1 With contributions from: M. Garcia-Munoz (University of Seville),Energetic particle diagnostics in present tokamaks and challenges towards a burning plasma M. Nocente1,2 1Dipartimento di Fisica, Università degli Studi di Milano-Bicocca, Milan, Italy 2Istituto di Fisica del Plasma, CNR With contributions from: M. Garcia-Munoz (University of Seville), M. Salewski (Technical University of Denmark) and the Milano neutron/gamma-ray spectroscopy group M. Nocente, International School of Fusion Reactors Technology, Erice, 28th April – 4th May 2017 1

2 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 2

3 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 3

4 Auxiliary plasma heatingproduces fast ions How do we raise the temperature up to 100 million degrees (≈10-20 keV) in a tokamak?

5 Fusion reactions produce fast ionsD + T  a + n MeV

6 Fast ion measurements D + T  a +n +17.6 MeVThe energy distribution of fast ions needs to be known for reliable operation of a thermonuclear fusion reactor D + T  a +n MeV a particles play a key role in the self sustainment of a fusion reactor fast ion acceleration (NBI or RF) needs to be assessed in in specific heating schemes to quantify efficiency of auxiliary heating (H, D, T, 3He, 4He accelerated ions) fast ions can drive MHD modes that may lead to their redistribution and losses 6

7 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 7

8 Charge exchange based diagnostics very importantEnergy ranges are different in mid size and large size devices Mid size devices (eg. ASDEX Upgrade) ≈ 100 keV (heating systems) Fusion products not well confined (negligible) Charge exchange based diagnostics very important Energetic ion + Cold ion Cold neutral Energetic neutral + Collision

9 Nuclear based diagnostics very importantLarge size devices (eg. JET or, even more, ITER) ≈ 100 keV NBI (JET); ≈ 1 MeV NBI (ITER) RF heating drives ions to the MeV range Designed to confine the fusion products (MeV range) Nuclear based diagnostics very important Neutron emission Gamma-ray emission + 9Be  n + 12C* 12C*  12C + g (4.44 MeV)

10 Charge exchange based diagnostics in mid size devicesNeutral particle analyzers A fast ion in the core undergoes a charge exchange reaction and becomes a fast neutral If the neutral does not ionize along its path, it can travel to the edge where its energy and charge to mass ratio are analysed NPA Path of neutral towards the edge Charge exchange reaction in the core Kislyakov A.I. et al. Fusion Science and Technology vol. 53 (2008) ch. 8 “Particle Diagnostics”

11 Challenges towards a high performance plasma: NPA measures the energetic ion pdf with ICRH heating Challenges towards a high performance plasma: Edge dominated signal; less events from the core Effect of neutron and gamma-ray background on the NPA particle detectors NPA system foreseen for ITER

12 Fast Ion D-Alpha (FIDA) diagnosticFast ions: helical trajectories Neutrals from neutral beam Charge-exchange reaction Fast ion becomes neutral Some emit a Dα-photon (n=3 2) at nm along the line-of-sight Doppler shift ~ line-of-sight velocity W. W. Heidbrink Rev. Sci. Instrum D727 (2010)

13 (which is not necessarily classical!)Red shifted tail in the spectrum is the FIDA signal Careful subtraction of background components essential FIDA can measure the fast ion energy distribution… … but also the fast ion profile (which is not necessarily classical!)

14 Collective Thomson Scattering (CTS)Resolved fluctuations along k ωd = ωs – ωi kd = ks – ki Next slide: So what does it actually measure? Description must account for all fluctuating quantities (n, E, B, j) Debye sphere λD < 1/kd D. Moseev et al. Plasma Phys. Control. Fusion (2011)

15 Components of a CTS spectrumModel CTS spectrum from ASDEX Upgrade Information on thermal ions contained at low frequency shifts (→ T and vrot) Information on fast ions contained at high frequency shifts. Incident (probe) frequency Even though it’s 3-wave mixing: Can still be helpful to think of it as Doppler shift. CTS spectra depend on projected fast ion velocities and many nuisance parameters: Ti, Te, ni, ne, B,… Main challenge: background noise and subtraction M. Salewski et al. Nucl. Fusion (2010)

