ICRH assisted breakdown study on JET

Ion cyclotron (IC) wave assisted breakdown has the potential to increase the robustness of plasma initiation during the ITER pre-fusion operation phase. Studies were performed at JET at ITER relevant loop electric field, E loop ≲ 0 . 33 Vm − 1 , and a range of toroidal fields, including at the low toroidal field of 1 . 7 T for which breakdown had not been achieved previously on JET. The study covered a range of H 2 and D 2 gas prefill pressures and timings, pumping conditions, and residual impurity levels. IC assisted breakdown was achieved for a lower and wider range of gas prefill pressures. IC assisted breakdown works by activating wall pumping before the current rise, changing the relation between fuelling and torus pressure in this phase compared to Ohmic breakdown. IC assisted breakdown enables plasma initiation with a higher level and significantly wider range of injected plasma prefill gas. As the injected prefill gas is the controlled parameter, this significantly improves the robustness of plasma initiation operationally. IC assistance is found to be more robust at ITER-like E loop , succeeding with higher low-Z impurity content. Moreover, it does not introduce an impurity source that may hamper the subsequent burn though and current ramp-up phase. For both the IC assisted and pure Ohmic breakdown, the initial current rise rate is found to scale with n e / E loop . The results and implications for ITER are presented.

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Plasma initiation in a tokamak begins with a Townsend avalanche phase where residual electrons in the vacuum vessel are accelerated by an induced toroidal electric field, producing secondary electrons through collisional ionisation of prefilled neutrals. This is followed by a burn through phase where, locally, most atoms are stripped of their electrons and the temperature of the plasma is increased through Ohmic heating until the plasma energy losses from interaction with neutrals (electron impact reactions and charge exchange collisions) become small with respect to thermal losses. Predicting the criteria and evolution of breakdown and burn through requires detailed modelling [1], but can be approximated by the Townsend avalanche breakdown criteria: where L f is the is the average length of open magnetic field lines (i.e. effective connection length), λ i the mean length for an electron to be accelerated and make a collisional ionisation of a neutral particle, p is the prefill gas pressure, E loop is the toroidal (loop) electric field, and a and b are constants [2] which, at room temperature (300 K) for H 2 , D 2 , or T 2 , are taken as a = 0.258 mPa −1 and b = 950 Vm −1 Pa −1 [3]. Based on the equation, L f /λ i can be used as a predictor for successful plasma breakdown. The breakdown criteria of equation (1) can be seen to have an optimal pressure but increasing E loop is always favourable for breakdown and lower E loop unfavourable with the range of allowable prefill pressures decreasing with decreasing E loop until a value of E loop below which no breakdown is possible. The maximum E loop for ITER is expected to be 0.33 Vm −1 which is considerably lower than that of existing tokamak (for JET E loop ≲ 1 Vm −1 ). This implies relatively long λ i and so presents a challenge for plasma initiation in ITER. Purely Ohmic plasma initiation is sufficient for JET and many existing tokamaks. Studies indicate that Ohmic plasma initiation in ITER may be challenging in the case of a high hydrogenic background pressure or when there are large amounts of residual in-vessel impurities such as those that may be injected for plasma optimisation or disruption mitigation [1,4]. This has motivated the study of electron cyclotron (EC) wave assisted plasma initiation both applying EC power to form a pre-ionised plasma prior to Ohmic breakdown and applying EC heating power later to assist the burn through phase. These studies have demonstrated EC assisted plasma initiation on several machines [2,[5][6][7][8][9] and provided a physics basis from which it has been concluded that the method can improve the reliability of plasma initiation on ITER [1,3,10]. During the initial ITER pre-fusion plasma operation phase (PFPO-1), ITER will operate at low magnetic field (B t = 1.8 T) to enable H-mode access, for which the required heating power has been found empirically to scale near linearly with magnetic field, for H plasma. As the connection length scales linearly with B t , plasma initiation will be less robust than for the later ITER Fusion Plasma Operation phase. Moreover, studies have shown that the planned 170 GHz ITER EC system will not be able to effectively heat at the third harmonic resonance that would be required at B t = 1.8 T [11]. As is the case at higher field, Ohmic plasma initiation for ITER at 1.8 T is expected to be adequate, but ion cyclotron (IC) wave assisted breakdown would reduce risk of failed plasma initiation in the case of poor vacuum conditions [1]. Even though the present ITER research plan does not foresee IC heating capabilities for PFPO-1, a case for IC assisted breakdown needs to be developed. This motivates the present study which aims to demonstrate and optimise the method on JET and provide a physics basis for extrapolation to ITER. IC plasma formation has previously been demonstrated and optimised at JET for wall cleaning [12] and a reduction in flux consumption is observed when ICRH is appllied in the lower toroidal look voltage case [13].
