Turbulence characterization during the suppression of edge-localized modes by magnetic perturbations on ASDEX Upgrade

We now consider the broadband density fluctuations which appear at a moderate perturbation field amplitude and study their toroidal localization within the RMP field pattern.

We use a technique introduced in [34] which is to rotate the current pattern in the RMP coils toroidally as a function of time and take measurements at toroidally fixed diagnostic positions. The resulting data can then be attributed to the toroidal phase of the perturbation. For a given poloidal and radial measurement position, this also corresponds to a particular field-line. Figure 6 gives an overview of discharge #34548, in which ELM suppression was reached with maximum RMP field amplitude at $t = {2.7}~\mathrm{s}$ and a differential phase angle $\Delta \varphi = {90}{^\circ}$. At $t = {3.0}~\mathrm{s}$ $\Delta \varphi$ is changed to ${135}{^\circ}$ and the amplitude is decreased. A further small decrease in amplitude occurs at $t = {5.0}~\mathrm{s}$. During periods of constant amplitude, the toroidal phase Φ₀ of the RMP field pattern is rotated, with two full rotation periods being passed through while ELM suppression is maintained. For each rotation period, the pedestal parameters remain reasonably constant. In the following we use predictions from VMEC 3D equilibrium calculations (described in [34, 35]) to map different diagnostics on the helical field-line label α. We follow the definition of the helical field line label used in [36] which is $\alpha = \theta^{*} \cdot q + \phi$ with the poloidal coordinate $\theta^{*}$ in straight field line coordinates, the safety factor q and the toroidal coordinate φ. The toroidal coordinate is time varying during the rigid rotation and can be written as $\phi(t) = \phi_\mathrm{0} + \phi_\mathrm{rot}(t)$ with $\phi_\mathrm{rot} = \pi \cdot f_\mathrm{rot} \cdot (t-t_\mathrm{start})$. Given the $q\mathrm{-profile}$, the poloidal and toroidal coordinates of a certain diagnostic $\theta^{*}$ and $\phi_\mathrm{0}$, and the rotation frequency of the n = 2 perturbation field $f_\mathrm{rot} = {0.5}~\mathrm{Hz}$, one can map measurements of different diagnostics during the rigid rotation onto α. In this work α was given an offset of ${-0.1}\,{\pi}$ as compared to [36] in order to move the two zero crossings of the surface corrugation to ${0}\,{\pi}$ and ${0.5}\,{\pi}$. In order to verify the use of VMEC in this discharge, a comparison between the measured and predicted surface corrugation is shown in figure 7. The radial displacement $\delta R$ of the plasma surface in the line of sight of the lithium beam [37] diagnostic is calculated by following the radially moving density profile at a constant value of ${1.2}{\times 10^{19}~\mathrm{m}^{-3}}$. A correction to the fit is applied which takes into consideration the effect of the plasma control system moving the plasma in radial direction due to the modulated signal in the magnetic probes used for real time control during a rigid rotation of the RMP field (described in [35]). It shows a good agreement between the measured surface displacement by lithium beam measurements and the predicted surface corrugation by VMEC.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Density fluctuations are measured by the O-mode reflectometry system with a perpendicular incidence angle and probing frequencies of ${37}~\mathrm{GHz}$ and ${18}~\mathrm{GHz}$ in the Ka- and K-bands, respectively. One advantage of O-mode reflectometry for this application is that the measurement location is tied to the O-mode cut-off layer which depends only on density, irrespective of the plasma deformation produced by the RMP field. Therefore, during the RMP field rotation the measurement position remains at a fixed position within the plasma, in this case the center of the ETB for the Ka-band, and the SOL for the K-band. The measurement of the Ka-band system on the LFS during both rotation periods is shown in figure 8 as a function of the helical field-line label α. The top panel shows the radial plasma surface displacement $\delta R$ in the line-of-sight of the Ka-band measurement as predicted by VMEC. The middle panel and the bottom panel show the integrated power spectral density and the spectrogram of the reflectometry homodyne signal, respectively. Over a wide range of α the reflectometry signal stays approximately constant, however, in a window between roughly $\alpha = {0.25}\,{\pi}$ and $\alpha = {0.55}\,{\pi}$ (${0.49}\,{\pi}$ in the second period) the PSD drops significantly by up to ${10}~\mathrm{dB}$. The minimum is located at approximately $\alpha = {0.41}{}$–${0.46}\,{\pi}$ which is close to one of the zero crossings of the surface displacement with a small shift towards the maximum of the surface displacement. This phase within the surface corrugation where the broadband fluctuations are reduced will hereafter be referred to as the 'fluctuation gap'. In the K-band measurement the fluctuation gap is also present, but only in a very narrow range in α where the minimum is also observed in the Ka-band measurement, at $\alpha = {0.41}{}$–${0.46}\,{\pi}$. Figure 9 shows spectra of the Ka-band measurement taken inside the fluctuation gap at $\alpha = {0.45}\,{\pi}$ (blue) and outside the fluctuation gap at $\alpha = {0.75}\,{\pi}$ (orange) for both rotation periods. The reduction of fluctuations can only be observed in the low frequency range up to about ${150}~\mathrm{kHz}$. Within this range, the magnitude of the reduction decreases with increasing frequency. At higher frequencies the spectra inside and outside the fluctuation gap overlap. On the HFS, such a toroidal asymmetry present in both rotation periods is not observed.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Turbulent transport perpendicular to the magnetic field arises from E × B-convection of turbulent eddies. This can be quantified experimentally if, in addition to the density fluctuations, perpendicular electrical field fluctuations can be measured, including the phase between the two quantities. Unfortunately, such a measurement is at present not available for the hot H-mode pedestal in ASDEX Upgrade. We therefore investigate the spatial and temporal correlations of the broadband density fluctuation in the ETB with other phenomena related to particle transport, i. e. (a) the formation of a density shoulder in the scrape-off-layer (SOL), and (b) particle fluxes into the divertor.

