Experimental studies of C-Mod ICRF minority tails via a multi-channel compact neutral particle analyzer

APS-DPP 2005

V. Tang, R. Parker, J. Liptac, P. Bonoli, R. Harvey*, C. Fiore, R. Granetz, J. Irby, Y. Lin, E. Marmar, A. Parisot, S. Wukitch, K. Zhurovich, MIT PSFC, Cambridge, MA

*CompX, Del Mar, CA.

Abstract• Recent experimental results from the upgraded multi-channel Compact Neutral

Particle Analyzer (CNPA) system are presented and compared with numerical simulation. The CNPA system uses Si diode detectors to measure charge exchange (CX) induced energetic (>50 keV) neutral particles for ICRF hydrogen minority tail distribution diagnosis. Currently, the diagnostic is operated primarily in passive mode. The discussed data involve measurements of the line-integrated Hydrogen-minority energy spectrum during low to moderate (nel~2x1020/m2) density ICRF plasmas with up to ~3 MW of power and BT from 4.4 to 6.2T. During these discharges, evidence of fast ion resonance localization and tail temperatures greater than 100keV is observed. The energy spectra are compared with a newly implemented synthetic diagnostic based on CQL3D Fokker-Planck simulations. Discrepancies between the experiment and simulated spectra are discussed.

Motivation• Ion-cyclotron RF(ICRF) minority heating is the main auxiliary heating scheme on

Alcator C-Mod.

• In this scheme, the RF power heats the minority species in vPerp, creating an anisotropic energetic minority ion distribution, or tail, which slows down on the bulk plasma.

• 3 Antennas have provided up to 5 MW of on-axis power into D(H) plasmas at 5.4T and D(He3) plasmas at 8.0T.

• Historically, the C-Mod minority species temperature is inferred from electron temperature measurements1 and TRANSP2 analysis. These estimates give temperatures up to ~300keV.

• The CNPA measures CX neutrals stemming from this hot ~on-axis ion distribution. These measurements give the tail distribution directly and are useful for verifying RF physics and heating deposition. The diagnostic is also potentially useful in studying interactions of such fast ions with TAE and other Alfvén gap modes3.

Introduction• The recently upgraded four-channel CNPA system4 utilizes fast digitization techniques

and silicon diode detectors to measures charge exchange (CX) induced energetic (>50 keV) neutral particles for ICRF heated hydrogen energy distribution diagnosis.

• Three of the channels view the plasma vertically, with the remaining channel mounted horizontally. These view cords allows the system to simultaneously measure the anisotropy of the energetic hydrogen species.

• Currently, the detectors are operated primary in passive mode, relying on the plasma background neutral density for CX. Active experiments will resume after the new Alcator long-pulse DNB passes its shakeout period.

• These line-integrated measurements indicate significant resonance localization and anisotropy of the fast ions.

• A synthetic NPA diagnostic is implemented for CQL3D in order to compare these measurements with ICRF heating theory. The implementation is native to CQL3D and will be available for the general community.

The CNPA Design and Hardware• Traditional fusion NPAs are based on mass

spectroscopy principles. They give good energy resolution and can separate ions with different q/m ratios. However, they are physically very large per channel.

• One compact alternative, employed by the CNPA, is to operate silicon diodes in pulse-height analysis (PHA) mode and detect the neutrals directly5. This method can not identify particle type easily but can still give decent resolution. Multiple channels can easily fit on one port. Because they have a direct view of the plasma, these systems are more susceptible to noise.

CNPA Sightline and Setup

The F-port top threechannel Si array

F-Port vertical detectors(Ch1-3), Ro~65 to 70 cmDetectors are ~3.5m from plasma mid-plane

Al EMF shield houses detector array, preamps, and shaping amplifier

C-Port horizontal detector at mid-plane

C-Port 1mm2 Si diode detector (Ch4) w/Al filterOn SMA connector

Alcator C-Mod Cross SectionMachine Ro=66.8cmMagnetic Ro~69cm

Isolation bellows and micrometer aperture

•Both detectors system see only particles with vpar~0 in their respective viewing cones

*An additional fifth channel is intermittently available and not shown or discussed here.

CNPA Specifications and Features• F-Port Top Vertical Detectors

– 3 x 1mm2, ~25µm thick AXUV-3ELA Si detector array.