The structure of this paper is as follows. Section 2 outlines the experimental method used for the studies. Section 3 presents the experimental results related to the criteria for successful plasma initiation. Section 4 presents the results related to the subsequent evolution of the plasma. In section 5, the results are summarised, and their implications discussed.

Experimental method
JET plasma initiation is performed [14] by using a set of poloidal field coils to generate a hexapole null in the poloidal field inside the vacuum vessel at the instance when the loop voltage, induced by the central solenoid, has reached its set steady value. In the region around the null, poloidal fields are small, hence the connection lengths are long and so favourable for breakdown. Once an initial plasma carrying a current higher than a few kA is established, the central solenoid controls the plasma current and the poloidal field coils control the plasma position and shape until the required plasma configuration is established. Figure 1 shows the waveform used for the IC assisted breakdown experiments. Breakdown is initially avoided by applying, from t = −1 s, a vertical magnetic field which ensures a short connection length. The required E loop and prefill gas pressure is then developed. The applied vertical magnetic field is removed at t = 0.45 s and breakdown attempted. The exact time of the breakdown and its development depends upon the experimental conditions. A plasma current waveform is requested from t = 0.45 s and the JET plasma control system essentially limits E loop to deliver this current. Before t = 0.8 s, the requested plasma current is kept low to avoid a strong increase in the loop voltage. The feed forward request of the loop voltage is gradually reduced over 200 ms to obtain a fully feedback controlled electric field at or before t = 0.8 s. During this time, the breakdown and burn through phases evolve freely. From t = 0.8 s, the requested plasma current is raised and E loop is controlled to deliver this. As the plasma resistivity is low at this time, the required electric field is well below the 0.33 Vm −1 limit of ITER.
Following previous studies of IC discharges for IC wall conditioning (ICWC) in JET [12], the IC frequency is selected for an on-axis fundamental H resonance with the, toroidally distributed, antenna straps all in phase-referred to as 'monopole' phasing. The resulting IC heating scheme is H minority for D plasma and H majority for H plasma. The antenna protection settings are identical to those for JET ICWC operation, following [12]. The interlock for the pressure in the IC vacuum transmission lines is lowered to 5 mPa with the same limit for the torus pressure. The number of trips per IC amplifier is limited to 10 per discharge and 100 per experimental session. The maximum permitted Voltage Standing Wave Ratio for the IC antenna is increased, but the maximum permitted voltages on the transmission lines is limited to 20 kV. The external conjugate-T matching system is not used. The same injected power was used throughout.
Following the JET ICWC scheme, pre-fill gas fuelling is from valves in the divertor for all discharges to avoid fuelling gas too close to an IC antenna and to obtain a uniform torus pressure. This contrasts with the mid-plane pre-fill gas fuelling more commonly used for breakdown. Compared to mid-plane fuelling, divertor fuelling on JET requires more injected gas to produce the same change in vessel pressure. This is experimentally and operationally advantageous, as it allows finer control of vessel gas pressure. Divertor fuelling on JET results in toroidally symmetric gas pressure in contrast to mid-plane fuelling where vessel gas pressure is larger in the toroidal location close to the injecting valve. The toroidally symmetric gas pressure distribution means that divertor fuelled breakdowns are easier to interpret which is experimentally and operationally advantageous.