The appearance of a density shoulder, i. e. a radial region with an elevated flat density plateau in the near SOL, can be attributed to field-aligned plasma filaments ('blobs') which quickly travel radially outward and lead to increasing perpendicular transport [38]. Here, we use the lithium beam diagnostic, with a sightline located slightly above the midplane at the LFS, which measures the SOL density profile with high spatial accuracy and an integration time of ∼1 ms that averages over filament events.

On the left plot of figure 10, edge electron density profiles are shown for four different times (or values of the field line label α) during each of the two RMP rotation periods in shot #34548. Line colors correspond to the values of α marked up in the right plot, which shows the electron density at fixed flux radius $\rho_\mathrm{pol} = 1.035$ (dashed line in left plot) as a function of the α.

Zoom In Zoom Out Reset image size

The plasma density is essentially independent of α on flux surfaces in the confined plasma ($\rho_\mathrm{pol} \lt 1$) and the density profiles converge in the far SOL ($\rho_\mathrm{pol} \gt 1.08$). At radii in between, there is a clear formation of a density shoulder, for $\alpha \lt {0.3}\,{\pi}$ and $\alpha \gt {0.7}\,{\pi}$. The shoulder height drops significantly in the range of α = 0.4–${0.6}\,{\pi}$, which overlaps well with the location of the fluctuation gap at α = 0.25–${0.55}\,{\pi}$ (see previous section). We therefore conclude that the toroidally localized density shoulder is linked to the edge turbulence that we detect as density fluctuations in the gradient region at the same toroidal positions. The increased density in the near SOL can only be caused by an increased radial efflux from the confined plasma, as competing flows along the magnetic field drain plasma from filaments and reduce the SOL density. This suggests that the plasma edge turbulence associated with the observed broadband density fluctuations during RMP application causes radial transport across the ETB.

The second indication for particle transport comes from Langmuir probes mounted on the outer divertor which allow for local measurements of density fluctuations. From top to bottom, figure 11 shows the toroidal phase Φ₀ of the MP field, the divertor shunt current as well as the electron density and temperature measured by one of the divertor Langmuir probes. In the ELMy phase before $t = {2.7}~\mathrm{s}$, ELMs can be recognized in the Langmuir probe measurements as large spikes in n_e and T_e which originate from the transient ELM particle and energy bursts arriving in the divertor. Transient transport can still be observed during the ELM suppression time interval (between the two dashed lines). During the two rotation periods of the RMP field between t = 3 and ${7}~\mathrm{s}$ the density and temperature peaks in the divertor vary in amplitude. Phases with increased or reduced amplitude occur in both rotation periods at the same phase Φ₀ (top panel).