16 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 16

17 Interaction of fast ions with instabilities can lead to redistribution and losses Alfvén waves naturally occur in plasmas. Shear Alfvén waves are especially important. In uniform plasmas, Alfvén waves are dispersionless (vf=vg=vA) Phase velocity 𝒗 𝒇 =𝝎/𝒌 Group velocity 𝒗 𝒈 =𝝏𝝎/𝝏𝒌 W. W. Heidbrink Physics Of Plasmas (2008)

18 Shear Alfvén waves in tokamaks have a complicated spectrumThe toroidal and poloidal periodicities make different branches of frequencies appear Modulations of B along the field line trajectory make a gap appear between the different branches Alfvén waves in the continuum are highly dispersive (w changes significantly with R): very difficult to excite. Not a concern! m=5 Wave packet m=4

19 However… Defects can introduce modes in the gap (eg. in solids, impurities introduce energy levels between the valence and conduction bands) There is a large number of “defects” in tokamaks (non monotonic q profile, coupling of poloidal harmonics, ellipticity, non circularity…) Wave packet Modes in the gap are little dispersive. When excited, they are hard to damp. Mode in the gap (Alfvén eigenmode) Main concern: fast ions can naturally excite Alfvén eigenmodes in the gap

20 Fast ion orbits and Alfvén eigenmodesFast ion orbits are described by poloidal and toroidal frequencies 93 keV trapped NBI ion at ASDEX Upgrade Poloidal (bounce) frequency wq Toroidal (precession) frequency wf

21 Resonance condition 𝝎=𝒏⋅ 𝝎 𝝓 −𝒑⋅ 𝝎 𝜽Fast ion orbital frequencies can match those of the eigenmodes in the gap and excite them Resonance condition 𝝎=𝒏⋅ 𝝎 𝝓 −𝒑⋅ 𝝎 𝜽 If this is matched, then there can be net power transfer 𝑬⋅ 𝒗 ⊥ between the wave and the ion. This can lead to a redistribution or loss of the fast ions. FIDA measurements of the fast ion profile in the presence of multiple Alfvén Eigenmodes at DIII-D. Flattening of the fast ion profile observed.

22 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 22

23 Fast ion loss detector (FILD) measure MHD induced lossesFILD measures the pitch-angle and energy of lost fast ions Large bandwidth allows measurements at Alfvén Eigenmode frequencies (~100kHz) – key for identifying coherent losses and impact of individual modes Local velocity-space measurements like these help to isolate fundamental mechanisms Installed in virtually all fusion devices. Conceptual design for ITER on-going (challenging project!) aperture ion Plasma Gyroradius fast ions Pitch angle 1S. J. Zweben et al, Nucl. Fusion (1988) 2M. Garcia-Munoz et al, Rev. Sci. Instrum (2009)

24 Typical FILD setup Safety &Simultaneous imaging of scintillator with double system CCD camera (slow but high spatial resolution) Array of 20 photomultiplier tubes (MHz - Alfvénic temporal resolution) ~ 100m Most modern FILD systems are installed on reciprocating arms following AUG design Safety & IR-Camera view

25 of the lost ion velocity spaceFILD measurements of the lost ion velocity space Typical CCD Scintillator Image Identification of loss orbit topology FILD M. Garcia-Munoz et al, Phys. Rev. Lett. 104, (2010) PMT signal (2 MHz sampling rate)

26 - Why fast ions in tokamaks?Outline - Why fast ions in tokamaks? - Confined fast ion diagnostics - Fast ion physics - Lost fast ion measurements - Nuclear diagnostics for fast ion measurements in high performance plasmas - Conclusions 26

27 Neutron emission from thermonuclear plasmasNeutron production Neutrons are produced by fusion reactions d +d  n + 3He d + t  n + a In a cold plasma (Ereactants ≈ 0) En = 2.45 MeV for DD reaction En = 14.0 MeV for DT reaction Neutron energy spectrum The neutron energy depends on the energy of the reactants

