Spectroscopic imaging diagnostics for burning plasma experimentsD. Stutman, M. Finkenthal, G. Suliman, K. Tritz, L. Delgado-Aparicio et al. Citation: Rev. Sci. Instrum. 76, 023505 (2005); doi: 10.1063/1.1852317 View online: http://dx.doi.org/10.1063/1.1852317 View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v76/i2 Published by the American Institute of Physics. Additional information on Rev. Sci. Instrum.Journal Homepage: http://rsi.aip.org Journal Information: http://rsi.aip.org/about/about_the_journal Top downloads: http://rsi.aip.org/features/most_downloaded Information for Authors: http://rsi.aip.org/authors

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

Spectroscopic imaging diagnostics for burning plasma experimentsD. Stutman,a! M. Finkenthal, G. Suliman,b! K. Tritz, and L. Delgado-AparicioDepartment of Physics and Astronomy, Johns Hopkins University, Baltimore, Maryland 21218

R. Kaita and D. JohnsonPlasma Physics Laboratory, Princeton University, Princeton, New Jersey 08543

V. Soukhanovskii and M. J. MayLawrence Livermore National Laboratory, P.O. Box 5508, Livermore, California 94551

sReceived 10 September 2004; accepted 29 November 2004; published online 20 January 2005d

Spectroscopic imaging of plasma emission profiles from a few electron volts to tens of kilo-electronvolts enables basic diagnostics in present day tokamaks. For the more difficult burning plasmaconditions, light extraction and detection techniques, as well as instrument designs need to beinvestigated. As an alternative to light extraction with reflective optics, we discuss normal incidence,transmissive-diffractive opticsse.g., transmission gratingsd, which might withstand plasma exposurewith less degradation of optical properties. Metallic multilayer reflectors are also of interest for lightextraction. Although a shift of the diffraction peak might occur, instrument designs thataccommodate such shifts are possible. As imaging detectors we consider “optical” arrays based onconversion of the short-wavelength light into visible light followed by transport of the visible signalwith hollow lightguides. The proposed approaches to light extraction and detection could enableradiation resistant diagnostics. ©2005 American Institute of Physics.fDOI: 10.1063/1.1852317g

I. INTRODUCTION

The diagnostic of “burning plasma” experiments inwhich deuterium-tritium fusion reactions are the main heat-ing source, will pose a major challenge. Although at presentthe choices for a next-step device in the US fusion programare still open, such an experiment is deemed essential for theprogress of fusion research.

Among the basic measurements,spectroscopic imagingof plasma emission profilesin the energy range from a fewelectron volts to tens of kilo-electron voltsskeVd svisible to xraysd can play an essential role for machine control and op-eration, as well as for plasma performance evaluation andphysics studies.1 As one will approach the burning plasmaenvironment, these measurements will, however, be consid-erably more difficult, due to intense plasma and nuclear ra-diation, as well as to long plasma exposure. As an example,the “first mirrors” directly viewing the International Thermo-nuclear ReactorsITERd plasma will be exposed for hundredsof hours each year to energetic neutron and gamma fluxes of<1012 cm−2 s−1, charge-exchangesCXd neutral fluxes of1013 cm−2 s−1, and heat fluxes of<1.5 kW/m2.1 In addition,the optical elements in proximity to the burning plasma canbe serviced or replaced only remotely, due to activation ofthe adjacent structural materials.

The present designs for burning plasma spectroscopicsystems extrapolate the large tokamak experience. Typicallyplasma light is extracted using a metallic first mirror andthen sent to a remote spectrometric system using secondary

mirrors arranged in a labyrinth path in the radiation shield.There are several difficulties with this approach. First, theprimary mirrors are subjected to intense sputtering and coat-ing by neutral atoms escaping the plasma.2 This causes mir-ror reflectivity loss and changes in the spectral and polarizingproperties, which can severely impact critical diagnostics,such as for instance active beam spectroscopy.3,4 Second,maintaining the accurate optical alignment over longsù10 m in ITERd multiply folded beam paths in the harshburning plasma environment is also difficult. Finally, usingthe remote spectrometry approach, wide-angle measurementsof the plasma profiles often require using multiple, expensivebeam-lines.1

It is therefore important to investigate also ideas andtools for burning plasma spectroscopic diagnostic. In particu-lar, spectroscopic imaging instruments that could function incloser proximity to the burning plasma could provide morecost effective and robust techniques for some of the criticalmeasurements, such as for instance impurity content in thedivertor and main plasma. In the present article we focus onelements for the extraction and detection of extreme ultravio-let sXUV d light fultrasoft x ray to vacuum ultravioletsVUV d,or the few angstromssÅd to <2000 Å rangeg in the burningplasma environment. Some of these elements can also be ofinterest also for visible light diagnostics, such as active beamspectroscopy. Many of the discussed devices are made pos-sible through recent advances in nanofabrication technology.

