Long-pulse high-performance H-mode plasmas achieved on EAST

Long-pulse, steady-state tokamak research aims at efficient low-cost fusion reactors. Steady-state operation (SSO) in ITER is foreseen to be with fully non-inductive current at Q_DT ≥ 5 for a burning time of ∼3000 s.¹ Long-pulse operation requires good particle control and heat flux exhaust. Steady-state operation requires that all the plasma current must be driven by 100% non-inductive current drive methods. Efficient operation needs high self-driven bootstrap current, which requires high poloidal beta (β_p). High fusion gain (∝β_tτ_EB₀²) requires high toroidal beta (β_t). The high β_p steady-state tokamak scenario has several attractive features in support of ITER and CFETR² steady-state operation: (1) low plasma current (β_p ∝ ⟨P⟩/I_P²) yielding low risk of disruptivity³ and reduced edge localized mode (ELM) challenge;⁴ (2) high bootstrap current fraction $fBS∼εβP$⁠, allowing to minimize the power for external current drive;⁵ (3) high Shafranov shift $Δ∼βP∼ α ∼ dβP/dr$ producing $α$ stabilization of turbulence for excellent energy confinement.⁶ Since the high β_p reactor first proposed by Kikuchi in 1990,⁵ the first high β_p steady-state scenario was developed in JT-60U⁷ and extended to high-performance H_98y2 ∼ 1.7, β_N ∼ 2.7, f_BS ∼ 92% at high Greenwald fraction n_e/n_G ∼ 87%.⁸ KSTAR has achieved ∼90 s H-mode plasma in 2018 with high β_p sustained until ∼50 s with core T_e ∼ 6 keV and core T_i ∼ 2 keV.⁹ Joint DIII-D/experimental advanced superconducting tokamak (EAST) high β_p experiments¹⁰ were started in 2013 (Ref. 11) and are based on previous DIII-D experiments by Politzer.¹² To address the needs for a fusion pilot plan design, joint DIII-D/EAST experiments on DIII-D have demonstrated high normalized beta β_N > 4, poloidal beta β_P > 3, toroidal beta β_T > 3% with q_min > 2, and q₉₅ ≤ 8 sustained for more than six energy confinement times.¹³ This joint steady-state long-pulse initiative aims at extending the physics understanding from DIII-D experiments^14–20 to true steady-state long-pulse demonstrations on EAST (experimental advanced superconducting tokamak)^21–26 with metal walls, a feature relevant for future reactors. Recently, a duration of 310 s H-mode plasma (H_98y2 ∼ 1.3, β_P ∼ 2.5, n_e/n_GW ∼ 0.7, f_BS > 50%) has been achieved with the EAST high β_p scenario.

The rest of this paper is organized with Secs. II–V. Section II describes the progress of EAST on long-pulse fully non-inductive H-mode plasma, and facility enhancements allowing advances in SSO research. Section III gives key physics understanding (energy transport, H&CD at high density, power and particle handling, pedestal characteristics) in long-pulse H-mode plasma. Section IV discuses and presents future research plan toward ITER and CFETR SSO. A summary is given in Sec. V.

EAST is the first fully superconducting tokamak in the world, with the major goal to demonstrate a long-pulse high-performance regime for scientific understanding in support of CFETR and ITER. With features, such as dominant electron heating, zero/low torque during neutral beam injection (NB), and an ITER-like tungsten divertor, EAST can offer unique contributions relevant to ITER and CFETR research on (1) high-performance steady-state operation and (2) suitable plasma exhaust for steady state. Since 2006 with the first plasma on EAST,²⁷ there have been three strategies for EAST to establish the scientific basis for long-pulse high-performance operation in engineering technology and fusion science. The first strategy is to enhance heating and current drive (H&CD) efficiency, relevant fundamental physics understanding and key diagnostics.²⁸ The second strategy is to demonstrate long-pulse (≥100 s) H-mode plasma and develop fully non-inductive high-β scenarios.²² The third strategy and stage is to extend EAST operation regime to demonstrate steady-state high-performance plasmas and deliver relevant physics for ITER and CFETR.²⁴ 

