CXRS diagnostics, did not exceed 1%. This effect was specially studied in [15]. It was shown there that the low efficiency of lithium penetration into the central regions of the plasma is determined by the high sorption of lithium by the circular and rail diaphragms. At the same time, the diffusion of lithium does not differ qualitatively from other impurities in the region of closed surfaces.

Fig. 13.
figure 13

Changes in discharge characteristics in an experimental campaign depending on the intensity of the LiI line.

Fig. 14.
figure 14

Dependence of the reciprocal of the intensity of the OII line on the intensity of the LiI line.

Fig. 15.
figure 15

Dependence of the effective plasma charge on the intensity of the LiI line.

Fig. 16.
figure 16

Dependences of the signals of the AXUV central chord and the intensity of the WI line in the cross section of the diaphragm location on the intensity of the LiI line.

Fig. 17.
figure 17

Change in the intensity of the WI line in the first pulses after lithium deposition.

Similar experiments with the introduction of lithium into SOL of the plasma were carried out at a number of tokamaks. Experiments almost identical to those at T-10 were carried out at the Italian FTU facility [33]. The CPS-based lithium diaphragm was also manufactured by Krasnaya Zvezda Corporation. It was introduced into SOL of the plasma at a distance of 2 cm from the last closed surface. The main result was the complete suppression of the inflow of light and heavy impurities, which made it possible to obtain discharges with a density by 30% higher. The same as at the T-10, the lithium concentration in the plasma was low and the effect of reducing impurities was maintained when the diaphragm was removed. Experiments on the introduction of a plate with flowing liquid lithium into SOL of the plasma were carried out at the EAST facility [34]. The plate was inserted to a distance of 3 cm from the separatrix. At the same time, a decrease in recycling and a decrease in light impurities in the plasma were also observed. The effective charge decreased from 2.1 to 1.6. The lithium concentration in the center was not measured. However, insertion of the plate increased the inflow of heavy impurities from the lithium substrate and led to the ejection of lithium droplets. Plates coated with lithium and CPS were used at the HT-7 facility [35]. Lithium elements were introduced to a distance of 1 cm from the last closed surface. In these experiments, multiple ejections of lithium droplets were observed, but their number was less when using CPS.

In experiments at DIII-D [36] and T-10 [37], this required large injections of powder that led to a significant lithium concentration in the central regions. At the NSTX facility [38], aerosol was injected using a special device based on the piezoelectric effect. It was noted that it was possible to inject a lithium flow five times higher than the deuterium inflow. However, no data are provided on the lithium concentration in the plasma. At the EAST facility [39], the lithium dust was injected at a level of 13.4 mg/s in a series of long-term discharges, which were preceded by the deposition of 25 g of lithium on the chamber walls. This injection made it possible to maintain the sorption capacity of lithium in a series of discharges, ensuring a reduction in recycling and the inflow of light and heavy impurities. However, the work does not report the lithium concentration in the plasma.

Thus, we can conclude that lithium capillary-porous structures manufactured by Krasnaya Zvezda Corporation also showed the high efficiency in reducing recycling and impurities in experiments at the FTU tokamak [33]. It is important that the conclusion about the main role of lithium accumulated in the chamber was confirmed. It should also be noted that there is no splashing when the diaphragm is sectionalized, even at high heat fluxes. Experiments at the EAST [34] and HT-7 [35] facilities with capillary-porous structures of a different design and plates with flowing lithium led to a decrease in recycling and impurities, but were accompanied by a significant ejection of lithium droplets. Injection of lithium dust at the DIII-D [36], T-10 [37], NSTX [38] and EAST [39] facilities also reduced recycling and impurities, but led to a significant increase in the lithium concentration in the plasma at DIII-D and T-10.