– Cremat CR-110 Preamps w/ in-house 6-pole Gaussian shaping amplifier on integrated motherboard and EMF housing.

– 1500Å Al filter for light and low energy particle shielding.

– A micrometer is used for aperture selection and collimation.

– D-TACQ ACQ216CPCI 14bit 10 to 40Mhz +/-10V Digitizer Card w/3sec memory for storage and post-shot pulse-height analysis.

• C-Port Horizontal Detector– 1mm2 ~35µm thick AXSUVHS5

detector – Ortec 142 Preamp with Ortec 572

Shaping amplifier.– 1000Å Al filter for light and low

energy particle shielding.– A 2.5” long passivated tube is used

for collimation and aperture – D-TACQ ACQ216CPCI 14bit 10

to 40Mhz +/-10V Digitizer Card w/3sec memory for storage and post-shot pulse-height analysis.

CNPA System• Detector electronics follows an analog approach with digitization and post-shot PHA6.• PHA is traditionally done in real-time. The main limitations of such analysis are pulse

pile-up, baseline shift, and the constraints of real-time analysis. By storing the amplifier voltage and doing post-shot PHA, more complex PHA routines could be employed. Additionally, since the voltage discriminator and bin settings are determined in software, different values could be tailored to each spectrum. Finally, electronic pickup noise can be easily discerned.

CrematCR-110 preamp

In-house GaussianShaping amplifier

D-tacq10-40MhzDigitizer Card

Storage and in-house software PHA (Post-Shot)

Pulse Shapes

Si Diode Detector

Energy

Counts

Pile-up Normally, this shaped pulse in a traditional real-time PHA system would be rejected. If the data is saved and analyzed later, a fitting routine could be used to extract the voltage peaks.

Detector Characteristics and Operation• For the 2005 campaign, the diagnostic was successfully upgraded to four channels with

integrated electronics and significantly thinner detectors. This essentially eliminated the plasma induced background noise from neutrons and gammas and allow the use of the CNPA in passive mode without background subtraction.

• The system count rate is limited by soft x-rays which do not result in detectable counts. These soft x-rays increases the baseline fluctuations which results in a decrease in detector resolution for the neutral particles. Typically, the soft x-rays in moderately heated (~2MW) low density plasmas (nel~1020/m2) limits the system to counts rates of <10k/sec for each channel at ~20keV resolution.

Extracting the Minority Distribution• With line-integrated measurements, the CNPA spectrum can be modeled based on the

following equation7:

• The on-axis minority energy distribution can not be easily extracted without some spatial localization of the signal, for example, through the use of a DNB. The signal could also be naturally localized if the power deposition is very peak and transport small.

• Otherwise, the distributions must be inferred by solving the above equation and finding the best match to the experimental line-integrated data.

• This approach is used here via a new synthetic NPA diagnostic for the CQL3D Fokker-Planck code8.

. to from travelscord viewingThe density. neutral background the)( and

path, freemean inverse the),( section,-cross CX the)( on,distributiminority the),( etendue, the

ty,ion velocifast thedetector, thehitting secondper dEE toEEnergy with particles offlux theis )( where

)()()(),(),(exp )( 2

aaxan

ElEcxxivifAivdEEF

dxxnEvExvfdlElAdvvdEEFa

aaicxii

a

xii

−

Ω

+

⎥⎦

⎤⎢⎣

⎡−Ω= ∫ ∫

−

ασ

σα

CQL3D

• General purpose Fokker-Planck computer code for modeling auxiliary heating in tokamks.

• Currently coupled to GENRAY for wave propagation data.

Sample CQL3D fast ion distribution

In mid-plane velocity coordinates

Trapped passing boundary

Cut of v-space seen by C horizontal port detector

Cut of v-space seen by F port vertical detectors

•Sample phase space contour plot in mid-plane coordinates of a low power density Alcator C-Mod ICRF heated minority distribution from CQL3D, shown here to illustrate qualitatively the different types of spectra expected for the C and F Port detector system.

“Rabbit ears” from resonance localization-stems from anisotropic heating and tokamak magnetic geometry.

2005 Campaign Results Summary• New detectors verified to have minimum response to plasma neutron and gamma

background.

• Fast ion resonance localization phenomenon seen on previous machines now documented in C-Mod.

• Very energetic line-integrated spectra are detected.