The structure of the article is as follows. Section II dis-cusses short wavelength light extraction using diffractive op-tical elements such as transmission gratings. Sec. III focuseson light extraction with multilayer mirrors, while Sec. IVdiscusses possible approaches for in-vessel detection ofXUV light from the burning plasma. Throughout the article

adElectronic mail: stutman@pppl.govbdPresent address: “Horia Hulubei” National Institute of Physics and Nuclear

Engineering, Department of Nuclear Physics, Bucharest, Romania.

REVIEW OF SCIENTIFIC INSTRUMENTS76, 023505s2005d

0034-6748/2005/76~2!/023505/9/$22.50 © 2005 American Institute of Physics76, 023505-1

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

we use ITER radiation numbers for exemplification pur-poses, since at present these provide the best description ofthe burning plasma environment.

II. XUV LIGHT EXTRACTION WITH DIFFRACTIVEOPTICAL ELEMENTS

A first idea we discuss is usingnormal incidence,transmissive-diffractive optical elementsas an alternative tothe grazing incidence mirrors conventionally used for shortwavelength light extraction. The basic principle is to usediffraction off a freestanding metallic structure such as atransmission grating or a Fresnel zone plate, in order to de-flect a usable portion of the incident light out of the directplasma view and into a particle and radiation shielded mea-surement region.

The advantage offered by this approach might be that theactive light-deflecting element is not an extended materialsurface, but an array of thin metallic wires. Thus even underheavy neutron, gamma, and particle bombardment, and evenwith some plasma impurity deposition, there is a betterchance that such a device will withstand the direct plasmaview without significant altering its optical properties. Fur-thermore, even if slow efficiency degradation would occur, itwould be easier to use interchangeable or movable gratingsin the beam path, since at normal incidence the sensitivity tomisalignment is low.

Transmission gratingssTGsd have been used for quitesome time as soft x-ray dispersive elements in the spectros-copy of laboratory and astrophysical plasmas.5,6 Freestand-ing metallic gratings with up to 10 000 lines/mm can nowa-days be produced by electron beam lithography in asubmicron thick substrate.5

Using such gratings we recently tested on the NationalSpherical Torus ExperimentsNSTXd and Current DriveExperiment-UpgradesCDX-Ud tokamaks simple and com-pact imaging spectrometers for the ultrasoft x-raysfew Å to<300 Åd range. These are in essence spectrally resolved pin-hole cameras, in which two narrow slits and a normal inci-dence grating disperse and image the light onto a two-dimensional detector. As illustrated by the CDX-U spectrumin Fig. 1, spectral resolutionDl /l<0.03–0.06 and spatialresolutionDr /a<0.05–0.10sa, device minor radiusd wereobtained in this simple setup.7

Similar instruments could be of interest for the spectros-copy of the burning plasma. In particular, these nonfocusingdevices would be well suited for survey spectrometry ofthe divertor region, where bright XUV line emissions1015–1017 photons cm2 sr−1 s−1d can be expected from thecool and dense plasma. Multichordal XUV spectroscopy willbe essential for the burning plasma divertor, due to the likelyuse of high-Z cooling gases or first wall components. Two-dimensional measurements will also be important in the di-vertor due to the complex plasma shape. Building such adiagnostic using multiple conventional spectrometers andbeam lines would be costly and difficult.8

The possible layout of a TG spectrometer for the burningplasma divertor is shown in Fig. 2. Assuming the ITER ge-ometry, adequate spatial coverage could be obtained by plac-

ing the instrument a few meters away from the divertor. Asecond instrument vertically viewing the divertor could beused for tomographic reconstruction of the local emissivity.As sketched for the second device, at VUV wavelengths thelight could be further folded or focused after being diffractedby the grating, using for instance broad band SiC mirrors.Finally, high throughput polychromators based on grid colli-mator, transmission gratings, and focusing mirrors can beenvisioned along the lines discussed in Sec. III.

To obtain spectral and spatial resolution as above, theplasma is viewed through narrow entrance and imaging slitsse.g., 120mm36 cm and 120mm30.6 cm, respectivelyd,spaced 10 cm apart, and positioned behind a thick radiationshield. In addition to collimating the incident light, the slitsserve also to drastically reduce the amount of sputtering anddeposition on the TG.