Dedicated Ohmic breakdown references were performed using the same setup as for the IC assisted breakdowns only with the IC power turned off.
Since JET began operating with the ITER-like Be/W wall (JET-ILW), plasma initiation has been robust with failed plasma initiations being rare [15,16]. This is when operating with breakdown E loop ≈0.9 Vm −1 and B t ⩾ 2.3 T to ensure sufficiently long connection lengths. To study an ITER relevant regime where E loop and connection lengths may be low enough that plasma initiation becomes marginal, these studies were performed with E loop ≲ 0.33 Vm −1 and B t ≈1.7 T. JET plasma initiation had not previously been achieved in such conditions.
Diagnosis follows the approach used in previous JET breakdown studies [16]. The loop electric field is inferred from the central solenoid (P1 coil) voltage, V P1 . The P1 coil has N T = 710 turns, so V P1 = 2π R 0 N T E loop , where R 0 = 2.96 m is the JET geometric radius. In-vessel neutral gas pressure is measured with Penning gauges [17,18]. The line integrated density is measured with the JET far-infrared interferometer using a vertical line of sight passing through the centre of vessel with a time resolution of 10 µs [19,20]. Bremsstrahlung and D Balmer-α emission is measured by a visible spectroscopy system with multiple lines of sight and 1 ms time resolution [21]. Z-effective is inferred from Bremsstrahlung radiation. The JET VUV emission spectroscopy system is used to measure line radiation from impurities, including Be III, C III, C IV, O VI, Ne VII, Ne VIII, Ne IX, Ar XVI, Fe XXIII, Ni XVIII and W (multiple lines) [22].

Overview of the JET IC assisted breakdown results
Plasma initiation with IC assisted breakdown has been successfully established in the regime of interest for ITER, E loop < 0.33 Vm −1 , for a range of fields B t = 1.7 − 2.3 T.  Figure 2 shows a typical discharge (red, #100624) compared with an Ohmic breakdown (blue, #100636) at matched E loop ≲ 0.33 Vm −1 and B t = 1.7 T. In the period before the main breakdown phase, a low density, n e ⩽ 5 × 10 18 m −3 , plasma is produced in the vacuum vessel by powering the IC antennas. After an initial increase of the torus gas pressure due to gas fuelling in vacuum, the pressure falls at the onset of the IC discharge due to wall pumping. This process eventually saturates as shown by the gradual pressure increase from ≈0.25 s. The, so-called, 'pre-ionisation' plasma is maintained until the start of the main breakdown phase. The torus pressure at breakdown can be tuned by the gas fuelling rate and the launched IC power. The plasma density rise rate through the breakdown and plasma burn through phases is much higher than for the Ohmic reference. The following analysis will look more carefully into successful and unsuccessful plasma initiation attempts for both Ohmic only and IC assisted scenarios. It is important to note that most of pulses achieved breakdown while many failed in the burn through phase. Hence, a successful plasma initiation is hereafter defined by a successful completion of the plasma burn through phase. Consistent with previous studies, the time point at which the JET plasma current reaches 100 kA is used to define the loop electric field and neutral gas pressure of the plasma breakdown [16]. Figure 3 shows the full set of IC assisted breakdown discharges in this study. A range of breakdown pressures and toroidal (loop) electric fields were attempted with D 2 fuelling at three different toroidal magnetic fields, B t = 1.7, 2.0, and 2.3 T. A range of breakdown pressures and toroidal (loop) electric fields were also attempted with 33 Vm −1 with the pressure range for successful plasma initiation identified as 0.5 − 2.4 mPa. The minimum pressure for successful IC assisted initiation appears to be independent of toroidal magnetic field (≈0.5 mPa), but IC assisted initiation was still successful for the highest pressure attempted at B t = 2.0 T (≈5 mPa). It is preliminarily concluded that the operational space of pressure is larger for higher B t , in line with the fact that connection length scales linearly with field. H 2 breakdowns appear to show the same trends as for D 2 ones, although the dataset of H 2 breakdowns is more limited.