Zoom In Zoom Out Reset image size

Figure 13 shows spectrograms of the electron density measured by four radially separated Langmuir probes (8ua4, 8ua6, 8ua8, 8ua9) located at the same toroidal angle in the outer divertor with increasing distance from the strike point (see figure 12). In the spectrograms plotted as a function of time broadband fluctuations can be observed as well as fluctuation gaps during the rigid rotation. As compared to the reflectometer, the Langmuir probe measurement has a smaller bandwidth of ${11}~\mathrm{kHz}$. Therefore, the full frequency range of the broadband fluctuations observed in the ETB (up to ∼150 kHz) can not be covered, but existence or disappearance of fluctuations are well observed. Despite all probes being located in a narrow toroidal angle range, the fluctuation gap appears in each probe at different times. This effect can be explained by tracing the field lines starting at the various Langmuir probe locations up stream until the horizontal plane of the reflectometry measurement is reached. Here the field line tracing is based on the reconstruction of the magnetic force equilibrium for #34548 and the vacuum perturbation field. Since the small RMP field does not add significantly to the rotational transform, and the plasma beta in the SOL is small, the plasma response to the RMP can be neglected for this calculation. As shown in figure 14, the field lines originating from different probes intersect with the plane of the reflectometry measurement at significantly different toroidal angles. Hence, if particle transport through the ETB into the scrape-off layer is toroidally asymmetric at this plane, the asymmetry will be seen in each Langmuir probe shifted in time during the rigid rotation. If the data is mapped again onto the helical field line label (right column of figure 13) the fluctuation gaps align very well and appear at roughly the same α as the flucutation gap observed in the ETB via reflectometry (compare to figure 8). For sake of readability we only showed spectrograms for four different probes, but the fluctuation gaps are observed in all probes that are shown in figure 12 and align with the observed trends as well. The only exception is probe 8ua1 which is located in the private flux region and is hence not connected to the horizontal plane of the reflectometry measurement. Furthermore, while the fluctuation amplitude is comparable for 8ua2 to 8ua8, a significant drop of the fluctuation amplitude can be observed from probe 8ua8 to 8ua9. The flux surface intersecting with probe 8ua9 is at $\rho_\mathrm{pol} = 1.075$ and as shown in figure 10, this agrees with the outer boundary of the density shoulder in the SOL.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In addition to the non-axisymmetric fluctuations described above, several plasma parameters show characteristic non-axisymmetry which may have some bearing on the turbulence drive mechanism.

For shot #34548, we present and discuss measurements of the inverse gradient lengths of electron density and ion temperature as well as the radial electric field. Since the temperature measurements from Thomson scattering and ECE in the steep gradient region are less reliable, we do not include an analysis of variation in the electron temperature gradient. The density and temperature gradients pointing inwards are here defined positive.

4.3.1. Electron density and ion temperature.

On the left side of figure 15, contour plots of the inverse gradient length ${{1}}/{{L}}$ are shown for the electron density as a function of the radial position R and the helical field line label α during both rotation periods. Here, the dashed black line indicates the separatrix location which is assumed to be at constant density $n_\mathrm{e} = {1.2}{\times 10^{19}~\mathrm{m}^{-3}}$. Inside the fluctuation gap at around α = 0.4–${0.6}\,{\pi}$, a local maximum of ${{1}}/{{L_\mathrm{n_\mathrm{e}}}}$ is located just outside the separatrix (bright green/yellow area around $R = {2.1}~\mathrm{m}$). Since the density shoulder is absent in this toroidal region, the density gradient is steeper in those profiles at the foot of the pedestal resulting in the observed maximum of ${{1}}/{{L_\mathrm{n_\mathrm{e}}}}$. Furthermore, outside the fluctuation gap where strong broadband density fluctuations prevail, ${{1}}/{{L_\mathrm{n_\mathrm{e}}}}$ is smaller at the pedestal (dark blue area around $R = {2.075}~\mathrm{m}$) possibly resulting due to the increased radial particle transport in this toroidal region. On the right side of figure 15, profiles of ${{1}}/{{L_\mathrm{n_\mathrm{e}}}}$ are plotted for cases inside (blue) and outside (orange) the fluctuation gap together with the ELM filtered profile from a reference discharge (grey) without the RMP field (#33 998 at t = 2.8–${3.1}~\mathrm{s}$). Due to the absence of the density pump-out effect, the electron density in the reference discharge is about 3 times higher as compared to the ELM suppressed discharge. The inverse density gradient length in the reference discharge is comparable to the inverse density gradient length measured during the rigid rotation in ELM suppression or can even exceed it when compared to profiles outside the fluctuation gap (orange). Inside the fluctuation gap (blue), ${{1}}/{{L_\mathrm{n_\mathrm{e}}}}$ is about 1.4 times larger than the reference case.