3.4 Experiments with Introduction a Lithium Diaphragm into the Region of Closed Magnetic Surfaces

A decrease in recycling and impurity levels associated with lithium accumulation in the chamber was observed at the beginning of the first experimental campaign in the spring of 2016, when the diaphragm was introduced only into the SOL region of the plasma. The situation changed in the following experimental campaigns due to the deterioration of vacuum conditions from 2 × 10–5 to 7 × 10–5 Pa. No long-term integral effect of lithiation was observed in these campaigns. The very next day after cleaning with a Taylor discharge, the impurity composition of the plasma returned to the initial level and the integral effect of lithium accumulation during the campaign was not observed. Plasma purification when introducing a lithium diaphragm into the SOL of the plasma and the long-term improvement shown previously suggested that there may be two mechanisms for reducing impurities. The first mechanism could be the capture of impurity flows into the SOL by the diaphragm itself as it is introduced. The second mechanism could be determined by the long-term lithium accumulation in the chamber and be associated with sorption of impurities by the lithium layer coating the chamber and diaphragm. To study these processes, an experiment was carried out with the introduction of a diaphragm into the plasma from discharge to discharge and its subsequent removal. In this experiment, the plasma was limited only by one circular tungsten diaphragm with a radius of 33 cm. In this case, the location of the lithium diaphragm at a radius of 32 cm corresponded to its insertion by 1 cm into the inner region of closed magnetic surfaces. The results of this experiment are shown in Fig. 18. The discharge parameters were constant in the series: the discharge current Ip = 220 kA, toroidal field Bt = 2.4 T, and average density 〈Ne〉 = (2–2.5) × 1019 m–3. A lithium diaphragm heated to 300°C was introduced into the plasma from discharge to discharge from the fully pulled out position. The abscissa axis in all plots shows the pulse number in the series. The distance of the lithium diaphragm from the last closed magnetic surface determined by the circular tungsten diaphragm is shown in Fig. 18a. It is clear from Fig. 18a that it was gradually withdrawn after its introduction by 1 cm into the closed surfaces. Figure 18b shows the intensity of the lithium line LiII 5485 Å in the equatorial plane of the diaphragm port. In this case, the emission is recorded from the inside of the circular tungsten diaphragm shown in the middle and left-hand photographs of Fig. 11. Figure 18c also shows the intensity of the lithium line LiII in a toroidal cross section 90° away from the diaphragm. It can be seen that both signals change from discharge to discharge similarly. This indicates that the intensity of this ion is well averaged along the torus and, as shown in [15], the intensity of the LiII 5485 Å line far from the diaphragm characterizes the complete lithium inflow into the plasma. Inserting a diaphragm does not lead to a sharp increase in lithium flows, and its removal does not reduce this flow. On the contrary, the intensity of the lithium lines gradually increases after the diaphragm is inserted, reaches its maximum value by the fifth pulse and remains at a high level even with the diaphragm removed in the 8th pulse. Thus, it can be concluded that the lithium flows mainly reflect the integral effect of lithium accumulation on the diaphragms and the wall during the experiment. It should also be noted that in addition to lithium sputtering in each discharge, one should also take into account lithium sputtering from a lithium diaphragm heated to 300°C for 15 min between pulses, which also increased the integral lithium deposition from discharge to discharge.

Fig. 18.
figure 18

Results of an experiment with the introduction of a lithium diaphragm into the plasma from pulse to pulse and its subsequent removal.

It should also be noted that the diaphragm position has a small effect on the lithium flow. Thus, the flow increased from the fourth to the fifth pulse, when the diaphragm was inserted by 0 to –1 cm, and it decreased in the sixth pulse when it was withdrawn to 0 cm. However, the main effect is determined by the integral deposition during the series, since the flow remains maximum at the complete withdrawal at the eighth pulse. Figures 18d–18g shows changes in the loop voltage, integral radiation losses, and intensity of the CIII and OII lines. It can be clearly seen that all these characteristics characterizing the level of impurities gradually decrease during pulses with the diaphragm introduced, reach a minimum at the last discharge with the diaphragm introduced and begin to increase slightly after its removal. That is, plasma purification from impurities is determined by the completely integral effect of lithium accumulation in the chamber. Asterisks in Fig. 18 also show data from a control pulse with the same parameters performed next day after overnight cleaning with a Taylor discharge. It can be seen that although the intensity of the lithium line (Figs. 18b, 18c) decreased by less than half with respect to the last pulse in the previous series, all characteristics completely returned to the state before the diaphragm was introduced. This may be due to lithium poisoning during night training, i.e., the effect of plasma purification is determined not only by the integral lithium deposition in the chamber and lithium flows, but also by the efficiency of sorption of impurities by lithium. The experiment, as well as previous series with a lithium diaphragm in SOL, showed that its introduction helps plasma purification from impurities.

The following experiments with deep insertion of the diaphragm took place under conditions where the plasma was limited by both circular and rail tungsten diaphragms. Several series of experiments were carried out with deep insertion of the diaphragm, and a number of specific features of the operation of the lithium diaphragm in contact with the main hot plasma were discovered.