28 See lecture by M. Cecconello for more detailsNeutron Energy Fast ion contributions appear in the high energy tails of the neutron spectrum Neutron cameras measure the neutron emission profile and its changes (for ex. due to MHD activity) See lecture by M. Cecconello for more details

29 Gamma-ray emission from fusion plasmasg-rays are produced by nuclear reactions between fast ions and impurities They can be produced in fusion reactions (I step reaction) or result from the de-excitation of a nucleus (II step reaction) d + p → 3He + g (5.5 MeV) (fusion reaction) ii) a + 9Be → 12C* + n, 12C*→ 12C + g (4.44 MeV) (two step reaction) Ions need to be energetic for significant g-ray emission (E > 0.5 MeV typically): natural process for MeV range ions Examples: M. Tardocchi, M. Nocente and G. Gorini Plasma Phys. Control. Fusion 55 (2013) V. Kiptily et al. Nucl. Fusion 42 (2002) 999

30 Diagnostic information from g-raysGamma rays are emitted from several reactions in the plasma. Main advantage: spontaneous process Main disadvantage: indirect information on fast ions. Detailed modeling is needed for each gamma ray emitting reaction Diagnostic information of increasing detail level can be in principle obtained: i) Reaction assessment (peak energy identification) ii) Temperature of fast ions (spectroscopy), iii) Effect of MHD instabilities (count rate) 12C(3He,pγ)14N 9Be(3He,nγ)11C 9Be(3He,pγ)11B 14N 11C 11B 5104 4440 4440 SE 4440 DE 2313 2000 1635 40K 60Co g-ray spectra are often complicated and made up to many lines. Information on multiple ions simultaneously is possible but adequate modelling is necessary

31 Why do we need neutron and gamma-ray spectroscopy?In present middle size devices (eg. AUG, DIII-D, NSTX, MAST) detailed information on fast ions are provided by several diagnostics: confined ions: FIDA, NPA, CTS lost ions: FILD But... what about JET, ITER and beyond? n + g n + g n + g Neutron and g-ray fluxes Edge temperature Signal to background Others Others Others

32 Gamma Ray Spectrometers installed at JETLaBr3:high rate (MHz) at high energy resolution HpGe:very high energy resolution HPGe and LaBr3 ~4m M. Nocente et al. Rev. Sci. Instrum. 81 (2010) 10D321 Gammas & Neutrons More spectrometers coming online for the DT campaign; see A. Murari’s talk

33 GRS measurements of a fast a particle beamBeam of a particles accelerated with ICRH at the 3rd harmonic Stix distribution describes ICRH at 3w4He a + 9Be → n + 12C* (→ 12C + g) Ea > 1.8 MeV → 0 (Eγ = 4.44 MeV) Ea > 4.0 MeV → 4.44 (Eγ = 3.21 MeV) Parasitic absorbtion by D ions The expected energy distribution has a sharp cut-off (E*) corresponding to a decrease of the RF power absorption efficiency 33

34 The GRS peak shape changes with E*Spectrometer 4MeV a 2MeV a Ea = 3 MeV E* = 6 MeV Shape does change with E* Limited by statistics R LOS ICRH kp in agreement with values found from analysis of peaks from 12C(d,pg)13C M. Nocente et al, Nuclear Fusion 52 (2012) 34

35

36 Simultaneous measurements of the d and 4He ion profile from ICRH heatingV. Kiptily et al. NF 45 (2005) L21 4He and d have same q/m ratio 4He ions lead to 4.44 MeV gamma-rays through 9Be(4He,ng)12C d ions lead to 3.1 MeV gamma-rays through 12C(d,pg)13C The d and 4He energies responsible for this emission are related to resonances in the underlying cross sections

37 New compact detectors developed for gamma-ray imaging of fusion DT plasmas with high energy resolution and MHz capabilities Multi Pixel Photon Counter (silicon photomultipliers) + LaBr3 PMT MPPC are inherently insensitive to magnetic fields Need to separate signal from neutron background in DT plasmas D. Rigamonti, M. Nocente et al. RSI 87 (2016) 11E717 M. Nocente et al. RSI 87 (2016) 11E714 See poster by D. Rigamonti