Zoom In Zoom Out Reset image size

As seen in contour plots of the ion temperature on the left side of figure 16, the maximum of ${{1}}/{{L_\mathrm{T_\mathrm{i}}}}$ is located as well around α = 0.4–${0.6}\,{\pi}$ inside the fluctuation gap (bright yellow area). A comparison of profiles inside (blue) and outside (orange) the fluctuation gap to the reference case without RMPs (grey) is shown on the right side of figure 16. As opposed to the electron density, for the case of the ion temperature the inverse gradient length is increased throughout the whole surface corrugation pattern at least slightly as compared to the no-RMP reference and at its maximum is increased by a factor of roughly 2.5.

Zoom In Zoom Out Reset image size

High absolute inverse gradient lengths of temperature are thought to destabilize turbulent transport, however, here we find the highest inverse gradient lengths for the ion temperature in the α range of the fluctuation gap where the density fluctuations disappear. A more natural interpretation of this observation is that the broadband turbulence produces additional radial particle and heat flux which lead to a flattening of the density and ion temperature profiles outside the fluctuation gap.

4.3.2. Radial electric field.

The radial electrical field at the plasma edge is determined from the radial force balance of impurity ions i (i is $B^{5+}$ in this case):

$$\begin{equation} E_\mathrm{r} = \frac{1}{e n_\mathrm{i} Z_\mathrm{i}} \frac{\partial p_\mathrm{i}}{\partial r} - v_\mathrm{\phi,i} B_\mathrm{\theta} + v_\mathrm{\theta,i} B_\mathrm{\phi} \end{equation}$$

with the toroidal rotation $v_\mathrm{\phi}$, poloidal rotation $v_\mathrm{\theta}$ and pressure p of boron $B^\mathrm{5+}$ measured by Charge Exchange Recombination Spectroscopy (CXRS) [39] as well as the radial coordinate r, elementary charge e, ion charge Z, toroidal $B_\mathrm{\phi}$ and poloidal $B_\mathrm{\theta}$ magnetic field. Figure 17(a) shows contours of the radial electric field $E_\mathrm{r}$ as a function of major radius R and field line label α. Since the ion temperature gradient contributes significantly to $E_\mathrm{r}$, the strongest $E_\mathrm{r}$ well is located at the same α range as the extremum of the inverse ion temperature gradient length (see figure 16), which also coincides with the toroidal position of the fluctuation gap. At α ≈ 0.45–${0.65}\,{\pi}$ a maximum depth of the $E_\mathrm{r}$ well of nearly ${80}~\mathrm{kV}~\mathrm{m}^{-1}$ is reached, which is 3 times larger than in the reference case without RMPs. In figure 17(b) two $E_\mathrm{r}$ profiles outside ($\alpha = {0}\,{\pi}$, blue) and inside ($\alpha = {0.46}\,{\pi}$, orange) the fluctuation gap are shown, and in figure 17(c) the corresponding $E_\mathrm{r}$ shear ($\omega^{^{\prime}}_\mathrm{ExB} = \frac{r}{q} \frac{\partial}{\partial r}(\frac{q}{r}\frac{E_\mathrm{r}}{B})$ [40]) profiles are plotted. Not only is the $E_\mathrm{r}$ well deeper in the region of the fluctuation gap, but also narrower. This in combination leads to a strongly increased positive and negative $E_\mathrm{r}$ shear on the outside and the inside of the $E_\mathrm{r}$ well, respectively.