For the instrument viewing only the divertor plasma theenergy of the escaping neutrals will be too low to sputter ahigh-Z grating.2 For the device viewing the divertor throughthe main plasma some sputtering will occur. Considering agrating made of high-Z refractory metal such as W or Ta,assuming ITER first wall conditionss<231015 cm−2 s−1 en-ergetic CX atomsd, and using effective sputtering yields pre-dicted for ITER in Ref. 9, one obtains that the<5310−6 cm2 sr throughput of the collimator limits the gratingsputtering to several Å per ITER operation years<400“burn” hoursd.

FIG. 1. Space resolved spectrum from the CDX-U tokamak obtained with atangentially viewing TG spectrometer equipped with 5000 l /mm gratingand MCP charge coupled devicesCCDd camera detection. The CCD cameraintegration time was 20 msssee Ref. 7d.

023505-2 Stutman et al. Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

More significant in the divertor could be deposition ef-fects. Recent ITER calculations predict that the net C depo-sition rate would range from<1014 atoms cm−2 s−1 near themain plasma, to<231015 atoms cm−2 s−1 near the outer di-vertor leg.10 Assuming these rates, the C coating on the grat-ing viewing the divertor would reach<600 Å per ITER op-eration year. It is likely that such a thin low-Z overcoat on ahigh-Z grating will not significantly affect its diffractiveproperties. In addition, since as above mentioned the sensi-tivity to misalignment at normal incidence is low, one could“refresh” the <120 m wide grating area exposed to theplasma by laterally translating the grating.

The above devices would require transmission gratingsoperating also in the VUV range. While such gratings are notcommonly used for VUV spectroscopy, where less costlyreflection gratings are preferred, VUV instruments based onlarger period transmission gratingss500–2500 l /mmd havenevertheless been demonstrated.11,12

Todays advances in nanofabrication would make pos-sible optimizing the grating performance through shaping ofthe grating bars and choice of materials. For instance, thespectral resolution of the TG imaging spectrometer is in thefirst order5

Dl/l < 2sw/Ldsd/ld, s1d

wherew andL are the collimator width and length, andd isthe grating period. The device in Fig. 2 would have for in-stanceDl<12 Å at 600 Å, when using ad=5000 Å grating.This would be adequate for VUV survey spectrometry of thedivertor plasma, where the spectrum is typically dominatedby strong resonance transitions from low charge states oflow-Z impurities, spaced some tens of Å apartse.g., C III2s2-2s2p at <977 Å and O VI 2s-2p at <1032 Åd. It might,however, be possible to increase the resolution of the TGinstrument by deeply shaping the grating bars and preferen-tially channeling the radiation in high diffraction orders. Thiseffect has been observed since the early days of TG researchand was attributed to internal reflection on the grating bars.13

Neglecting phase effects the diffraction efficiency of atransmission grating is14

Im

I0 sld < F3sinSmpa

dD

mp4

2

s2d

with m=diffraction order,d=grating period,a=gap betweengrating wires, andF=fraction of grating open areasoutsidethe support structured. With open area fraction up to<65%the TG efficiency can thus reach several percent in the firstorder.6 Assuming for instance a ten chord TG spectrometerhaving 120mm wide by 0.6 cm high detection “pixels” lo-cated at<10 cm behind the grating, one obtains for the di-vertor range of XUV brightness an incident power of<0.2–20 nW per pixel, at a wavelength of 400 Å for in-stance. Detection possibilities for these signal levels are dis-cussed in Sec. IV. A TG/focusing mirror polychromatorwould achieve much stronger signals per detection channel.

Other diffractive optical elements of interest could beFresnel zone plates. In addition to spectrally dispersing theincoming light, these devices also focus radiation frommulti-keV to visible light.15,16 For instance, arrays of “pho-ton sieve” metallic lenses are currently being investigated forvisible light imaging in harsh environments.16 Such devicescould conceivably be used also as “first lenses” for lightextraction and focusing in the burning plasma environment.Similarly, micropatterned Fresnel mirrors which change lightdirection by diffraction rather than reflection could be of in-terest as replacement for conventional mirrors.17

In conclusion, freestanding diffractive optical elementsin collimated optical designs might withstand exposure to theburning plasma with possibly less degradation of their lightcollecting and polarizing properties than reflective “firstmirrors.”

III. USXR LIGHT EXTRACTION WITH MULTILAYERMIRORS

Other potentially important tools for XUV light extrac-tion from the burning plasma could be synthetic multilayermirrors sMLM d. These are high throughput devices that de-flect and disperse USXR light through Bragg diffraction.18 Inrecent years the multilayer mirrors found more and moreapplications as USXR light extractors in tokamaks. The in-

FIG. 2. Conceptual layout of imaging TG spectrometers for the burningplasma divertor. The XUV light imaged and dispersed by the pinhole gratingcamera is detected by a thin phosphor, amplified and transported out of thevessel using hollow fiber waveguides, as discussed in Sec. IV. Further de-flection or focusing would be possible for VUV light using, e.g., broadbandSiC mirrors.