To enhance the plasma performance for fully noninductive steady-state operation, Fig. 1 shows that the total available source power of the auxiliary heating and current drive systems has been gradually increased up to 32 MW including various radio frequency (RF) lower hybrid current drive (LHCD) system with the nominal injected power of LH waves P_LH ∼ 6 MW at the launched frequency of 4.6 GHz and P_LH ∼ 2 MW at 2.45 GHz, ion cyclotron resonant frequency (ICRF) system with P_ICRF ∼ 12 at 37 MHz, electron cyclotron heating (ECH) system with the total source power P_ECH ∼ 4 MW equipped with the four gyrotrons at 140 GHz, and two NBI systems with the total source power P_NBI ∼ 8 MW and beam energy up to 80 keV. Figure 1 also shows the record pulse-lengths of discharges gradually increased with the development of long-pulse fully non-inductive H-mode operation on EAST. The first H-mode plasma with LHCD was obtained in 2010.²⁹ In 2012 with the upper graphite divertor replaced with an ITER-like W mono-block divertor, reproducible H-mode plasma with a duration over 30 s was achieved using predominantly LHCD assisted with ICRF heating.^30,31 In 2016, 1 min H-mode discharge was realized³² and then the duration increased up to 101.2 s in 2017 with zero loop voltage and an ITER-like tungsten divertor, exhibiting stationary good confinement (H_98y2 ∼ 1.1, n_e/n_GW ∼ 0.4, f_BS ∼ 23% with the LH current fraction f_LHW ∼ 75% and the EC current fraction f_EC ∼ 2%, β_P ∼ 1.1, β_N ∼ 0.9, W_MHD ∼ 120 kJ).²² Recently, the record duration of 310 s fully non-inductive H-mode plasma with improved performance (H_98y2 ∼ 1.3, n_e/n_GW ∼ 0.7, f_BS ∼ 57% with f_LHW ∼ 27% and f_EC ∼ 16%, β_P ∼ 2.5, β_N ∼ 1.6, W_MHD ∼ 160 kJ) was achieved with full metal wall in 2022. As shown in Table I, both pulse-length and performance largely improved toward CFETR steady state scenario at fusion power near 1 GW.²⁵ 

Figure 2 shows selected time traces of the record duration ∼310 s stable H-mode plasma. It achieved β_P ∼ 2.5, H_98y2 ∼ 1.3 at n_e/n_GW ∼ 0.7 with line-averaged density ⟨n_e⟩ ∼ 3.2 × 10¹⁹ m⁻³, and toroidal magnetic field B_T ∼ 2.3 T. Fully non-inductive plasma is obtained with pure RF heating power P_rf ∼ 4.8 MW (P_LHW ∼ 2 MW at 4.6 GHz, P_EC ∼ 1.6 MW, P_IC ∼ 1.2 MW). A new water-cooled tungsten lower divertor significantly enhanced the particle and power exhaust. An advanced double-null (DN) configuration is applied for the plasma equilibrium with robust iso-flux control, which helps to reduce the target heat flux. Grassy ELMs with a frequency about 500 Hz lead to negligible transient heat load. This grassy ELM regime is characterized by high-frequency small-amplitude perturbations, a relatively high plasma density near separatrix, and low density gradient in the pedestal region. A small ELM regime can facilitate the RF power coupling to the H-mode edge and reduce the divertor sputtering/erosion.^4,33 It also can enhance the boundary impurity screening and facilitate the particle exhaust, improving impurity flush-out and plasma density control.^34,35 Thus, much lower W sputtering rate leads to low W source from the divertor with good tungsten control till ∼300 s. The limitations for this long-pulse H-mode plasma are due to hot spots localized at the lower divertor due to leading edges and also from sputtering and erosion of the RF-antenna and guard limiter.

Since the middle of 2020, to improve EAST flexibilities and capabilities, there were several major upgrades, making it possible to achieve a milestone on the path to steady-state long-pulse high-performance scenarios toward ITER and CFETR.