Comparison of IC assisted and Ohmic initiation
To understand whether IC assistance improves the robustness of plasma initiation, comparison pulses with Ohmic breakdown only were selected from previous JET-ILW pulses. There have been over 20 000 attempted Ohmic breakdowns in the JET-ILW. As it is the more robust approach, most attempted breakdowns in JET are with a toroidal voltage induced by a current interruption in the primary coil (referred to as 'mode D' breakdown in JET). These breakdowns typically have 0.7 < E loop < 1.0 Vm −1 and are not comparable to the IC assisted breakdowns here. The Ohmic pulses selected for this study have the primary coil voltage under direct control (referred to as 'mode B' breakdown in JET), as was the case for the IC assisted ones. These Ohmic breakdowns have similar E loop ⩽ 0.5 Vm −1 to the IC assisted ones. Amongst these pulses, most have the prefill gas injected immediately before the breakdown time and so the torus pressure is highly asymmetrical. Such pulses are excluded from the dataset here and only pulses with prefill gas 0.5 s earlier than breakdown time, where the torus pressure is equalised and suitable for data analysis, are selected. The resulting dataset comprises 77 JET-ILW Ohmic breakdown pulses suitable for comparison with the IC assisted breakdown pulses introduced in section 3.1. Figure 4 shows the operational range of the reference Ohmic only breakdowns. Figure 4 includes both succeeded and failed pulses. Many of the attempted breakdowns are part of a previous study to achieve ITER-like (E loop ⩽ 0.33 Vm −1 ) breakdown. In that study, breakdown could only be achieved at B t = 2.8 − 3.0 T which is associated with longer connection lengths. The success rate of Ohmic only breakdowns with B t ⩽ 2.3 T for E loop ⩽ 0.5 Vm −1 is much lower than that of the IC assisted breakdowns and the achieved operational pressure range is narrow, ≈2.5 − 4.0 mPa. The majority of successful Ohmic breakdowns at higher B t = 2.8 − 3.0 T also fall within a similar operational pressure range ≈2.0 − 4.5 mPa.
The successful IC assisted breakdowns and Ohmic only breakdowns are show in figures 5(a) and (b) respectively. For the same B t , the operational pressure range of the IC assisted breakdowns is wider than for the Ohmic only breakdowns and the minimum E loop is lower for the IC assisted breakdowns. The minimum achieved E loop for plasma initiation is 0.25 Vm −1 . The majority of the B t = 2.0 T Ohmic breakdowns were from a set of experiments where the E loop and prefill pressure were optimised to ensure robust breakdown and their operational range for successful breakdown was not explored.
Although the Ohmic breakdown references have been selected so as to best match the IC assisted breakdowns presented in section 3.1, they still represent breakdowns which were often performed many years before, potentially under different machine conditions and with different prefill gas delivery. The previous Ohmic breakdowns also contained relatively few discharges with connections lengths comparable to the IC assisted breakdown discharges, that is comparable B t . As it  was not the focus of the experiments in which many of the previous Ohmic breakdowns occurred, the operational range for successful breakdown was often not explored and so cannot clearly be deduced. For these reasons, as part of the current experiment, Ohmic only breakdowns with D 2 fuelling at E loop ⩽ 0.33 Vm −1 and B t = 1.7 T were performed as dedicated references for IC assisted pulses with otherwise identical settings. Some of these were also successful, representing the first Ohmic breakdowns established at ITER-PFPO-like E loop and B t in JET. As for the IC assisted breakdowns, a full (seven-point) pressure scan was completed at B t = 1.7 T, figure 6(a). The pressure range for successful plasma initiation with Ohmic breakdown at B t = 1.7 T was observed to be 2.5 − 4.0 mPa, consistent with thatof the wider Ohmic breakdown dataset as noted earlier. Hence, IC assisted breakdown enables plasma initiation with five times lower gas prefill pressure and a much wider range (0.5 − 2.4 mPa). This gives access to the 1 mPa prefill pressure that is proposed for ITER [4]. The amount of injected prefill gas for successful plasma initiation with IC assisted breakdown (7.7 − 10.9 kPa l) is greater than for Ohmic breakdown (4.3 − 4.4 kPa l) with a considerably larger range, figure 6(b). As the amount of injected prefill gas  is the controlled parameter experimentally, Ohmic breakdown at E loop ⩽ 0.33 Vm −1 and B t = 1.7 T requires careful gas control, and is presumably sensitive to vessel condition, whilst IC assisted breakdown is more robust.