Figure 19 shows a comparison of changes in plasma characteristics in three series of experiments with deep insertion of the diaphragm. The data were obtained in a discharges with a current Ip = 250 kA, a magnetic field Bt = 2.4 T, and an average density of (2–2.5) × 1019 m–3. The top plots show the distances from the edge of the lithium diaphragm to the last closed surface of the plasma defined by the rail diaphragm (30 cm). In the first series, the diaphragm was gradually introduced from the removed state to a radius of 29 by 1 cm inside the closed magnetic surfaces (Δr = ‒1 cm). In the second series, it was introduced from 42 to 30 cm, and in the third series, from 34 to 30 cm. As can be seen in the plots of the second row, insertion of the diaphragm is naturally accompanied by a sharp increase in the intensity of the LiII line in the diaphragm cross section (in the first series, spectroscopy did not function). However, changes in the plasma characteristics were different in these series. It can be seen that in the second series, there is a gradual decrease in the loop voltage, bolometric losses and carbon flows, i.e., plasma purification occurs in accordance with the lithium accumulation similar to previous experiments with gettering and insertion of a diaphragm into SOL of the plasma. However, in the first and third series, there is a strong increase in bolometric losses and loop voltage, although carbon flows decrease. In this case, radiation losses are maximum in the central regions of the column, which is typical for tungsten. Moreover, in the first series, the increase in losses is especially large, and there are indications in the last pulse of the series that an electron temperature profile with a dip in the center is formed at the end of the pulse. Figure 20 shows the evolution of the radial bolometric loss profiles. It is clearly seen that from 440 to 500 ms there is an increase in losses while the radial profile is maintained, then starting from 500 ms, the growth of the periphery is terminated and a predominant increase in losses in the center begins to a record radiation power density of 0.52 W/cm3 with an ohmic input of 0.6 W/cm3, which leads to a drop in the electron temperature at the center as shown in Fig. 21. Unfortunately, the electron temperature profile was not measured using EC radiation in this pulse. However, the temperature profile with a dip is confirmed by measurements of soft X-ray radiation in the 3 keV region. Figure 22 shows the time variation of the radial profile of the soft X-ray radiation. It can be seen that an X-ray profile with a dip is formed at 735 ms. Since the density profile always remains peaked, and the tungsten impurity concentration is maximum in the center, so the drop in X-ray radiation in center clearly indicates a hollow temperature profile. The strong increase in radiation losses from the center when a lithium diaphragm is introduced in the first and third series of experiments can be explained by sputtering of the tungsten capillary structure. This may be associated with a poor lithium coating of the tungsten felt in the first and third series. The thing is that the first series was carried out immediately after mounting a diaphragm in the tokamak chamber. However, it was probably not well prepared, and the lithium did not covered the entire surface of the tungsten. The third series, on the contrary, took place at the end of the campaign, when almost all the lithium was consumed, which confirms a significant decrease in the luminescence of the LiII line in this series. This also resulted in poor lithium impregnation. At the same time, the second series was carried out in the middle of the campaign, when the diaphragm, on the one hand, was well prepared and, on the other hand, there was quite a lot of lithium. Thus, an important conclusion can be drawn that the capillary lithium structure can work well at high heat fluxes from the plasma, but the tungsten structure should be completely impregnated with lithium. Heat fluxes were estimated on the basis of calculations using known plasma parameters at a radius of 30 cm [15] and from thermocouple data. Figure 23 shows a typical heating time course of thermocouple T2 located on the back side of the lithium diaphragm and opposite the place of the maximum interaction with the plasma. It can be seen that the temperature reaches a maximum 20 s after the pulse and then drops to a stationary value until the next pulse. Since this time is much longer than the discharge duration, the temperature is averaged over the diaphragm within 20 s. Therefore, the temperature increase across thermocouple T2 should be proportional to the total energy transferred by the plasma per discharge. Figure 24 shows the dependence of the maximum increase in heating of thermocouple T2 for a series of 30-cm diaphragm insertion. It can be seen that heating is described by an exponential with an increment of 2 cm.

Fig. 19.
figure 19

Comparison of plasma characteristics in three series of experiments with deep insertion of the diaphragm.

Fig. 20.
figure 20

Time evolution of radial profiles of bolometric losses in pulse 71474.

Fig. 21.
figure 21

Time evolution of the central electron temperature in pulses 71473 and 71474.

Fig. 22.
figure 22

Radial distributions of SXR intensity for two times of pulse 71474.

Fig. 23.
figure 23

Typical time course of the heating discharge of the thermocouple T2 located on the reverse side of a lithium diaphragm.

Fig. 24.
figure 24

Dependence of the maximum heating increase of thermocouple T2 on the distance to the last closed surface for a series of the diaphragm insertion of 30 cm; (squares) temperature increase values, (line) exponential with an increment of 2 cm.

However, the longitudinal plasma heat fluxes were carried out according to the formula Q = 7Te × 0.5Csne, where Cs is the ion-sound velocity. Thus, they should have grown with an increment of about 0.86 cm, since the decrease in the density in the initial plasma occurs with an increment of 1.2 cm, and the temperature with an increment of 3 cm [15]. The discrepancy may be explained by the fact that the introduction of the lithium diaphragm itself further reduces the plasma parameters in SOL. The decrease in the heat inflow upon introduction of the diaphragm may be also associated with its cooling by lithium evaporation, which was observed in experiments at FTU [24]. In addition, it is not excluded that the introduction of a diaphragm causes the appearance of an island structure of magnetic surfaces observed in experiments at the T-11M [40]. The island reduces the thermal load on the diaphragm. Calculations using the given formula with the parameters of undisturbed plasma give a longitudinal heat flux at a radius of 30 cm of 6 MW/cm2. Estimates from a three-dimensional model in the ANSYS program with a temperature increase of T2 of 55°C give a maximum flux of 2.5 MW/cm2. In other series, the maximum temperature increase over thermocouple T2 was 80°C and, accordingly, the longitudinal heat flux at a radius of 30 cm can be estimated as 3.6 MW/cm2. The ANSYS software used a complete 3D model of the lithium diaphragm with molybdenum mesh and tungsten felt and the actual position of thermocouple T2. The model included a longitudinal energy flow from the plasma, which decreases exponentially with the increasing minor radius. The model fully described the temporal behavior of thermocouple T2 when the diaphragm was located at a radius of 32 cm. The model did not take into account the decrease in the energy flux due to lithium evaporation, which is significant when introducing into the region of closed surfaces. However, it adequately related the integral heat flux to the diaphragm with the readings of thermocouple T2.