38 High resolution neutron spectrometersITER plans to have a combined system of neutron and gamma-ray spectrometers Biological shielding Plasma RGRS Biological shielding RGRS RNC In-port detectors RNC Ex-port detectors HRNS RNC Ex-port detectors High resolution neutron spectrometers A diagnostic port that combines a) neutron camera b) gamma-ray camera c) high resolution neutron spectrometer system is being designed for ITER

39 Conclusions Fast ions must be studied to understand plasma heating but also because they can lead to instabilities which must be understood and controlled to achieve fusion performance In mid-size machines, diagnostics relying on charge exchange reactions are dominant and essential in the ≈ 100 keV energy range Fast ion losses can also be measured with edge probes High performance (burning) plasmas have fast ions predominantly in the MeV range. Nuclear processes emitting neutrons and gammas, albeit indirect, become essential as other methods experience limitations ITER will employ a combination of the above mentioned techniques (except FIDA). Redundancy is the keyword for reliability at ITER. 39

40 Thank you for your attention!

41 Interaction of fast ions with instabilities can lead to redistribution and losses Alfvén waves naturally occur in plasmas. Shear Alfvén waves are especially important. In uniform plasmas, Alfvén waves are dispersionless (vf=vg=vA) A tokamak is a periodic system, hence k║ must have a discrete spectrum Phase velocity 𝒗 𝒇 =𝝎/𝒌 Group velocity 𝒗 𝒈 =𝝏𝝎/𝝏𝒌 n,m: integers associated to the toroial and poloidal periodicity q: safety factor (function of radius) W. W. Heidbrink Physics Of Plasmas (2008) In a tokamak, Alfvén waves are dispersive: k║= k║(R)

42 these waves are very difficult to excite !The non uniform magnetic field introduces modulations to vA, hence to the refraction index n=c/vA Modulations lead to the appearance of gaps in the frequency spectrum (same as in solids: modulations of the potential lead to the gap between the valence and conduction bands) m=5 m=5 m=4 Modes outside the gap (continuum) are highly dispersive (vg changes dramatically with r): these waves are very difficult to excite !

43 Defects can introduce modes in the gapHowever… Defects can introduce modes in the gap (eg. in solids, impurities introduce energy levels between the valence and conduction bands) There is a large number of “defects” in tokamaks (non monotonic q profile, coupling of poloidal harmonics, ellipticity, non circularity…) Modes in the gap are little dispersive. When excited, they are hard to damp. Main concern: fast ions can naturally excite Alfvén eigenmodes in the gap

44 Diagnostic integration with the Radial Neutron CameraBiological shielding Plasma RGRS Biological shielding RGRS RNC In-port detectors RNC Ex-port detectors HRNS RNC Ex-port detectors High resolution neutron spectrometers ITER is big: we can accommodate the high resolution detectors we use at JET (3’’x6’’ LaBr3 and/or HpGe) on each of the multiple sightlines. At ITER, RGRS can combine energy, space and time resolution in one single system ITPA EP – June 2016

45 Expected gamma-ray spectrum at the detector4.44 MeV 3.21 MeV S/B at the 4.44 MeV peak ≈ 2 background mostly due to non signal gamma-rays Independent assessment of the neutron rate possible by integrating events in the Eg>10 MeV region from t(d,g)5He RGRS Technical Review Meeting 23/3/2017

46 Velocity-space observation regionsWeight functions w: Signal/ion Discretize: Reject data if w is is outside the target velocity space. covers the velocity space below 120 keV. M. Salewski, M. Nocente et al. Nucl. Fusion 57 (2017)

47 Velocity-space observation regionsTOFOR 5 detectors 3 typical Doppler shifts Diamond NE213 GRS keV GRS keV Low Medium High

48 Inversion by 1st-order Tikhonov regularizationForward problem: Inverse problem: Add Presentation Title in Footer via ”Insert”; ”Header & Footer” M. Salewski et al 2017 Nucl. Fusion

49 Velocity-space tomography vs. simulationBasic features in excellent agreement: Tail length, tail width. Barrier region suggests low densities above 2 MeV. Velocity-space tomography confirms the barrier experimentally. M. Nocente – PKU – Weight Functions for Neutrons and Gammas