023505-3 Burning plasma spectroscopic diagnostic Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

struments using them range from single chord polychroma-tors, to multichordal arrays of monochromators, and to two-dimensional 2D imaging devices.19–24

A class of multilayer based devices that could havestrong potential for burning plasma diagnostics are the mir-ror monochromator arrays.20–22 These high throughput de-vices can image with high spectralsù1.5 Åd, spatialsù2 cmd, and temporalsù2 msd resolution the USXR lineemission from intrinsic or injected impurities, enabling mea-surements of the impurity density, transport and even mag-netohydrodynamicsMHDd activity. As an illustration, Fig. 3shows the emission of the C VILya shell in a beam heatedNSTX plasma, measured by a 16 chord mirror arrayequipped with photodiode detection.20 In addition, to a con-tinuous measurement of the C VI profile of interest for trans-port assessment, the high throughput of the system enableobtaining the mode structure of MHD perturbations localizedin the C VI shell. Similar measurements were performed onthe CDX-U tokamak using a MLM array for the Li IIILya

emission at 135 Å.21

A mirror polychromator using grid collimators for in-creased throughput and resolution and microchannel platesfor increased sensitivity was also prototyped on LHD for

transport measurements using the faint beam charge-exchange CX emission from tracer embeddedsTESPELdpellets.22 The layout of the device together with illustrativetraces of injected MgHa emissionsl<45.5 Åd is shown inFig. 4. As seen, good signal-to-noise and background rejec-tion can be obtained with this design even for faint USXRtransitions.

In-vessel arrays of grid collimator/mirror monochroma-tors resembling that in Fig. 4 might enable measuring thelow and high-Z impurity line emissionspossibly includingthe helium ashd in the edge and core of the burning plasma.Both electron collision excited and beam CX excited transi-tions are of interest. For instance, theLya or Ha USXR linesof low-Z hydrogen-like ions have orders of magnitude higherCX excitation rates than the high-n visible lines typicallyused in active charge exchange spectroscopy.22,25 This couldbe important in burning plasma conditions, where due tolimited beam penetration and intense bremstrahlung emis-sion the signal-to background ratio of visible light CX tran-sitions is very poorse.g.,ù1/100 for the C VIn=7–8 tran-sition in the ITER core.25d

A possible layout for the “unit-cell” of a mirror mono-chromator array for the burning plasma is shown in Fig. 5.

FIG. 3. sad Layout of a MLM/diode monochromator array prototyped onNSTX ssee Ref. 19d. Sixteen Ti/Cr mirror monochromators having AXUVdiodes as detectors measure the C VILya emission at<33.7 Å in NSTX.sbdThe plot of theLya intensity on transport time scales shows the formation ofa narrow emission shell, indicative of the low particle diffusivity in beamheated NSTX discharges. The same data plotted on MHD time scales evi-dences a 2/1 perturbation localized in the C VI shell.

FIG. 4. sad Layout of grid-collimator MLM monochromator tested for tracerembedded pelletsTESPELd measurements at LHDssee Ref. 21d. The pelletcreates a dilutes<0.1%d shell of Mg impurity, which is measured throughits beam excitedHa emission at<45.5 Å. sbd Measured and computedtraces of Mg emission atl<45.5 Å. The high throughput and narrow spec-tral bandpasssfew Åd of the mirror monochromator enables transport rel-evant measurements, even when using faint CX excited transitionssLHDbeam energy<190 keV/amud.

023505-4 Stutman et al. Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

The mirror is protected from direct plasma effects and fromthe background light by the combination of a grid collimatorand a thin metallic filter. The use of grid collimators in XUVinstrumentation is discussed in Ref. 26. These enable obtain-ing in a compact layout spatial resolution of a few centime-ters at several meters distance, together with a few percentspectral resolution and high optical throughput.22 A differ-ence from the device in Fig. 4 is that a focusing rather thanplanar multilayer would be used in the burning plasma in-strument, in order to increase the ratio between the usefulphoton flux and the nuclear background measured by thedetectorssee Sec. IVd.

The effects of sputtering and deposition on the gridcollimator/MLM instrument can be estimated as above. Con-sidering the geometry in Fig. 4se.g., 0.130.1310 cm col-limator channels having<10−6 cm2 sr throughput eachd andassuming as above ITER first wall conditions, the sputteringrate for a common XUV filter material such as Be will bearound 15 Å per operation year. Since the typical filterthickness is a few kilo-angstroms, the effect is negligible.Similarly, assuming the main plasma deposition rates<1014 cm−2 s−1d the carbon coating will also be minimals<5 Å/yeard.