These results demonstrate the first feature of experiments with a deep insertion, namely, the need to control good impregnation of the capillary structure with lithium in order to avoid the material of the capillary structure entering the discharge, which was implemented in the second series of experiments. Such control can be done by the reflection coefficient. However, this series of experiments also revealed a second feature of operation with a deep input. It occurred that under conditions of good lithium impregnation of the capillary structure, when the diaphragm was inserted to a radius of 30 cm, the discharges did not reach the end and resulted in disruptions. The analysis showed that the disruptions occurred due to massive injections of lithium droplets from the diaphragm. Fast camera observations showed that the droplet injection occurs from the edge of the diaphragm, while the droplet injection occurs from the opposite edge when the direction of the toroidal magnetic field changes. In the third series of experiments, the current from the diaphragm to the chamber was measured. It was shown that the current passes from the plasma to the diaphragm and its value was up to 11 A. Figures 25 and 26 show photographs of the diaphragm at the times of the beginning of the droplet injection and after 1 ms for two magnetic field directions. They also show the current flow pattern and the resulting Ampere force. It is clearly seen that the drops fly out from the edge of the diaphragm where the Ampere force [J × B] is directed, i.e., the Ampere force acting along the diaphragm created the excess pressure of liquid lithium and led to the extrusion of lithium at the corresponding edge of the diaphragm when the pressure exceeded the capillary forces. The emission of droplets occurred during the stationary part of the current in the final stage of the discharge from 600 to 900 ms. It should be noted that the diaphragm temperature between pulses was maintained at 335°C in these experiments. The temperature increase after the pulse was 69°C according to the readings of the second thermocouple. Calculations using the ANSYS program showed that heating of the edge of the diaphragm facing the plasma is 2.85 times greater than the increase in the temperature of thermocouple T2, i.e., the temperature of the diaphragm at the end of the pulse could reach 530°C. Since the capillary forces fall with increasing temperature, it is likely that the droplet ejection occurred when during the discharge, the temperature was reached, when the capillary forces became less than the lithium pressure due to the Ampere force. Estimates [41] show that the surface tension force decreases with a temperature increase from 330 to 530°C by only 8%. Therefore, one can assume that initially the process was on the verge of stability. This is also confirmed by the fact that the time of the droplet ejection varied stochastically from 600 to 900 ms from discharge to discharge. To prevent the effect of the droplet ejection, it is necessary either to keep the temperature level significantly below 530°C, or to sectionalize the capillary structure in the direction of the Ampere force. This can be done by dividing a long capillary structure into separate independent sections in the direction of the Ampere force. Since the pressure gain at the edges increases with the length of the structure, this can reduce the maximum lithium pressure at the ends and prevent the droplet ejection. For example, the lithium diaphragm at the FTU tokamak [33] had a structure close to that of T‑10 and the poloidal dimensions. However, it was divided into three independent sections and droplet exit was never observed, although the discharge duration and diaphragm temperature were higher than those at the T-10.

Fig. 25.
figure 25

Photo of the diaphragm at the time of the beginning of the droplet ejection and after 1 ms when the toroidal magnetic field is directed clockwise as observed from above.

Fig. 26.
figure 26

Photo of the diaphragm at the time of the beginning of the droplet ejection and after 1 ms when the toroidal magnetic field is directed counterclockwise as observed from above.

Experiments with a deep insertion of the diaphragm also showed a strong dependence of lithium sputtering on the temperature of the plasma interacting with the diaphragm. Thus, in one of the series of experiments, to control plasma recycling, the gas inlet was switched off at 850 ms of the discharge. Figure 27 shows the results of such a switch for different radial positions of the diaphragm. It can be seen that when introducing a diaphragm up to 31 cm, a rapid density decay occurs after turning off the gas inlet as shown in the upper plots. Moreover, as the diaphragm is inserted, the intensity of the LiII line does not grow much when the gas influx is turned off. However, the situation changes when the diaphragm is inserted at 30 cm into the hot plasma region. It can be seen in the right-hand plots that when the gas influx is turned off, the sharp density decrease is terminated, and, at the same time, a strong increase in the intensity of the LiII line is observed. This allows the conclusion that the decrease in the flow of working gas is compensated by a strong lithium inflow when the gas influx is turned off. The increase in lithium sputtering may be associated with an increase in the temperature of the plasma periphery when the valve is turned off. An increase in the lithium sputtering coefficient with plasma temperatures in the range up to 50 eV was noted in [30]. This conclusion is confirmed by an experiment in a discharge with the gas influx turned on and off shown in Fig. 28. It can be seen that the luminescence of lithium and tungsten decreases when the gas influx is turned on and increases when it is turned off. Interestingly, the intensity of the carbon and oxygen lines changes in the opposite way and this may be due to the fact that their inflows depend more on the density than on the plasma temperature. Thus, lithium sputtering increases with the plasma temperature, the same as for tungsten.

Fig. 27.
figure 27

Changes in plasma characteristics after turning off the gas influx at different radial positions of the diaphragm.

Fig. 28.
figure 28

Changes in discharge characteristics when gas injection is turned on and off.

This effect is largely associated with a strong increase in the lithium concentration in the central regions of the plasma when a diaphragm is introduced into the region of closed magnetic surfaces. In [15], the balance of lithium in a discharge with a lithium diaphragm at a radius of 30 cm was analyzed in detail. It was shown that the plasma almost entirely consists of lithium ions in these discharges. Moreover, the experimental data are fully confirmed by the simulation with the coefficients used for the diffusion of other impurities, i.e., the diffusion of lithium does not differ from the diffusion of other impurities. In this work, a sharp increase in the concentration of lithium in the discharge is explained by the exponential increase in the lithium inflow and the decrease in the screening of these flows by the main rail diaphragm.