To reduce interference among the H&CD systems, they were rearranged as shown in Fig. 3. Figure 4 shows the effect of ICRF at I-port and previous B-port (Fig. 3) on the reflected coefficients (RCs) (%) of 4.6 GHz LHCD on E-port against the counts representing the number of points for a given time interval (0.1 s). Comparing ICRF on B-port/N-port and ICRF off, it can be seen that ICRF at B-port resulted in the highest reflection coefficient value of 4.6 GHz LHCD and reduced its efficiency. Thus, the ICRF system previously on B-port was moved to N-port, and correspondingly, the 2.45 GHz LHCD previously on N-port was moved to B-port.³⁶ Meanwhile, aiming to improve the NBI heating efficiency, NBI sources previously on F-port (Fig. 3) for countercurrent injection were moved to D-port re-directed for co-current injection, to avoid significant orbit losses from countercurrent beams.³⁷ 

To improve antenna coupling and H&CD capabilities, several upgrades were also made to the RF systems. For the 2.45 GHz lower hybrid wave (LHW) system, to improve long-distance coupling, the old FAM (fully active-multijunction) launcher was upgraded to the PAM (passive active-multijunction) launcher with efficient cooling.³⁸ This type of launcher can be located far away from the plasma, which is favorable for long-pulse operation. For the ICRF systems, to improve the antenna coupling, two newly developed antennas were deployed at N-port.³³ The key design features of the new antennas are lower parallel wave number $k∥$ of the power spectrum, two current strap configuration, and field aligned Faraday screen. According to the wave propagation theory,³⁹ the fast wave cutoff density can be decreased by decreasing $k∥$⁠. By decreasing $k||$ from 13–14 to 7.5  $m−1$ for (0, π) phasing, the cutoff density of the fast wave decreases from 8  $×1018$ to 2.8  $×1018 m−3$⁠. Consequently, the antenna coupling resistance in H-mode is increased significantly as shown in Fig. 5, which compares the new ICRF antennas with the old antenna. For the ECH systems, to improve control flexibility of the electron temperature and current profiles on EAST, the total power was increased to 4 MW from a four gyrotrons at 140 GHz.

To enhance particle and heat flux load and removal capability, EAST has been equipped with full metal wall including a new lower W/Cu divertor. The inner wall and baffles on EAST were upgraded in 2012 to water-cooled molybdenum (Mo) alloy tiles⁴⁰ with heat removal capability of ∼1 MW m⁻². The upper divertor was upgraded to actively cooled tungsten (W) in 2014 with ITER-like W/Cu monoblock plasma-facing units (PFUs) and cassette bodies.⁴¹ A W/Cu monoblock is composed of tungsten, oxygen-free high conductive copper and copper–chrome–zirconium (CuCrZr) as armor, interlayer, and heat sink, respectively, which can withstand high heat loads up to 10 MW m⁻² on the top surface. Instead of the previous lower W-shape carbon (C) divertor, the new lower water-cooled W/Cu divertor wase installed in 2020. Its structure consists for 75% of a monoblock structure and 25% of a flat-type structure [see Fig. 18 in Ref. 42] with more closed geometry with larger slot, which helps to increase the flow conductance by ∼36%. The W/Cu flat-type components based on new explosively welding and brazing technologies are expected to withstand steady-state heat loads as high as 10 MW m⁻² and transient heat loads up to 20 MW m⁻² on the top surface.⁴³ A special cooling system structure was designed in a copper-stainless steel composite plate for a high heat removal efficiency. For the new flat-type structure design, the calculations show that transient heat loads ∼20 MW m⁻² can lead to 1190 °C on the top surface with the inlet flow rate of ∼650 ton/h, and the thermal equilibrium can be achieved in 10 s. It is consistent with thermal fatigue test result. Figure 6 shows that the maximum temperature of W surface is ∼926 °C using a heat load of ∼20 MW m⁻², duty cycle ∼50% with 15 s on, 1000 cycles, and inlet flow rate of ∼280 ton/h. In order to mitigate the leading-edge-induced high thermal loads, chamfer shaping structures were employed.⁴⁴ 