• CQL3D synthetic passive NPA diagnostic spectra are too soft to match the experimental data. The differences increase with increasing spectrum energy.

• These deviations at higher energies are passably due to a broader than calculated power deposition profile, resulting in larger than calculated banana widths, or errors in the assumed background neutral densities.

6e13Neutrons(/s)

Gammas(/s)

Soft X-rays(kW/m2)

Density(/m2)(line integrated density)

2e20

4e13

20

Ch1 (R~66cm)

Ch2 (R~67cm)

Ch3 (R~68cm)

50keV

50keV

50keV

Detector Response to Gamma and Neutron Background

•Background data taken with gate valve closed on several days. Almost no response except for very high performance shots.

•Basically detectors are not susceptible to neutrons and hard x-rays (Detectors are ~25um thick; mfpof 20keV photon is 1000um in Si or 95% transmission)

•Soft x-rays increases baseline noise but does not result in pulse.

•Left: Shot 1050705018 (A shot with very high neutron background)

t=0 t=1.5s

Raw CNPA amplifier voltage; each pulse indicates a count.~4MW of RF was applied from t~0.8 to 1.25s

•Fast ion resonance localization9: one aspect of anisotropic distribution and tokamak geometry.•As ions are heated in vperp each time they hit the RF resonance, they go from passing to trapped. Their banana tips are thus localized near the resonance.•3 Channel F-port top system should see highest signal during shots where the banana tips and RF resonance are in the viewing cone, since vpar~0 there.

R=67cm (~5.4T)R=62cm (~5T) R=74cm (~6T)

Fast Ion Resonance Localization Physics

ResonanceLocation:

Polodial Cross section of 200keV C-mod fast hydrogen ions orbits.

Light blue band indicates detector view coneDashed blue line indicates RF resonance location

Resonance Localization Experimental Results• As expected, only shots with the heating

resonance in the detector view cones (5.4T) resulted in significant counts, indicating that most of the fast ions have their banana tips localized around the RF heating resonance.

• Shots are of various currents (400kA-1.5MA), BT(4.4-6.2T), nel(~2x1020/m2) and RF power (<3MW).

• 5.4T counts are lower for higher current shots. For 1.2MA the rate goes down to ~45 counts/shot.

• Ratio of counts/field for constant currents stays roughly the same. (10% for 6.2T/5.4T)

Counts w/ E >70keV vs. Field, 1050624

0

50

100

150

4.4T 5.4T 6.2T

B(T)

Ave

rage

Cou

nts/

shot

7108Number of shots used for averaging at each field

Comparison with CQL3D• Analysis is ongoing for a series of L-mode discharges (1050818) that scan density from

nel~0.5-1x1020/m2 and RF power from 1 to 2.5MW.

• Here, shot 1050818005 is shown in detail with its preliminary CQL3D simulation.

• Currently, the CQL3D synthetic NPA is setup for passive simulations with arbitrary sightlines. Orbit effects such as finite banana widths are important for accurate passive NPA simulations since the background neutral density can vary dramatically over these widths. These effects will be implemented into the synthetic diagnostic in the near term. For now, they are accounted for in an ad-hoc manner via banana (2 x v_par/ω_pol) and gyro-orbit width estimates.

• After additional testing and validation, the synthetic NPA will be included in the CQL3D package.

• The background neutral density profile is taken from TRANSP, based on the FRANTIC code. In the near future, KN1D10 could be used to provide the required neutral density profile.

Shot 1050818005•L-mode low density (neo~1.5x1020/m3) limited shot with 78Mhz, ~1.2MW RF power. H/D is estimated at 3%+/-1%.

•Yellow band (t=0.8 to 0.9s) indicates steady-state period where CNPA spectra are summed for comparison with CQL3D simulation.

•No CNPA signal is detected before RF is switch on. The CNPA signal decays exponentially after RF is turned off.

(Counts above ~70keV)

CNPA Spectra for shot 1050818005

•The Cut off here is caused by the voltage limit on the shaping amplifiers.

•Ch1 to Ch3 data shows similar slopes with the strongest spectrum at Ch2.

•The C-Port channel, as excepted, has the weakest spectra.