The impact of pumping
The impact ofdivertor cryo-pump condition was explored by comparing breakdowns attempted with the cryo-pump cooled by liquid N 2 and liquid He, which affect the pumping efficiency of hydrogen. For both H 2 and D 2 gas fuelling the JET divertor cryo-pump is found to pump at ≈150 m 3 s −1 when liquid He cooled and at ≈ 5 m 3 s −1 when liquid N 2 cooled. At both temperatures, scans of pre-fill gas pressure were performed with IC assisted breakdown with E loop < 0.33 Vm −1 at B t = 1.7 T. A pre-fill gas pressure scan was also performed with the cryo-pump cooled by liquid N 2 for Ohmic breakdown with the same E loop and B t . The impact of this on the breakdown studies can be seen by comparing the IC assisted D breakdown discharges at B t = 1.7 T with the two divertor cryo-pump coolants, figure 7. For both pumping conditions, a similar range of gas prefill pressure, ≈0.6 − 3.0 mPa, is observed for successful IC assisted breakdowns, figure 7(a). The range of pre-fill gas fuelling that is required to achieve these pressures is also similar with the two cryo-pump coolants, ≈75 − 110 mbar l, figure 7(b). Thus, for IC assisted breakdown, cryo-pump condition does not seem to obviously affect the breakdown pressure or the total injected pre-fill gas.

The role of IC wave injection in breakdown
Comparing IC assisted with Ohmic breakdown with the cryo-pump at liquid N 2 temperature, clearly the breakdown pressure and the total injected pre-fill gas ranges differ significantly, figure 8. The effective pumping speed, S, can be calculated from the expression for the time evolution of the gas prefill pressure p (t) where V is the total pumped torus volume, ≈180 m 3 ; p L is the stable torus pressure, ≈0.03 mPa for the N 2 cooled pump divertor; and F (t) is the time dependent gas injection rate. The total pumping speed is found to be ≈20 m 3 s −1 for Ohmic breakdown pulses and ≈1000 m 3 s −1 for the IC assisted breakdown pulses. The applied IC acts as a pump which greatly increases the amount of gas consumed by breakdown. As outlined in [23], IC power applied to breakdown the plasma creates additional pumping through its impact on normal neutral particle processes (fluxes to the wall and disassociation of molecular hydrogen or deuterium) and through the production of fast neutrals by charge exchange between protons or deuterons accelerated in the IC resonant layer and the background neutral gas. There is no strong trend across the dataset of a correlation between IC coupled power and successful plasma initiation. Some pulses with poor coupling (lower ICWC n e ) during IC pre-ionisation phase succeeded in plasma initiation, while some failed with well coupled power. There was no strong trend between the IC preionisation plasma density and breakdown either. Figure 9 illustrates this for a D 2 breakdown assisted with 33 MHz IC. Discharge #100416 (blue trace) is a successful E loop < 0.33 Vm −1 , B t = 1.7 T IC assisted discharge. Discharge #100408 (red trace) with similar loop voltage failed in plasma initiation. Coupled IC power is higher for this discharge and is associated with considerably (≈2 times) higher preionisation electron density. Despite this, the current fails to take off.