Optimized particle control is the key issue to sustain long-pulse H-mode plasma,⁴⁵ which requires highly efficient fueling, enhanced pumping capability with adequate particle exhaust in the divertor region. For the fueling part, an advanced gas puffing (GP) system on EAST has been installed at various positions, including low field side and high field side of midplane, and in/out/dome region of both upper and lower divertors.⁴⁶ A supersonic molecular beam injection (SMBI) system and a pellet injection (PI) system⁴⁷ were developed for plasma fueling from low field side of midplane and from both the horizontal and vertical plates of the lower outer divertor region. A systematic investigation of the fueling efficiency shows that SMBI at midplane (SMBI-M) is ∼40%, and GP at the lower divertor dome region (GP-LD) is ∼16%. Therefore, in order to achieve high fueling efficiency and to reduce excessive particle retention on the first wall, GP-LD and SMBI-M are exploited for EAST long-pulse operation, which is good for plasma density control in the feedback mode.⁴⁸ For the pumping part, in-vessel divertor cryopumps in the low field side of both the upper and lower divertor regions provide an enhanced pumping capability for particle exhaust, each with the nominal pumping speed of 75 m³/s for deuterium.⁴⁹ External cryopumps installed in the upper and lower divertor region provide additional capability for divertor particle exhaust. In addition, lithium coating assisted by wall conditioning plasmas with the consumed Li 5–15 g every time and real-time lithium powder with the injection rate of 1–2 mg/s during the plasma are routinely used to improve the wall pumping rate and to lower the outgassing from the wall surface, which is helpful to achieve good fuel recycling control for long-pulse plasmas.⁵⁰ Furthermore, Fig. 7 shows the influence of different divertor magnetic configurations on the particle exhaust, with the external fueling terminated during the period at 3.0–4.5 s. It indicates that double null (DN) configuration with lower outer strike-point on the horizontal divertor plate can provide the highest particle exhaust capability, probably due to particles that could be exhausted from both upper and lower divertor region via cryopumps; moreover, the strike-point on the horizontal plate of the lower divertor region is closer to the pumping slot than that on the vertical plate, which is more suitable for particle exhaust.⁵¹ 

The steady-state real-time plasma control requires integrated accurate and reliable control of plasma equilibrium, current, and loop voltage.⁵² The fiber optic current sensors (FOCS), based on the Faraday Effect, thus, with no signal drift, are used for the first time to provide higher precision current measurement compared with Rogowski coils in plasma current control.⁵³ Low or zero drift integrators are applied for magnetic measurements used to reconstruct and control the plasma position and shape.⁵⁴ Small drifts were observed during the dry-run operation and corrected in the EAST plasma control system (PCS) for long-pulse operation. Magnetic measurement signal selection and fitting uncertainty in real-time equilibrium reconstruction are based on error analysis with 500 s dry-run operation. With the fine-tuning optimization of the loop voltage PID controller and 4.6 GHz LHW feedforward power, the loop voltage is controlled within ±5 mV for fully non-inductive operation as shown in Fig. 8.

With enhanced flexibility and capabilities for EAST, the high β_P scenario with high bootstrap current fraction and extended performance was further explored. Figure 9 shows that improved confinement quality is achieved with the extension of fully non-inductive H-mode plasmas to higher β_P at higher density. The improved confinement quality with the reduced turbulence at higher β_P was also observed in previous EAST experiments,²² DIII-D,⁵⁵ and JT-60U.⁵⁶ Transport simulations⁵⁷ show that the energy transport is insensitive to E × B shear flow and point to the strong effect of Shafranov shift (α-stabilization) on the turbulent transport in the higher β_P regime. Furthermore, Fig. 9 also suggests a synergism between high density and high β_P, in agreement with theoretical predictions that a high density gradient is important to leverage the Shafranov shift turbulence stabilization and achieve higher confinement quality.⁵⁸ 

Figure 10 shows the experimental electron temperature and density profiles of shot 110488, the calculated safety factor profile. Here, the radial coordinate ρ used in all figures about profiles in this paper is the square root of the normalized toroidal magnetic flux. Similar to previous long-pulse H-mode discharge developed in EAST,²² it is characterized by the existence of the weak internal transport barrier (ITB) in the electron temperature profile and the calculated current density profiles for each component. The bootstrap current (“BS”) is calculated with the Sauter model.⁵⁹ The current driven by EC waves (“EC”) is simulated with the coupled TORAY⁶⁰ and CQL3D⁶¹ modeling, and the current driven by LH waves (“LH”) is simulated with the coupled GENRAY⁶² and CQL3D model. The total non-inductive current profile (“sum”) is taken as the total current profile to update the fixed-boundary plasma equilibrium with the boundary reconstructed beforehand by the EFIT code⁶³ and the pressure profile calculated from the experimental temperature and density profiles. The q profile of the updated equilibrium is shown in Fig. 10, which is in agreement with the Faraday angles measured by the polarimeter interferometer (POINT) system.⁶⁴ Since this equilibrium is consistent with the kinetic pressure and non-inductive current profiles, it is used in the following modeling and analysis for plasma transport.