The overall throughput of a grid collimator is neverthe-less high, since the combined area of the collimating chan-nels can reach<70% of the cross section.26 For instance,assuming a 333 cm cross section the device in Fig. 4 canhave <5310−4 cm2 sr geometrical throughput. Further on,high performance multilayers typically haveù20% reflectiv-ity, while the long wavelength blocking filters can be opti-mized for<20% transmission.20–22 The brightness of low-nUSXR transitions from the main plasma may be expected tospan the<1013–1015 photons cm−2 sr−1 s−1 range swith thelower end representing beam excited transitionsd. Assumingthen that the spherical mirror would focus the diffracted pho-

ton flux on a 333 mm detector, the device in Fig. 3 wouldproduce<10–1000 nW signals, at 30 Å wavelength for in-stance.

In addition to resisting exposure to the burning plasma,the in-vessel mirrors need to resist also the intense neutronand gamma irradiation. One can presume for instance thatthe mirrors in the above monochromator array will be ex-posed to neutron fluences comparable to those at the ITERfirst mirror locationsø1019 cm−2d.1 Our earlier assessmentof the radiation hardness of the multilayers shows that evenat such high fluence the reflectivity is maintained for sometypes of mirrors. For instance, W/B4C mirrors exhibitedonly a few percent decrease in their peak reflectivity afterbeing irradiated with <1.131019 cm−2 neutrons of1–2 MeV characteristic energy.27

While maintaining good reflectivity, a small shifts<2%d in the wavelength of the Bragg peak was neverthe-less observed for the irradiated W/B4C mirrors.27 Similarconclusions have been obtained in other irradiation tests.28

Based on such observations the use of multilayers is pres-ently considered only for remote instrumentation onITER.1,28

As noted earlier, however, the above irradiation experi-ments have been performed at high temperature, due tonuclear heating of the mirror substrate and insufficient cool-ing. This left open the question if the observed Bragg peakshift was a thermal or a neutron damage effect. Our estimateswere that the thermal effects are dominant.27 Irradiation testsof cooled, or preannealed mirrors are therefore needed for aconclusive resolution of this issue.

Even if a small shift of the Bragg peak would occur intime, however, we observe that optical schemes that cancompensate for such shifts are in any case possible. A simpleexample is illustrated in Fig. 6, where there the parallel beamfrom the grid collimator is incident on a mirror having cur-vature radius of approximately ten times the beam diameter.As shown by the USXR ray tracing calculations, even if themirror d spacing varies by a few percent, a nearly constantfraction of the incident photons is focused on the detector,since the region of peak reflectivity also “shifts” on the mir-

FIG. 5. Conceptual layout of grid collimator/spherical multilayer monochro-mator for the burning plasma. High performance multilayers can nowadaysbe manufactured with very tight curvature radii, or with precisely gradeddspacing.

FIG. 6. Example of optical scheme compensating for small changes in theBragg diffraction angle at a given wavelength. As shown by the ray tracingcalculations with theSHADOW code on the right, a practically constant frac-tion of the incident beam is focused on the detector as the mirrord spacingincreases by a few percent.

023505-5 Burning plasma spectroscopic diagnostic Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

ror surface. Using mirrors with laterally gradedd spacing isanother option, which would reduce the curvature require-ments.

Transient thermal effects on the mirrord spacing aremore difficult to quantify. It is likely, however, that multilay-ers functioning in closer proximity to the burning plasmawould require temperature control in order to maintain a con-stantd spacing and reflectivity.

In conclusion, in conjunction with protective collimatorsand filters the multilayer mirrors might also offer useful pos-sibilities for in-vessel light extraction from the burningplasma.

IV. IN-VESSEL XUV LIGHT DETECTION, SIGNALAMPLIFICATION, AND TRANSPORT

A. XUV light detection

The above instruments use diffractive elements to extractXUV light in closer proximity sfew metersd to the burningplasma. The extracting element views the plasma through anarrow collimator in the primary radiation shieldse.g., thetritium breeding blanket in ITERd, while deflecting the usefulradiation towards a protected detector. Even behind thisshield the nuclear background will be, however, quite high,requiring approaches to in-vessel light detection.