To analyze the origin of the sharp increase in the lithium inflow in experiments with a deep insertion of the diaphragm, we consider the evolution of the density formation shown in Fig. 29. It shows the average densities, the control voltage on the gas valve, and the intensity of the LiII line for several pulses of the second series presented in Fig. 19. Pulse 71661 was carried out before the diaphragm was inserted. The rest pulses were carried out with the diaphragm inserted to a radius of 30 cm. The gas valve was regulated by the feedback to maintain the average density at the level of 2 × 1019 m–3, which is implemented in a pulse without an inserted diaphragm. In this case, the gas valve is only slightly open. In the first pulse with the diaphragm inserted, the density course and gas inlet at the beginning of the discharge coincide, however, at 500 ms the lithium inflow begins to increase and the valve closes, as the density increases above the program. In pulse 71 663, the density increases more slowly, despite the larger valve opening. In pulse 71 666, the valve is opened completely, but its influx is not sufficient to increase the density. This is due to a strong drop in recycling which results from the lithium accumulation in the chamber during previous pulses. However, at 400 ms, the lithium inflow associated with heating of lithium begins, and after this the density begins to increase because of the lithium inflow. Moreover, Fig. 30 shows the dependence of the density in discharge 71666 on the intensity of the lithium line. An almost linear relationship is visible starting from a density of 0.5 × 1019 m–3, this shows that the density is completely determined by the lithium influx. These data are in complete agreement with the statement about the dominance of lithium in the plasma [15]. The main origin is the insufficient gas injection performance. As a result, the gas injection plays no role in the particle balance due to the sharply decreased recycling and the periphery of the plasma remains hot, which leads to strong lithium sputtering caused by the increase in the temperature of the plasma incident on the diaphragm as shown above. This effect further increases the already high evaporation level associated with heating of the diaphragm in the discharge. That is, to purify the plasma and reduce the lithium inflow, it is necessary, firstly, to reduce the temperature of the plasma periphery by the intense gas injection. If an intense deuterium flow does not lead to a significant decrease in the temperature of the plasma periphery, then an additional injection of impurities, e.g., neon or argon, is necessary. Secondly, lithium sputtering increases exponentially with its temperature. Therefore, it is necessary to reduce the lithium temperature as much as possible using an effective cooling system.

Fig. 29.
figure 29

Time evolution of the density, voltage on the gas inlet valve, and intensity of the LiI line in a series of discharges with a lithium diaphragm at a radius of 30 cm.

Fig. 30.
figure 30

Dependence of the evolution of the average density on the intensity of the LiII line for discharge 71666.

In a series of experiments with the insertion of a diaphragm shown in the right-hand plots of Fig. 19, the current to the diaphragm was measured using a shunt. The results of these measurements are shown in the bottom plot of Fig. 31. It can be seen that after the current reaches a stationary value of 250 kA for 400 ms of the discharge, the current flows from the plasma to the diaphragm even at a radius of 32 cm. This corresponds to the positive potential of the plasma interacting with the diaphragm. The current increases with the insertion of the diaphragm from 32 cm (pulse 72 157) to 30 cm (pulse 72 158) and also with an increase in the temperature of the diaphragm before the discharge from 300 (pulses 72 158 and 72 161) to 376 (pulse 72 162) and 390°С (pulse 72 165) and reaches the magnitude of 11 A. It can be seen that the current to the diaphragm at an initial temperature of 300°C increases during the discharge, which correlates with heating of the diaphragm. This assumption is confirmed by a strong increase in the current with increasing initial temperature of the diaphragm. Note that a strong increase in the current occurs only at the beginning of the stationary phase, but then the current increases no longer and reaches saturation. In a similar way, there is a change in the intensity of the LiII line, which characterizes the lithium flows in the discharge. It can be seen that when the diaphragm is entered, the intensity of the LiII line increases in four times at 500 ms, but then it reaches saturation. In the next pulse at the same initial temperature, saturation is achieved earlier by 500 ms, and, finally, by 450 ms with an increase in the initial temperature of the diaphragm. It is extremely important that in all cases the intensity of the LiII line reaches the same maximum level. The saturation of lithium flows from the diaphragm is also directly confirmed by observations of the glow of the lithium diaphragm using a color camera. These results are shown in Figs. 32 and 33. Figure 32 shows a snapshot of the diaphragm glow observed tangentially by a high-speed color camera. It shows two lines which were used for data reading. Figure 33 shows data taken along the red line. The data along the black line are identical to those along the red line. The brightness reading was carried out in the red region (LiI line 6105 Å) at 740 ms of the discharge, when the intensity of the LiII line in all discharges reaches saturation. The glow in discharge 72158 is not shown, since the exposure was too long and the signals were saturated. It can be seen that the glow in discharges with an inserted diaphragm changes slightly from discharge 72 161 to discharges 72 162 and 72 165, despite the increase in the initial temperature. Since the intensity of the LiI line at constant plasma flows depends exponentially on temperature (Fig. 12), it can be stated that thermal stabilization of the diaphragm temperature by evaporating lithium occurs. This effect was also observed in experiments at the FTU tokamak [33] when the surface temperature reached 450°C. In the experiment at T-10, the maximum temperature can be estimated from measuring the temperature increase of thermocouple T2. The increase was 73°C at the initial temperature of 298°C in pulse 72 158. Calculations using the ANSYS program showed that the temperature increase on the surface of the diaphragm without thermal stabilization should be 2.85 times greater. Thus, the upper estimate of the thermal stabilization temperature is 506°C. This estimate is close to that of 450°C obtained at FTU [33]. Assuming that in pulse 72 158 the maximum temperature was 506°C and lithium flows increase in accordance with the experimental data given in [31] and confirmed in the experiment at T-10 (Fig. 12), it is possible to recover the temperature evolution in pulses of this series. During the recovery, it was assumed that the maximum intensity of the LiI line in pulse 72158 corresponds to 506°C. The time course of the temperature in the remaining pulses was recalculated from the intensity of the LiI line in accordance with the plot shown in Fig. 12. The temperature estimated in this manner is shown in Fig. 34. It is clearly seen that the temperature reaches saturation the earlier, the higher the initial heating temperature before the discharge. It should be noted that the estimate of the maximum temperature in pulse 72158 was made on the assumption that there was no decrease in the heat flux in this pulse due to the thermal stabilization effect. However, it is clear that already at 650 ms the lithium intensity reaches its maximum value, i.e., the effect has taken place and the temperature estimated at 506°C should be overestimated, therefore the estimate of the limit for the beginning of the intense lithium evaporation of 450°C at FTU seems more realistic.