The temperature and density profiles in the core region are predicted by the TGYRO code,⁶⁵ which calls the TGLF code⁶⁶ and the NEO code⁶⁷ for the calculation of turbulent and neoclassical transport, respectively. Three turbulence saturation models (i.e., the conventional model SAT0,⁶⁸ recently developed models SAT1,⁶⁹ and SAT2⁷⁰ in TGLF are used in our modeling. The predicted results are in good agreement with the fits of the experimental measurements, as shown in Fig. 11. The difference of predicted profiles due to different saturation rules is small for this experiment. This indicates that the conventional model SAT0 can well estimate the turbulent transport in this experiment, consistent with the success of SAT0 in describing baseline inductive H-mode scenarios on the other machines discussed in Ref. 71.

The profiles obtained by TGYRO with SAT0 are used as the reference case in the following analysis for plasma transport. The turbulent thermal diffusivity profiles are shown in Fig. 12. χ_e and χ_ei drop from the outer region toward the core region. The turbulent energy transport dominates over neoclassical energy transport. Figure 13 shows the growth rate and frequency of the most unstable drift-wave mode in the core region calculated by the TGLF code. The frequency larger than zero indicates that the mode is propagating along the electron diamagnetic drift direction, suggesting that trapped electron modes (TEM) dominate in the core region. Furthermore, the energy turbulent transport decreases as the collisionality increases, which is consistent with the feature of TEM induced transport (see more discussion in previous analysis for the EAST β_P scenario with only RF heating⁷²).

For 310 s H-mode plasma, fully non-inductive plasma is obtained with pure RF heating power P_rf ∼ 4.8 MW (P_LHCD ∼ 2MW, P_ECRH ∼ 1.6 MW, P_ICRF ∼ 1.2 MW) at n_e/n_GW ∼ 0.7. Figure 14 shows the density limit for LHCD existed for both 2.45 and 4.6 GHz waves on EAST. Previous experiments also showed that the current profile driven by LHW becomes more off-axis at higher density indicating a reduced radial penetration of the wave and a wave absorption closer to the plasma edge.^21, Figure 14 shows that the higher frequency lower hybrid wave with 4.6 GHz has a higher LHCD density limit than the 2.45 GHz LHW.⁷³ This is because the parametric instability (PI) is much lower for higher frequency wave, especially at higher density, which enhances the CD efficiency.^74,75 Thus, for long-pulse H-mode operation, higher power is systematically provided by the higher frequency 4.6 GHz launcher. In addition, good wall conditioning can enhance LHCD efficiency. Experimental results show that strong lithium wall conditioning can reduce PI intensity due to lower edge recycling and higher temperature at the edge.^76,77 Furthermore, to improve LHCD efficiency, ECRH is applied to increase the electron temperature. According to CD theory,^39,78 higher temperature corresponds to higher velocity of resonant electrons, which means stronger absorption of LHW power and lower collisionality, thus higher CD efficiency. As shown in Fig. 15, the central T_e is increased by ∼1.0 keV and the HXR count rate is higher by ∼73% with an additional 0.9 MW of ECRH, indicating higher CD efficiency.⁷⁹ To improve the ICRF antenna coupling, the new ICRF antenna at the N-port is updated with smaller parallel wave number $k||$ ∼ 7.5 m⁻¹. By comparing with the previous antenna, Fig. 16 shows that the heating efficiency is increased significantly for the ICRF power.

The time evolution of the peak heat load (q_t) in Fig. 17(a) and the temperature in Fig. 17(b) for the lower outer (LO) divertors are obtained through the infrared camera.⁸⁰ During the long-pulse duration, the heat load for the target is controlled and kept lower than ∼3 MW/m² with temperature maintained at <700 °C. This indicates that thermal equilibrium can be achieved quickly at the W/Cu divertor target. Figure 17(c) shows the stable evolution of the two-dimensional distribution of ion saturation current density (j_s) measured by the divertor Langmuir probes, indicating the good stability of the divertor configuration in this discharge. The double null configuration is stably controlled, with the two stable strike points shown as the red and black lines with circles in Fig. 17(c) calculated by the EFIT magnetic equilibrium reconstruction code. This advanced DN divertor configuration helps to reduce the target heat flux.