Energy-error bar: 10%Baseline noise: ~15keV

CQL3D Results for 1050818005

• Sharp peaks in the power deposition reflects discrete nature of ray-tracing. • Peaks in the power deposition results from multiple cycles of fast wave reflections and

damping within the plasma.• Phase space distributions clearly show resonance localization of the fast ions at the RF

resonance.• Vnorm corresponds to 2.5MeV Hydrogen ions.

Sample mid-plane distributions, R=74cm

Nphi~+/-11 to 15,w/ peak at +/-13

Choice of wave solver important

• CQL3D power deposition is narrow compared with either TORIC or SPRUCE.• More accurate Nphi modeling for GENRAY using multiple modes could broaden the

profile.• Minority temperature are better matched, although TORIC/FPP is generally broader.

Profile details are important since background neutral density varies rapidly near the edge and make banana width effects significant for high energy CX.

Power Deposition, Various Solvers-shot 10050818005

0

5

10

15

20

25

30

35

40

45

50

68 73 78 83 88Major Radius(cm)

Pow

er d

epos

itio

n(W

/cc) TORIC/FPP

SPRUCE/FPPGENRAY/CQL3D

Temperatures, Various Solvers-shot 1050818005

0

50

100

150

200

250

300

68 73 78 83 88Major Radius (cm)

Tem

per

atu

re(k

eV)

TORIC/FPPSPRUCE/FPPGENRAY/CQL3D

Simulated NPA data and comparison• Predicted spectra with ad-hoc orbit

effects are generally too soft.

• The significant deviations at higher energy could indicate a broader than predicted temperature profile, since the energetic fast ions must be encountering background neutral densities about an order of magnitude higher than predicted.

• This effect could be simulated in CQL3D by using spatial diffusion.

• Comparisons of absolute experiment and simulated spectra should provide another constrain on the fitting process. 1.E+06

1.E+07

1.E+08

1.E+09

1.E+10

1.E+11

68 73 78 83 88

Major Radius(cm)

D N

eutr

al D

ensi

ty (

/cc)

FRANTIC neutral output

•2 x (v_par/ω_pol) for banana width and assume most of the signal is from R~72cm

Conclusion and Future Work• The recently upgraded and now multi-channel CNPA system is operational for both

passive and active CX experiments on low to moderate density Alcator C-Mod plasmas. The four channel system allow simultaneous line-integrated distribution measurements of the ICRF heated hydrogen minority species.

• Evidence for resonance localization of the fast hydrogen ICRF heated minority species is documented for C-Mod plasmas in a wide parameter range.

• Very energetic passive NPA spectra are recorded for low to moderate density C-mod plasmas with ~1-2MW of ICRF heating.

• A new natively implemented synthetic NPA diagnostic is written for CQL3D. Simulated NPA spectra are compared with experimental results. Further modeling is required to determine the sensitivity of the line-integrated spectra to profile details and hence the errors associated with inferring the minority distribution function via the synthetic diagnostic approach. Careful comparisons between the absolute spectra should provide an additional constrain.

References1) O’Shea, P., “Measurements of ICRF power deposition and thermal transport with an ECE

grating polychromator on the Alcator C-Mod tokamak”, PhD thesis, 1997.2) Onega, J., Evrard, M., and McCune, D., Trans. Fusion Technology. 33, 181, 1998.3) Snipes. J., Sears, J., Gorelenkov, N. , Session KP1, “A Comparison of Measured and

Calculated TAE Damping Rates in Alcator C-Mod”, Denver, 47th APS-DPP, 2005.4) Tang, V. et al. , “Initial Experimental Results from the Alcator C-Mod Compact Neutral

Particle Analyzer”, 46th APS-DPP, 2004.5) Osakabe, M., Yamamoto, T., Takeiri, Y. et al. Review of Scientific Instruments, 72, No.1,

2001.6) O’Connell, R. et al. Rev. Sci Instrum. 74, No. 3, 2001.7) Hutchinson, I.I., Principles of Plasma Diagnostics, Cambridge University Press, Cambridge,

2nd Ed., 2002.8) Harvey, R.W. and McCoy, M.G. IAEA TCM, 1992. 9) Hammett, G., “Fast ion studies of ion cyclotron heating in the PLT tokamak”, PhD thesis,

1986.10) LaBombard, B., “KN1D: A 1-D Space, 2-D Velocity, Kinetic Transport Algorithm for

Atomic and Molecular Hydrogen in an Ionizing Plasma”, PSFC-RR-01-3.