Fig. 31.
figure 31

Change in time of the average density, intensity of the LiII line and current per diaphragm in a series of discharges with a lithium diaphragm at a radius of 30 cm. (red solid curve) Discharge 72157, diaphragm at 32 cm, initial temperature of 300°C; (black dashed curve) discharge 72158, diaphragm at 30 cm, initial temperature of 300°C; (green dash-dotted curve) discharge 72161, diaphragm at 30 cm, initial temperature of 300°C; (purple short dotted curve) discharge 72162, diaphragm at 30 cm, initial temperature of 376°C; (blue short dash-dotted curve) discharge 72165, diaphragm at 30 cm, initial temperature of 390°C.

Fig. 32.
figure 32

Photo of the glow of the diaphragm as viewed tangentially by a high-speed color camera. Two lines, along which the data was read, are shown.

Fig. 33.
figure 33

Glow intensities in the red region, taken for the photo in Fig. 32 along the red line. (red solid curve) Discharge 72157, diaphragm at 32 cm, initial temperature of 300°C; (black dashed curve) discharge 72160, diaphragm at 30 cm, initial temperature of 300°C; (green dash-dotted curve) discharge 72161, diaphragm at 30 cm, initial te-mperature of 300°C; (purple short dotted curve) discharge 72162, diaphragm at 30 cm, initial temperature of 376°C; (blue dashed curve) discharge 72165, diaphragm at 30 cm, initial temperature of 390°C.

Fig. 34.
figure 34

Calculated changes over time in the temperature of the lithium diaphragm in a series of pulses with a diaphragm at a radius of 30 cm. (red circles) Discharge 72158, diaphragm at 30 cm, initial temperature of 300°C; (purple triangles) discharge 72161, diaphragm at 30 cm, initial temperature of 300°C; (blue stars) discharge 72162, diaphragm at 30 cm, initial temperature of 376°C; (black squares) discharge 72165, diaphragm at 30 cm, initial temperature of 390°C.

The effect of thermal stabilization of the diaphragm by intense lithium evaporation is also manifested in a decrease in the temperature increase of thermocouple T2 with an increase in the initial temperature of the diaphragm before the pulse (see Fig. 35). It can be seen that heating of the diaphragm decreases in 1.5 times with an increase in the initial temperature from 298 to 390°C. Thus, discharges with the complete dominance of lithium in the impurity composition were obtained subject to the condition of complete impregnation of the capillary structure with lithium and deep insertion of the diaphragm. This made it possible to estimate the maximum total radiation losses in the case of the purely lithium plasma. Figure 36 shows chord signal profiles of pyroelectric bolometers and AXUV semiconductor detectors for a lithium-dominated discharge. It can be seen that the AXUV signals change little along the chords, while the bolometer signals are skinned. This difference is explained by the fact that AXUV surface, after long operation at T-10, is coated with a sputtered layer and is insensitive to the soft X-ray spectral region, in which main radiation of light impurities of plasma and lithium occurs on the periphery. Therefore, AXUVs are sensitive to radiation from heavy impurities and, in part, to light impurities. AXUV is also insensitive to losses with fast neutrals of charge exchange. At the same time, the bolometer is sensitive to the entire spectrum. Therefore, the difference between the signals of the two diagnostics can provide an upper limit on the radiation losses of purely lithium plasma. However, when subtracting, it should be taken into account that the sputtered layer also weakens the AXUV sensitivity to losses caused by heavy impurities. This attenuation can be obtained in 71474 discharge with tungsten storage shown in Fig. 20. The coefficient found in this way is 0.65. The difference loss profile obtained in this way is shown in Fig. 36, while the upper limit of losses on lithium, taking into account charge exchange, is 56 kW, which is close to the sum of radiation losses of lithium and charge exchange PlossLi + Plosscharge exchange = 56–59 kW obtained in [15]. Another discharge from this series was analyzed in the same manner. In this case, the sum of radiation losses and charge exchange was 55.1 kW. Taking into account that the calculated losses with charge exchange neutrals [15] are 12–15 kW, the radiation losses on lithium, even in discharges with its dominance, are 40–44 kW. This coincides with the calculations [15] and corresponds to only 20% of ohmic heating, and in the case of pure discharges with the low lithium concentration, these losses are negligible. In this case, the lithium inflow was 2.8 × 1020 s–1 [15]. Thus, the “radiation cost” of a single incoming lithium atom is 0.94 keV/atom. This value can serve as an estimate of radiation losses for a known lithium inflow into the plasma. A close value of 0.8–1 keV/atom was estimated experimentally at the T-11M [42]. A theoretical estimate of 1–2 keV/atom is also given there.