Figure 18 shows that the 310 s H-mode plasma was mainly fueled by SMBI in the feedback control mode, and the SMBI injection rate was gradually decreasing along with the discharge duration. The global recycling coefficient (R_global) increases from 0.93 to ∼1 at the end of the discharge, indicating gradual increase in fuel recycling but still in control at ∼300 s. Particle balance analysis shows that the particle exhaust rate by the divertor cryopumps is ∼5 × 1020 D/s and almost constant during the discharge. The first wall changed from providing a pumping role to an outgassing role from t ∼ 160 s even with continuous real-time lithium powder injection, and the outgassing rate was increased to ∼5 × 1020 D/s by the end of the discharge. This rate is close to the exhaust rate of divertor cryopumps, resulting in R_global ∼ 1 at the end of the discharge. The increasing recycling along the discharge duration was attributed to the increased first wall temperature and accumulated fuel retention in the wall surface.

Figure 19 shows that high-Z impurity accumulations in the plasma core are well controlled at a low level by using on-axis ECH for long-pulse (>100 s) H-mode plasma. Modeling shows that the strong diffusion due to TEM turbulence in the central region increases anomalous diffusion sufficient to pump out tungsten impurity and avoid tungsten accumulation.²⁶ Small ELMs exist in this ∼310 s long-pulse H-mode discharge and lead to negligible transient heat load. Thus, much lower W sputtering rate leads to low W source from the divertor. However, ∼310 s long-pulse plasma is terminated with increasing tungsten and particle recycling. The limitations of the long-pulse H-mode plasma are from hot spots in the local area of the lower divertor, due to leading edges, and also from sputtering and erosion from the RF-antenna and guard limiter.

Small ELMs have been commonly observed in the high β_p discharges on EAST, which are thought to be the grassy ELMs.^34,35 By comparing with the type-I ELM regime on EAST, the small ELM regime is characterized by a wide density pedestal and a high ratio between the separatrix density and pedestal top density n_e,sep/n_e,ped ∼ 55% as shown in Fig. 20, which is thought to be critical for access to this small ELM regime. The wide density pedestal and high n_e,sep/n_e,ped in the small ELM regime would help to stabilize peeling-ballooning modes (PBMs) by reducing the pedestal pressure gradient and pedestal current density dominated by the bootstrap current density. Pedestal linear stability analysis with the ELITE code⁸¹ suggests that the operational point is located in the PBM stable region and near the PBM stability boundary as shown in Fig. 21. This is consistent with previous small ELM results,³⁴ where the operational point is also located near the PBM stability boundary. The stability boundary is set to be the value when the ratio of the growth rate of the most unstable mode to half of the ion diamagnetic frequency equals 1. The ELM size is determined by the nonlinear pedestal dynamics. The underlying physical mechanism for the observed small ELMs is that the PBM boundary would expand after the initial pedestal collapse during the small ELM burst, making the operational point move to more PBM stable region and preventing large amplitude pedestal collapse. In addition, the small ELM regime characterized by a low pressure pedestal gradient and a low pedestal bootstrap current density has a very localized instability in the pedestal region that would not trigger large-amplitude PBMs and affect the core region. These are essential for negligible transient heat load and to minimize the influence on the core confinement, which are beneficial for good core-edge compatibility in high β_p scenario.

Low particle source and strong particle exhaust in the pedestal region are thought to be two main factors leading to the flat density pedestal profile in the small ELM regime. Previous studies⁸² have found that when the divertor strike point is located on the horizontal target plate, trapping of recycling neutrals in the closed divertor corner area would result in a reduced pedestal fueling from the lower divertor, thus reducing the pedestal density gradient. In addition, with the aid of pedestal particle exhaust driven by the high-frequency small-amplitude edge localized instabilities, a flat density pedestal profile with high SOL density can be achieved more easily.

Based on the 310 s H-mode plasma, Fig. 22 shows the dependences of the electron energy transport flux by TEM on the Shafranov shift parameter α_MHD, the magnetic shear s, and q. All the dependences can be separated into two branches, i.e., the conventional confinement state branch (Conv.) and the advanced confinement state branch (Adv.). The transport flux is lower with lower α_MHD, q, and/or higher s in the Conv. branch, while in the Adv. branch, higher α_MHD, higher q, and/or reduced s can reduce the transport flux. The operation point of this EAST experiment stays in the Conv. branch as shown in Fig. 22. The previous high β_P experiments on EAST with only RF heating²² also stayed in the Conv. branch according to the analysis given in the integrated modeling work.⁷² The exploration on EAST toward advanced steady-state scenarios with higher H_98y2 is looking for methods to move the operation point to lower s, higher α_MHD, and higher q. As discussed in Ref. 83, increasing the heating power and/or broadening the current profile may be needed in order to climb over the peaks of the transport flux (indicated by dashed lines in Fig. 22) for the entry of the advanced confinement state.