Zoom In Zoom Out Reset image size

4.3.3. Turbulence drift velocity.

We now compare the E × B drift velocity $v_\mathrm{ExB}$ obtained from the CXRS analysis with the perpendicular velocity $v_\mathrm{\perp}$ of turbulent structures, as measured by the Doppler shift of microwaves diffracted at the corrugated cut-off layer with oblique incidence ('Doppler reflectometry', DRFL). As explained in [41], $v_\mathrm{\perp}$ can be interpreted as the Galileian superposition of the E × B drift velocity with a phase velocity $v_\mathrm{ph}$ of a drift mode in the $E_\mathrm{r} = 0$ reference frame. Since the phase velocity is often negligible, DRFL is often used to obtain an estimate of the $E_\mathrm{r}$ profile. When the phase velocity contributes significantly to $v_\mathrm{\perp}$, a comparison between CXRS and DRFL can be used to estimate a phase velocity, however, with large uncertainty. Figure 18 shows the $E_\mathrm{r}$ profiles as measured by CXRS (blue) and approximated by $v_\perp B$ from DRFL (orange) at two different phases α within the surface corrugation. For the evaluation of the DRFL measurement during the rigid rotation we used density profiles at the same α to calculate the backscattering location and we used poloidal cross-sections of the 3D equilibrium in order to calculate the probed wave number. In the case of $\alpha = {0.85}\,{\pi}$ both measurements show good agreement especially in the $E_\mathrm{r}$ well (minimum). On the other hand, the case of $\alpha = {0.55}\,{\pi}$ shows a strong discrepancy between $E_\mathrm{r}$ and $v_\perp B$, especially in the depth of the $E_\mathrm{r}$ well.

Zoom In Zoom Out Reset image size

While the $E_\mathrm{r}$ profile (as measured by CXRS) still shows the pronounced deep minimum with steep flanks, the $v_\perp B$ profile measured by DRFL shows a very flat minimum over a broad radial range, $\rho_\mathrm{pol} = 0.97 - 1.0$. The minimum of the CXRS $v_\mathrm{ExB}$ profiles is tracked during both rotation periods and plotted against α in figure 19(a) (blue). For the same radial positions as the $v_\mathrm{ExB}$ minima measured by CXRS also $v_\mathrm{\perp}$ measured by DRFL is plotted (orange). One can see that they agree only in a small region within the surface corrugation at around $\alpha = {0.75}\,{\pi}$ and differ at all other values of α. Following [41], this can be attributed to a finite phase velocity of the turbulence within the plasma frame (which moves with $v_\mathrm{ExB}$). In figure 19(b) a phase velocity $v_\mathrm{ph} = v_\mathrm{ExB} - v_\mathrm{\perp}$ at the same radial positions is plotted (orange) against field line label α. For comparison, we overlay half the main ion diamagnetic velocity $v_\mathrm{i,dia}$ (blue), evaluated using the CXRS T_i measurement. According to our measurement, the turbulence propagates in the ion diamagnetic drift direction with a significant phase velocity of up to ${25}~\mathrm{km}~\mathrm{s}^{-1}$. Roughly, it amounts to about ${25}{\%}$ of the ion diamagnetic velocity and varies in the same way, except for the region around $\alpha = {0.75}\,{\pi}$ where it vanishes. The backscattering of the microwave probing beam may be due to complex fluctuation structures resulting from the applied RMP field and not just linear flows. Therefore, a derived phase velocity might not be interpreted as the phase velocity of a simple turbulent structure. Nevertheless, the rotation direction of the underlying turbulent structures in ion direction with a substantial velocity is still identified.

Zoom In Zoom Out Reset image size

Broadband density fluctuations are observed in the ETB during ELM suppression at moderate RMP field amplitude. A toroidal scan of these fluctuations reveals a region of reduced density fluctuations at one zero crossing of the surface displacement induced by the RMP field, dubbed the 'fluctuation gap'. Two specific observations point at the relevance of the broadband mode for radial transport: (a) the appearance of a density shoulder in the near SOL, at all toroidal positions of the broadband mode, i. e. everywhere outside of the fluctuation gap, and (b) the observation of correlated particle and energy fluxes along field lines in the divertor, again with the exception of the fluctuation gap. Hence, the broadband density fluctuations are contributing to the density pump-out induced by the RMP field. The fluctuation gap region is also special in that the $E_\mathrm{r}$ well in the H-mode ETB is deepest there and a turbulence phase velocity (in ion diamagnetic direction, as obtained from $v_\mathrm{\perp}$ measurements by Doppler reflectometry) has a maximum.

The QCM is present in the ETB region when the applied perturbation field amplitude exceeds a certain critical value. The radial localization of the QCM in the ETB is determined by reflectometry measurements in stationary ELM suppression phases at maximum RMP field amplitude and $\Delta\varphi = {90}{^\circ}$. In order to cover most of the ETB region, varying probing frequencies were used in discharges #35310 (2.9–${3.2}~\mathrm{s}$), #35943 (3.3–${3.5}~\mathrm{s}$) and #36514 (3.0–${3.3}~\mathrm{s}$). These discharges all utilize the same scenario in terms of toroidal field, plasma current, shaping and heating just as described in section 2. The electron density profiles of those phases are plotted in figure 20, with QCM observations coded by symbol color. Again, red color indicates that the QCM was not observed while green color indicates that the QCM was observed at that position. From these measurements the QCM is localized in the steep density gradient region in the ETB up to the pedestal top. It is not observed in the far SOL where the density profile is flatter. This observation indicates that the drive of the QCM may be related to the steep gradient in the ETB.