A detector considered for in-vessel energy integratedx-ray measurements in ITER is the vacuum photodiode.29

This is in essence a high-Z metal surface that converts theincoming x-ray flux into photoelectrons, which after escap-ing into the vacuum are collected at an anode and detectedby a remote current preamplifier. While this approach mightbe of interest for theùmW signals expected from directx-ray detection, even here the transport of small electricalcurrents near the burning plasma remains a serious issue. Thedifficulties range from radiation induced electromagneticforces in the electrical circuits,1 to electromagnetic pickup,and to the large capacitance presented by a long cable to thecurrent preamplifier. It is therefore likely that for theùnWsignals estimated for our XUV devices the electronic detec-tion approach is not useful.

One concept we study and which may enable more ro-bust in-vessel XUV detectors is the “optical array” discussedin Ref. 30. In its burning plasma version it would consists ofan efficient and radiation resistant XUV to visible light con-verter, followed by radiation hard visible light wave guideswhich transport the light signal outside the vacuum. An in-termediate visible light amplification stage could be used toboost the signal before transport with the wave guides. Thisconcept is illustrated also in Fig. 2, where a phosphor basedoptical array relays out of the vacuum the space-resolvedspectra recorded by the TG instrument. The use of opticalrather than electrical signal path in the vacuum vessel couldhave a better chance of withstanding the harsh burningplasma environment.

The increased immunity of the optical array design tonoise and nuclear radiation in the tokamak environment isillustrated in Fig. 7, which compares MHD fluctuations mea-sured on NSTX with an USXR optical array having CsIsTldconverter, to the same measurement performed using a con-

ventional photodiode array.30 Both the noise estimated toarise from neutron interactions and the electromagneticpickup are much reduced in the optical device.

The first element of the in-vessel detector is a thin XUVphosphor. Phosphor converters have also been proposed forenergy integrated x-ray imaging in ITER.31 Our investiga-tions of XUV inorganic phosphors have identified some goodcandidates as concerns their efficiency.32,33 For instance, a2 mg/cm2 s<6 mm thickd P45sY2O2S:Tbd layer was foundto emit about 60 visible photons in 4p for a 525 eV incidentphoton and about 3.6 visible photons for a 75 eV photon.32

The active research towards the development of directionallyemitting, multilayered transparent phosphor films might leadto further efficiency improvement.34

Detailed radiation transport calculations are needed toassess the signal to nuclear background ratio for each par-ticular device and geometry. For an order of magnitude esti-mate one can nevertheless compare the XUV power incidenton a detector pixel, to the ITER absorbed dose predictions.For instance, for the TG spectrometer a few meters awayfrom the divertor, one can assume the radiation field at thelocation of the ITER secondary mirrorss<33109 cm−2 s−1

energetic neutron and gamma flux, producing<10−2 Gy/sabsorbed dose rate in silicon.1d Similarly, for the MLM ar-rays operating at a shorter distancese.g., <2 md from theplasma, one may assume as nuclear background the geomet-ric average between the first and secondary ITER mirrordose ratess<5310−2 Gy/sd.

Using these values and scaling the radiation interactioncoefficients from Si to the P45 composition, one obtains anabsorbed power of<0.2 nW for the 120mm36 mm pixelof the TG device and<12 nW for the 333 mm MLMmonochromator detector, respectively. This indicates thatboth instruments could measure the lower end of their re-spective brightness rangess1015 photons cm−2 sr−1 s−1 for the

FIG. 7. USXR signals from an optical array equipped with CsIsTld con-verter and from a re-entrant photodiode array during high powers7 MWdbeam injection on NSTXssee Ref. 30d. While both traces show the 1/1MHD perturbation, the optical array is practically free of the neutron spikesseen on the diode array signal. Electromagnetic pickup and electronic noiseare also significantly reduced due to the use of an optical rather than elec-trical signal path.

023505-6 Stutman et al. Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

divertor and 1013 cm−2 sr−1 s−1 for the main plasmad witharound unity signal-to-background ratio.

The phosphor would be deposited on a thin transparentsubstrate, such as for instance a fiber optic plate. Nuclearradiation induced transmission loss and luminescence is anissue for thick windows.1 However, ITER research has leadto the development of glass having improved radiation resis-tance, such as hydrogen hardened KU-1.35,36 Assuming theuse of such glass, the visible light attenuation in a1-mm-thick plate would be negligiblesø a few percent after531017 cm−2, E.0.1 MeV neutronsd.36

A more significant effect could have secondary electronsgenerated in the glass substrate by the energetic gammaflux. Assuming again ITER conditions, the flux of<33109 cm−2 s−1 MeV gamma rays at the secondary mirrorposition would translate into<105 s−1 Compton electrons offew hundred keV average energy impacting the 120mm36 mm pixel of the TG instrument. Using the stoppingpower approximation, the energy deposited in the<2 mg/cm2 phosphor layer would then be<0.1 nW, whichis also comparable to the lower range of XUV signals ex-pected from the divertor. A similar conclusion obtains for thedetector of the MLM device.