In order to develop the path toward ITER and CFETR steady-state advanced operation, EAST will aim at accessing more relevant dimensionless parameters to develop long-pulse high-performance plasma. Figure 23 shows the comparison of several dimensionless parameters between recently achieved 310 s H-mode plasma on EAST with the ITER/CFETR steady-state design. Three key challenges need to develop innovative physics understanding and approaches toward ITER and CFETR steady-state long-pulse operation.

The first challenge is to explore fully non-inductive high-β_p scenario at medium q₉₅ ∼ 4–6 with excellent confinement and low torque injection, which requires to increase the plasma current at the same toroidal field, sustained with the available RF-heating. The main H&CD systems will be further upgraded with improved efficiency at high density. In the next 3–5 years, (1) for LHW, we will use 4.6 GHz PAM instead of the current 2.45 GHz PAM and will increase its power to 4.0 MW, to enhance LHCD fraction with high frequency at high density, while leveraging the long distance coupling of PAM to avoid plasma interactions with the antenna at high power injection; (2) for ECH, we will increase the number of gyrotrons up to 6, for a total power up to 6.0 MW. Another two gyrotrons (total 2.0 MW) at dual frequency140/105 GHz.0215 and two gyrotrons (2.0 MW) with 170 GHz as the same frequency as ITER have been planned, which could provide more useful off-axis e-heating for current profile control; (3) for ICRF, we will have one more antenna for an additional 3.0 MW increasing the total ICRF power to 9.0 MW. The new antenna will be expected to provide higher heating efficiency at higher plasma current and lower q₉₅ by H minority heating in D plasmas,⁸⁴ enabling various transport studies (T_e/T_i, electromagnetic effects, magnetic shear, Shafranov shift, and E × B shear), the study of energetic particles (EP) from ICRH, EP interaction with MHD (internal kink, NTM control), and the study of EP effects on SSO and performance (including coupling with 3D structures and ELMs); (4) for NBI, we would update the injectors with new RF sources, with the total power up to 4.0 MW capable of long-pulse operation. Energetic particle physics from NBI and ICRF will play a crucial role for the development of extended steady state operations in EAST and of experimental database in support of the CFETR design. A potential high current driven by ICRF minority ion heating and improved heating capabilities enable access to ITER and CFETR relevant β_N and q₉₅. The facility enhancement will exploit high-performance core solutions and validate models in burning plasma relevant conditions.

The second challenge is to improve performance by broadening the current and pressure profiles with a large radius ITB in the high β_p regime. This, combined integrated scenario modeling, will enable us to explore the fundamental physics of operation with q_min > 2, large radius ITB, and high f_BS. The increase in ECH and ICRH powers will address profile requirements for ideal MHD limits, good EP, and global confinement. Recently, a broader current profile with q_min > 2 and li ∼ 0.7 was obtained by early heating using off-axis ECH and was sustained with combined LHW, ICRF, and NBI powers.⁸⁵ Based on this high q_min profile, future experiments will pursue the formation and sustainment of an ITB at larger radius, in order to further extend the fusion performance of long-pulse plasmas on EAST.

The third challenge is to demonstrate long-pulse operation with high power (P_loss/R) to extend fusion performance. To improve the particle and power handling, a scientific and technological solution will be developed. According to the damage from hot spots on limiters and leading edge on the divertor in recent experiments, new modular limiters and active water-cooling for RF-antennas and divertor upgrades (e.g., lead-edge, alignment, cooling capability) will be developed and installed. EAST has developed a number of heat flux control techniques⁸⁶ to dissipate heat load with high-performance plasma maintained in the core.⁸⁷ 