As outlined in section 3.2, Ohmic only and IC assisted plasma initiations differ significantly. Given that Ohmic only plasma initiations are equivalent to IC assisted plasma initiations with IC coupled and applied power of zero, clearly the amount of IC couple power does affect successful plasma initiation. The lack of correlation in the experiments likely results from the limited dataset and the interplay between plasma formation and IC power coupling. As discussed in section 1, the injected IC power and all other setting of the IC system are identical for all IC assisted breakdowns in this study. Hence, the coupled power depends only on the plasma conditions, including the plasma density.
Across all the IC assisted discharges, the IC preionisation plasma has low density, n e < 5 × 10 18 m −3 , and there is no correlation observed between IC plasma density and density at the end of the burn through. The time when I p = 100 kA may be taken as the approximate end of the burn through. At this time the IC assisted breakdown pulses have higher electron density than Ohmic breakdowns, even at similar prefill gas pressures, figure 10(a). Even though the total injected gas is lower in Ohmic only breakdowns than IC assisted ones, the density at breakdown is similar, figure 10(b).

The impact of impurities
Compared with Ohmic only assisted breakdown, IC assisted breakdown can initiate plasma in the presence of higher levels of impurities, such as Ne. This was demonstrated with a study of breakdown with D 2 fuelling in the presence of high invessel Ne concentrations. Ohmic breakdown was first attempted at E loop ≲ 0.33 Vm −1 . Despite three consecutive attempts   at different pressures around the optimal value, Ohmic breakdown was unsuccessful. Figure 11 shows one of the unsuccessful Ohmic only breakdowns (#100401, red) which was based on a reference successful Ohmic only breakdown (#98343, blue). Despite similar prefill fuelling, prefill gas pressure, and loop electric field to the reference at t = 0.45 s, the breakdown for #100401 is clearly un-sustained. IC assisted initiation #100404 was then attempted and was successful straight away. Figure 12 shows plasma parameters for the successful IC assisted breakdown discharge (#100404, red) against that of the previously successful Ohmic pulse (#98343, blue). The Ohmic reference has line radiation from Ne lines which are essentially at or below the background noise. #100404 has much higher neon levels, but these do not prevent plasma initiation. As no wall cleaning was attempted after the previous attempts (#100401-#100403), it is safe to assume that they would have had similar, high neon levels. The high neon content comes from the previous session which actively injected neon for experimental purposes. The failure of the Ohmic only breakdown and the success of the IC assisted breakdown for the same high Ne level vessel conditions shows that IC assisted breakdowns can tolerate higher impurity levels than Ohmic only ones in JET with the ITER-like wall.

The parametric dependency of the plasma current rise rate
The plasma current rise rate after breakdown is defined, for these studies, as the average of dI p /dt over the period between I p = 100 kA and I p = 200 kA. This is always before I p is in feedback control and is freely evolving. For both IC assisted and Ohmic breakdown, the plasma current rise rate after breakdown inversely correlates with the electron density at breakdown and positively correlates with E loop . This can be seen in figure 13 where breakdown n e /E loop is well, inversely correlated with the I p rise rate. Higher E loop drives faster I p rise rate. Higher n e is associated with lower T e for the same plasma energy, which results in higher resistance and so lower I p rise rate. For Ohmic breakdown, I p rises much faster, at similar pressure, than for IC assisted breakdowns. Comparing the breakdowns with the cryo-pump cooled by liquid N 2 and liquid He, there is no evidence of the divertor condition affecting the I p rise rate.
Although Ohmic breakdown happens at higher vessel pressure, n e is lower compared with IC assisted breakdown, as the examples (#100624 vs #100636) shown in figure 2. I p rises faster for higher B t for the same breakdown prefill gas pressure, figure 14(a). This is consistent with the observation that IC assisted breakdown pulses have higher electron density than Ohmic breakdowns at similar prefill gas pressures, section 3.4. There is no notable isotope effect. There is a strong, negative correlation between the I p rise rate and n e /E loop across the whole dataset, figure 14(b).