Fig. 35.
figure 35

Heating of the diaphragm in the discharge according to thermocouple T2 when the initial temperature changes.

Fig. 36.
figure 36

Chordal signal profiles of pyroelectric bolometers and AXUV semiconductor detectors for lithium-dominated discharge 71673. (rectangles) Pyroelectric bolometer data, (circles) AXUV data (magnified in 1.54 times), (asterisks) difference between them.

4 CHANGES IN DISCHARGE CHARACTERISTICS WHEN WORKING WITH A LITHIUM DIAPHRAGM IN SOL

As shown above, the prolonged operation of the lithium diaphragm in the SOL leads to plasma purification. The characteristics of such discharges are discussed in detail in [15]. In this case, the central electron temperature decreases slightly with a slight increase in the peripheral temperature. However, the energy time shown in Fig. 37 remains unchanged at low densities because of the strong drop in the ohmic power. However, under pure conditions, it is possible to significantly advance into the high-density region of up to 0.8 of the Greenwald limit. At the same time, the energy time increases significantly when operating in Improved Ohmic Confinement (IOC) modes. Figure 38 shows a comparison of the maximum densities obtained when working with graphite, tungsten, and lithium diaphragms for different currents. Data for graphite diaphragms are taken from [43]. They were obtained in Saturated Ohmic Confinement (SOC) modes at high densities. It can be seen that the use of a lithium diaphragm ensures high densities. It is possible to reach the Greenwald limit at a current of 150 kA, but at higher currents this cannot be done. It should be noted that these changes in discharge characteristics are typical for pure discharges, and are not determined by the presence of lithium.

Fig. 37.
figure 37

Comparison of density dependences of energy confinement times when T-10 is operating with carbon, tungsten and lithium diaphragms. (circles) Data with lithiation, (triangles) previous data with graphite diaphragms without lithium, (asterisk) with tungsten diaphragms without lithium.

Fig. 38.
figure 38

Discharge current dependences of the maximum achievable densities in ohmic discharges when T-10 is operating with (blue triangles) carbon diaphragm, (purple stars) tungsten diaphragm, and (red circles) lithium diaphragm.

5 CONCLUSIONS AND RECOMMENDATIONS

Experiments with lithium at the T-10 were carried out with graphite and tungsten diaphragms. In both cases, lithium gettering in the region where the diaphragms are located significantly reduces deuterium recycling and the level of impurities in the discharges. In this case, a significant decrease in recycling is observed five pulses after lithium sputtering. In subsequent pulses, recycling increases, but the lithiation effect persists after 25 discharges. The decrease in the impurity level under good vacuum conditions (typically 2 × 10–5 Pa) is determined by the accumulation of lithium during the experimental campaign and persists 150–300 pulses after the termination of the operation with lithium. In poor vacuum conditions, with a vacuum worse than 7 × 10–5 Pa, an improvement occurs during the experiment and the effect disappears the very next day after preparing the chamber with Taylor discharges. Thus, when working with lithium, it is necessary to ensure a good vacuum level, since otherwise lithium poisoning occurs.

Experiments with lithium purification in a helium glow discharge do not lead to the recovery of the sorbing properties of lithium.

Insertion of a lithium element with a capillary porous structure when it is poorly secured leads to lithium splashing. Therefore, the lithium element should be fixed well.

After a long-term exposure of a capillary-structure-based lithium diaphragm to the atmosphere, it is necessary to purify it from the formed oxides when it is mounted in the facility. In experiments at T-10, purification was carried out by introducing a capillary structure into a Taylor discharge with heating to 550°C. In this case, it is desirable to control the recovery of complete lithium impregnation of the entire surface of the diaphragm, e.g., by the reflection coefficient.

The insertion of a lithium diaphragm into the SOL region leads to an increase in lithium flows and its accumulation in the chamber. As a result, there is a significant decrease in the impurity level and the effective charge approaches unity. The effect of plasma purification from light impurities is associated mainly with the sorption properties of integrally accumulated lithium, similar to the first experiments on lithium deposition. Lithium gettering before the experiment reduces the influx of tungsten for several pulses, but does not prevent its influx to the discharge for a long period. At the same time, the decrease in the tungsten level in the plasma is associated with the lower efficiency of tungsten sputtering with the pure deuterium plasma and a decrease in the neoclassical effect of their accumulation in the center, but not with the formation of a protective lithium layer on the tungsten surface.

Despite the strong effect on impurities, the lithium concentration in the central regions remains less than one percent. This is explained by the strong screening of the lithium flow by the main diaphragms. Thus, the use of lithium diaphragm in the SOL region is efficient for the preparation of pure discharges with low re-cycling.