A record duration of 310 s fully non-inductive H-mode plasma has been achieved on EAST with H_98y2 ∼ 1.3, β_P ∼ 2.5, n_e/n_GW ∼ 0.7, f_BS > 50%, due to enhanced facility capabilities and improved relevant fundamental physics understanding. For the improved core energy confinement quality with zero torque injection, the analysis points to the strong effect of Shafranov shift on turbulence. Important synergistic effects are leveraged toward this result, which relies purely on RF powers for heating and current drive. Advanced double-null configuration is applied to the plasma equilibrium with robust iso-flux control, which helps to reduce the target heat flux. New water-cooled W/Cu lower divertor can significantly enhance particle and power exhaust. High-Z impurity accumulation in the plasma core is well controlled in a low level by using on-axis ECH. Small ELMs lead to negligible transient heat load and low W source from the divertor. Toward ITER and CFETR SSO, EAST will aim at accessing more relevant dimensionless parameters with the low single-null configuration to develop long-pulse high performance. Near term plan with upgrade of the inner components and augmented heating and current drive systems will aim at realizing 400–1000 s long-pulse H-mode operation with ∼60% bootstrap current fraction and demonstrate SSO with extended fusion performance at 15–20 MW power injection with the duration >100 s.

The authors would like to thank Dr. P. T. Bonoli for his helpful discussion and support. This work is supported by the project National Key Research and Development Program of China No. 2019YFE03020004, Anhui Provincial Natural Science Foundation No. 2008085J04, Anhui Provincial Key R&D Programmes No. 202104b11020003 and the Excellence Program of Hefei Science Center CAS No. 2021HSC-UE015, the National Natural Science Foundation of China under Grant No. 11975276, the U.S. Department of Energy, Office of Science, Office of Fusion Energy Sciences Cooperative Agreement Nos. DE-FC02–04ER54698, DE-SC0010685, DE-SC0014264, DE-AC52–07NA27344, and DE-AC02–09CH11466.

The authors have no conflicts to disclose.

Juan Huang: Data curation (equal); Formal analysis (equal); Investigation (equal); Writing – original draft (equal). Yifeng Wang: Formal analysis (equal); Investigation (equal). Yao Huang: Formal analysis (equal); Investigation (equal). Hang Si: Formal analysis (equal); Investigation (equal). Lingyi Meng: Formal analysis (equal); Investigation (equal). Tianqi Jia: Data curation (equal). Yanxu Sun: Data curation (equal). Long Zeng: Data curation (equal). Le Han: Data curation (equal). Yanmin Duan: Data curation (equal). Annika Ekedahl: Writing – review & editing (equal). X. Gong: Data curation (equal); Writing – review & editing (equal). Christopher Thomas Holcomb: Writing – review & editing (equal). Rajesh Maingi: Writing – review & editing (equal). Erzhong Li: Data curation (equal). Haiqing Liu: Data curation (equal). Bo Lyu: Resources (equal). Qilong Ren: Resources (equal). Youwen Sun: Resources (equal). Liang Wang: Data curation (equal). Liqing Xu: Data curation (equal). Damao Yao: Resources (equal). Andrea M. Garofalo: Data curation (equal); Writing – review & editing (equal). Qingquan Yang: Investigation (equal). Qing Zang: Data curation (equal). Bin Zhang: Data curation (equal). Ling Zhang: Data curation (equal). Xuemei Zhai: Software (equal). G.Z. Zuo: Resources (equal). G. Q. Li: Data curation (equal). Pengfei Zi: Resources (equal). Mao Wang: Resources (equal). Handong Xu: Resources (equal). Jinping Qian: Data curation (equal); Formal analysis (equal). Qiping Yuan: Resources (equal). Yahong Xie: Resources (equal). Liansheng Huang: Resources (equal). Jian Zhang: Resources (equal). Yanlan Hu: Resources (equal). Weibin Xi: Resources (equal). Zhiwei Zhou: Resources (equal). Zhengchu Wang: Resources (equal). Bin Guo: Resources (equal). Guosheng Xu: Resources (equal). Rui Ding: Data curation (equal); Investigation (equal). Jiansheng Hu: Resources (equal). Kun Lu: Resources (equal). Yuntao Song: Supervision (equal). Baonian Wan: Supervision (equal). J. G. Li: Supervision (equal). Xinjun Zhang: Data curation (equal); Formal analysis (equal). Jiale Chen: Formal analysis (equal); Investigation (equal). M. H. Li: Formal analysis (equal); Investigation (equal). Yaowei Yu: Formal analysis (equal); Investigation (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.