Finally, since during the discharge the phosphor wouldbe immersed in a low pressureD-T mixture, it is of interestto evaluate the potential effect of tritium beta decays. Thebeta emission rate at e.g., 10−5 Torr tritium pressure is<23103 cm−3 s−1, resulting in a few hundred beta electronsincident on the TG pixel froma<10310310 cm3 spec-trometer volume. With<5.7 keV mean beta energy, the ab-sorbed power in the TG phosphor pixel is negligiblespico-Watt ranged. While tritium retention in the phosphor mightincrease this value, results from spectrometers equipped withphosphor converters and exposed to tritium suggest that theeffect is not large.37

In conclusion, few micron thick phosphors would be“transparent” to energetic neutrons, gamma and secondaryelectrons, enabling in principle the detection of quite smallXUV signals inside the burning plasma vessel. The effects ofgamma and neutron irradiation on efficient XUV convertersneed, however, to be investigated. The research so far fo-cused on x ray and gamma scintillators.38 While irradiationtests of rare-earth phosphors with mega-electron-unit protonshave identified some radiation hard candidates, it is not clearwhether these results can be extrapolated to energetic neu-tron and gamma irradiationssee, e.g., Ref. 31, and referencesthereind.

B. Visible signal amplification

Assuming an efficient converter such as P45, the abovelevels of XUV power incident on the phosphor would trans-late into a large number of converted visible photonssfew108–1010 s−1 per pixel, or detection channeld. At the lowerend of this range it would be desirable to boost the lightintensity before transport to the outside world. One type ofdevice that could accomplish this task is the sealed proximityfocused image intensifier. Due to the combined gain of themultichannel platesMCPd and proximity focused phosphor

these devices can achieve light amplification up to<104 W/W.7

In neutron and gamma fluxesø several 108 n/cm2 s−1

the noise from conventional MCP intensifiers is low.7 Instronger fluxes the noise becomes a problem, as indicated byspectrometers operated during high powerD-T experimentsat TFTR and JETsneutron and gamma fluxes up to a few1012 cm−2 s−1d.37,39,40 The main source of intensifier noisewas estimated to be gamma interactions in the MCP.40

This might not be surprising, as conventional MCPs aremanufactured in lead doped glass having around 50% leadcontent by mass.41 Since in addition the glass volume signifi-cantly exceeds that of the amplifying channels, this makesthe MCP also quite sensitive to hard x rays and gamma raysse.g.,<2% detection efficiency at 500 keV41d.

Advances in nanotechnology lead, however, to the de-velopment of types of MCPs, likely much more “transpar-ent” to nuclear radiation than the conventional ones. Theseare silicon-MCPs, or quartz-MCPs, manufactured by directlymicromachining the MCP pores in thinstens of micronsdsilicon wafers, which can eventually be oxidized toquartz.42,43Although at a developmental stage, devices hav-ing ultrathins<1mmd walls, open volume in excess of 90%,and high gain have been demonstrated. At the same time, theSi–MCP is estimated to be more radiation resistant and in-sensitive to strong magnetic fields.42,43

For the stronger signals generated by focusing instru-ments such as the MLM monochromator, proximity focusedintensifiers consisting just of a photocathode and an electronphosphor would provide simple and radiation hard light am-plifiers. Finally, a type of light amplifier of potential interestfor the burning plasma could be the gaseous imageintensifier.44 This consists of an image intensifier in whichthe amplifying element is a gaseous electron multipliersGEMd.45 While the Kapton foil of the conventional GEM isnot radiation resistant and typically requires flowing theworking gas, sealed GEM amplifiers based on silica capillaryplates are currently being investigated.46

In conclusion, the recent advances in nanofabricationmight have also opened possibilities for radiation resistantlight amplifiers, which could be used to boost the visiblelight signal by<102–104 W/W.