The impact of IC assisted breakdown on plasma impurities
IC heating on JET has been observed to be associated with impurity production [24][25][26][27][28][29]. The impurities concerned include Ni and Be, resulting from interactions with the Ni components of the IC antennas or the Be limiters around them, as well as other low-Z and high-Z impurities. For IC assisted breakdown pulses, no sign of increased impurities level in current ramp-up or later phases. Figure 15 shows an example IC assisted breakdown discharge and its Ohmic only breakdown reference. The lines shown cover the impurities commonly found in JET discharges which can contribute significantly to Bremsstrahlung and total plasma radiation. Plasma initiation for both discharges is successful. Line radiation from low   Z and high Z impurities and the Z-effective for the two discharges is similar throughout plasma burn through and later flat-top phase. There is no evidence of increased Ni or Be impurities.

Summary and implication of the results
For the first time on JET, a series of IC wave assisted breakdown experiments have been performed to provide a physics basis for validating models for testing and optimising plasma initiation with IC assisted breakdown for ITER. RF operation and protection, fuelling time and location, and the range of breakdown pressure and E loop were studied. The IC assisted breakdowns were compared with Ohmic breakdowns performed under similar conditions from previous experiments as well as dedicated Ohmic breakdowns performed as part of the experiment in conditions as closely matched as possible to the IC assisted breakdowns.
For the first time, both IC assisted and Ohmic breakdown were achieved in D for the regime planned for the first prefusion operation (PFPO-1) phase of ITER, E loop ≲ 0.33 Vm −1 with B t = 1.7 T and where EC breakdown assistance with 170 GHz waves would not work. All discharges, IC assisted or otherwise, achieved breakdown; successful plasma initiation was determined by the success of the subsequent plasma burn through phase.
In the studies, plasma initiation with IC assisted breakdown was significantly more robust than Ohmic breakdowns with the same E loop and B t . For the IC assisted breakdowns at the lowest fields explored, B t = 1.7 T and E loop ≲ 0.33 Vm −1 , breakdown is achieved at lower pressures for a wider range and with significantly higher amounts of injected gas for a wider range than Ohmic only references. Plasma initiation with IC assisted breakdowns was also achieved with higher impurity content than for Ohmic breakdown references. To get the same prefill gas pressure, IC breakdowns required considerably higher injected prefill gas than Ohmic breakdowns showing that the IC acts as a pump. Thus, IC assisted breakdown enables plasma initiation with a higher level and wider range of injected plasma prefill gas. As the injected prefill gas is the controlled parameter, this significantly improves the robustness of plasma initiation operationally.
Divertor fuelling on JET provides a more symmetric torus pressure than the more common mid-plane fuelling, enabling easier control and analysis of breakdowns. It is recommended that divertor fuelling is used for future breakdown studies and that the toroidal asymmetry is considered when validating breakdown models on mid-plane fuelled discharges. No obvious influence of cryo-pump condition on breakdown pressure range was observed.
Turning to the evolution of the breakdown and plasma burn through, the I p rise rate is well (negatively) correlated with the n e /E loop across the full dataset which includes Ohmic and IC assisted breakdowns and a wide range of B t , bulk isotope species, and pumped divertor temperatures. There is no evidence of increased impurity level for IC assisted breakdown in the current ramp-up and later phases.
As successful burn through determines the success of plasma initiation, combining IC pre-ionisation with IC assisted burn through and ramp-up phases may further extend the operational range. Such a recipe on JET requires the use of different antennas one or more tuned for the pre-ionisation plasma and one or more for the later phases. This is intended to be the subject of future experiments.
Ohmic plasma initiation on JET-ILW has previously been shown to be consistent with the DYON plasma burn through simulator based on a parametrised confinement time and impurity fluxes based on a wall-sputtering model [30,31]. The data from all the studies presented here will be provided as a basis for model validation for plasma initiation on ITER and other devices. However, the demonstration in the ITER relevant regime that IC assistance extends the access to breakdown without any evidence for enhanced impurities already gives significant confidence that its use on ITER will also be important for assisting breakdown in the pre-fusion operation phase.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).