Zoom In Zoom Out Reset image size

Furthermore, the QCM is not only localized in radial direction, but also in the toroidal direction. In order to investigate the toroidal localization of the QCM, the RMP field is rotated in toroidal direction. ASDEX Upgrade has eight RMP coils in each of the two poloidally separated toroidal arrays. The large n = 2 RMP field amplitude required for the QCM is obtained if all coils are used at or near their maximum coil current, and the coil current polarity alternating after every two coils. This however means that the maximum RMP field amplitude can be achieved at only four discrete toroidal orientations separated by ${90}{^\circ}$ phase within the n = 2 perturbation field. These four orientations are scanned in the ELM suppression discharge #35862 shown in figure 21. When the RMP field is rotated to the next orientation, the RMP field amplitude drops and slowly recovers leaving 400–${500}~\mathrm{ms}$ long stationary phases for each orientation. In each of these phases, the ELMs are fully suppressed and the electron density becomes stationary again. However, in the ${270}{^\circ}$ orientation the electron density saturates at a slightly higher value as compared to the other orientations. For each of those four highlighted phases, the power spectral density of the reflectometry measurement taken radially at the center of the ETB is plotted with the respective color in figure 22. The QCM can only be observed at the orientation $\phi_\mathrm{0} = {0}{^\circ}$ (red color). However, due to the finite step width of Φ₀ in this experiment, and due to an unnoticed polarity reversal in one of the sixteen RMP coils at the time of this experiment (#35310, 35521, 35862, 35943, 36514) which pollutes the RMP spectrum, we cannot determine more precisely the localization with respect to the plasma surface corrugation pattern. This will be further discussed in section 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In order to investigate whether the QCM is causing radial particle transport, the RMP coil current amplitude $A_\mathrm{eff}$ is modulated sinusoidally during ELM suppression in shot #35521 (see blue line in the top panel of figure 23). Here, the field modulation is performed with a frequency of ${2}~\mathrm{Hz}$ and—accounting for the image currents in the PSL—the peak-to-peak current modulation amplitude is roughly ${270}~\mathrm{A}$. During the RMP amplitude modulation, the spectrum of the reflectometry measurement in the center of the ETB (top panel) shows transitions back and forth between states with broadband density fluctuations and with QCM. The QCM is centered around 40–${60}~\mathrm{kHz}$ (encircled in the figure, and not to be confused with a simultaneously occurring core tearing mode at around 20–${30}~\mathrm{kHz}$). During the RMP amplitude modulation also the electron density is modulated as seen in the middle panel. Here, the dark gray shaded time periods mark the phases in which the QCM can be observed and the light gray regions indicate the temporal uncertainty of these phases. The electron density reacts non-linearly to the RMP amplitude modulation. At high RMP amplitude, when the QCM is present, the density is low and varies little in time. When the QCM is lost due to decreasing amplitude the density starts to rise sharply, and decreases only after the RMP amplitude is raised again. This behavior indicates that the QCM participates in producing outward directed particle transport. The bottom panel of figure 23 shows density fluctuations measured at the outer divertor by an exemplary Langmuir probe. Even though none of the available Langmuir probe measurements (bandwidth ${11}~\mathrm{kHz}$) is fast enough to capture the peak frequency of the QCM and none of the probes is connected via a field-line to the line of sight of the reflectometer, an increase of the lower frequency spectrum can be seen every time the QCM is present in several probes. This observation supports the role of the QCM for particle transport into the divertor.

Zoom In Zoom Out Reset image size

If, in ELM suppression, the RMP field amplitude exceeds a threshold, the observed edge turbulence changes character and assumes a narrower bandwidth, so that it can be termed a 'quasi-coherent' mode. This mode is localized radially within the ETB and toroidally within the surface corrugation induced by the MP field. Simultaneously with the appearance of the QCM, a second pump-out occurs with the effect of decreasing the density further. Furthermore, density fluctuations measured on the outer divertor increase in presence of the QCM and the pedestal density increases when the mode disappears. Both observations indicate that the QCM is causing turbulent radial particle transport at least contributing to the'second density pump-out'.