Experiments with deep insertion into the region of closed magnetic surfaces showed the reliable operation of the lithium capillary structure at longitudinal heat plasma flows up to 3.6 MW/m2.

It has been shown that under the condition of good impregnation with lithium, the insertion of a diaphragm leads to plasma purification from impurities in accordance with the accumulation of lithium deposition similar to experiments with lithium gettering and the insertion of a diaphragm into SOL of the plasma.

However, in a number of experiments, a strong increase in bolometric losses and loop voltage was observed with a decrease in the level of light impurities. In this case, the radiation losses characteristic of tungsten are maximum in the central regions of the column, this leads to the formation of an electron temperature profile with a dip in the center. These phenomena occur due to sputtering of the tungsten base of the capillary structure when it is poorly impregnated with lithium. Thus, an important conclusion can be drawn: the capillary lithium structure can operate effectively at high heat fluxes from the plasma. However, to prevent tungsten sputtering, the capillary structure should be completely impregnated with lithium. In this case, it is desirable to control the recovery of complete lithium impregnation of the entire surface of the diaphragm, e.g., by the reflection coefficient.

Under conditions of good impregnation of the capillary structure with lithium, when the diaphragm was inserted into the hot region of the plasma, in some experiments, the discharges ended in disruptions due to massive injections of lithium droplets from the diaphragm. It is shown that the exit of droplets is associated with the squeezing out of lithium due to Ampere forces when current flows to the diaphragm from the plasma in a magnetic field. To prevent the droplet ejection, it is necessary to keep the temperature level significantly below 450°C, and also to sectionalize the capillary structure in the direction of the Ampere force.

Experiments with deep introduction of the diaphragm also showed a strong dependence of lithium sputtering on the temperature of the plasma interacting with the diaphragm, i.e., to reduce the lithium inflow, it is necessary to reduce the temperature of the plasma periphery by the strong gas influx. If a strong deuterium flow does not lead to a significant decrease in the temperature of the plasma periphery, then an additional injection of impurities, e.g., neon or argon, is necessary. In addition, since the lithium evaporation increases exponentially with increasing its temperature, it is necessary to reduce the lithium temperature as much as possible using an effective cooling system.

Experiments have shown that at high heat fluxes onto the lithium diaphragm, the thermal stabilization of the diaphragm by evaporating lithium occurs. Undoubtedly, the temperature at which this effect occurs should depend on the heat flux. Estimates showed that it was observed at T-10 when the lithium temperature reached about 500°C. However, an estimate of 450°C made in experiments at the FTU tokamak is more realistic.

The insertion of the diaphragm to the closed magnetic surfaces caused strong heating and lithium evaporation. In this case, the maximum gas influx level was insufficient to maintain the density in the discharge and cooling the periphery due to a strong drop in recycling. As a result, the increased lithium flows led to the almost completely lithium plasma.

The discharges obtained at T-10 with the complete dominance of lithium in the impurity composition made it possible to estimate the maximum total radiation losses in the case of purely lithium plasma. It is shown that radiation losses on lithium, even in discharges with its complete dominance, are 40–44 kW, this corresponds to only 20% of ohmic heating, and in the case of pure discharges with a low lithium concentration, they are negligible. In this case, the “radiation cost” of a single incoming lithium atom was 0.94 keV/atom. It should be especially noted that the simulation of lithium diffusion in [15] showed that in a plasma column, it does not differ from that for all other plasma ions. The simulated radiation loss value of 44 kW was obtained, which completely coincides with the experimental estimate, i.e., calculations confirmed the low emissivity of lithium.

In pure discharges obtained when working with lithium, the central electron temperature decreases slightly with a slight increase in the peripheral temperature. However, the energy time remains unchanged at low densities due to the strong drop in the ohmic power. But at the same time, it is possible to significantly advance into the high-density region of up to 0.8 from the Greenwald limit. It should be noted that these changes in discharge characteristics are typical for pure discharges and are not determined by the presence of lithium.

Experiments carried out at T-10 with lithium c-apillary-porous structures showed their high efficiency in obtaining pure plasma with low recycling when placed in SOL of the plasma. In [29], a possible scheme for using such structures in tokamak reactors was proposed, which is based on the concept of a closed circulation loop of lithium flows [16]. In this scheme, a structure deeply inserted into the SOL emits lithium into the plasma, while another, located in the peripheral region of the SOL, is used as a collector of lithium flows. Such a scheme should be effective in obtaining plasma with a low content of impurities and lithium. However, the results of experiments at the T‑10 showed that although it is possible to reduce the tungsten influx from the divertor plates, it cannot be completely eliminated. Therefore, to completely eliminate the influx of heavy impurities, it is necessary to use lithium capillary-porous structures directly as divertor plates. To this end, lithium capillary structures should be developed with efficient cooling at power levels on the order of 10 MW/m2. It is also necessary to experimentally show that lithium flows in a divertor configuration do not lead to a high lithium concentration in the plasma, in contrast to the limiter configuration. Currently, lithium capillary-porous structures with molybdenum meshes for divertor plates for heat fluxes up to 5 MW/m2 have been developed and should be tested on stands. It is planned to modernize them with tungsten felt for high heat fluxes. The developed structures are supposed to be used in experiments at the T-15MD divertor tokamak.