C. Visible signal transport

The main obstacle towards the development of in-vesseloptical diagnostics for the burning plasma might have beenthat the extraction of visible light signals using conventional,solid core optical fibers is not possible. As shown by manyirradiation tests as well as by tokamakD-T operation, thetransmission of solid core fibers significantly degrades at theradiation levels expected near the burning plasma.1,28,40Al-though as mentioned, the recently developed hydrogen hard-ened fibers have improved radiation resistance, their use inITER is presently restricted to outside the vacuum vessel.1,28

Finally, even if the transmission problem could be mitigated,a major obstacle would remain—the radiation induced lumi-nescence in the fiber core, which over long path lengths cangive raise to large background levels.1,36,40

023505-7 Burning plasma spectroscopic diagnostic Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

It was earlier suggested that hollow optical fibers mightalleviate these problems.40 The physical concept and technol-ogy for the production of efficient hollow fibers was notavailable at that time, however. Recent advances in nano-technology have changed this situation as well. Thus, pres-ently a revolution is underway in the fiber optic technology,with the advent of “photonic-crystal” hollow fibers.47,48

These carry light based on a completely different physicalprinciple than solid core fibers. Instead of total internal re-flection on an outer cladding and transmission of the re-flected light through a solid core, the wave is guided byconstructive interference on a layer of subwavelength holessurrounding a larger hollow core, as illustrated in Fig. 8.Another type of hollow fiber based on diffraction on coaxialmultilayers is the recently developed OmniGuide® Braggfiber.49

We advance that these types of fibers could offer a solu-tion to the problem of visible light transport in the burningplasma environment. The reason being that in these fibersmore than 95% of the light energy can be guided through thehollow core, instead of solid silica. Thus, even if the inter-stitial glass will darken due to irradiation, only the negligiblelight attenuation along the microscopic distance between theguiding holes might have an effect on the overall transmis-sion. In addition, very little of the fluorescence light gener-ated in the thin silica walls should be guided through thecore. In conclusion, one cana priori expect that the“photonic-crystal” fibers will perform well up to high levelsof irradiation. Finally, like all interferential devices, the hol-low fibers have a defined wavelength bandpass, which couldbe useful for the rejection of any parasitic light.

Presently the performance of these fibers at visiblewavelengths is relatively modestsnumerical aperturesaround 0.12, core diameters ofø15 mm and attenuation ofthe order of 0.8 dB/md, since they have been primarily op-timized for the near infraredsNIRd wavelengths of interest intelecommunications. It is, however, in principle possible toimprove these parameters and numerical aperturesù0.3 anddiameters of a few hundred microns are considered techni-

cally feasible.50 Conversely, XUV phosphors emitting in theNIR could be explored, in order to match the highly efficientNIR hollow fibers.

To conclude, combining efficient XUV converters withradiation hard light amplifiers and hollow waveguides itmight be possible to develop in-vessel detectors for burningplasma experiments. For instance, assuming 1% XUV con-version efficiency, 103 W/W light amplification, and<1%transmission for the hollow fiber bundles the number of vis-ible photons counted outside the vacuum for the TG devicewould be in the 107–109 s−1 range.

ACKNOWLEDGMENT

The present work is supported by U.S. Department ofEnergy Grant No. DE-FG02-99ER54523.

1A. E. Costley, D. J. Campbell, S. Kasai, K. E. Young, and V. Zaveriaev,Fusion Eng. Des.55, 331 s2001d.

2V. Voitsenyaet al., Rev. Sci. Instrum.72, 475 s2001d.3K. Yu. Vukolov, M. I. Guseva, S. A. Evstigneev, L. S. Danelyan, V. M.Gureev, A. A. Medvedev, and S. N. Zvonkov, 30th EPS Conference onContr. Fusion and Plasma Phys., St. Petersburg, 7–11 July 2003, ECA Vol.27A, p. 2.76

4A. Malaquias, M. von Hellermann, P. Lotte, S. Tugarinov, and V. S. Voit-senya, 30th EPS Conference on Contr. Fusion and Plasma Phys., St. Pe-tersburg, 7–11 July 2003, ECA Vol. 27A, p. O-3.4C.

5T. Wilhein, S. Rehbein, D. Hambach, M. Berglund, L. Rymell, and H. M.Hertz, Rev. Sci. Instrum.70, 1694s1999d.

6C. Brinkmanet al., Proc. SPIE4012, 81 s2000d.7B. Blagojevic, D. Stutman, M. Finkenthal, H. W. Moos, R. Kaita, and R.Majeski, Rev. Sci. Instrum.74, 1988s2003d.

8N. C. Hawkeset al., in Diagnostic For Experimental Fusion Reactors 2,edited by P. E. Stottet al. sPlenum, New York, 1988d, p. 297.

9M. Mayer, R. Behrisch, C. Gowers, P. Andrew, and A. T. Peacock, inDiagnostic For Experimental Fusion Reactors 2, edited by P. E. Stottetal., sPlenum, New York, 1988d, p. 279.

10A. S. Kukushkin, H. D. Pacher, D. Coster, G. W. Pacher, and D. Reiter,30th EPS Conference on Contr. Fusion and Plasma Phys., St. Petersburg,7–11 July 2003, ECA Vol. 27A, p. 3.195.

11P. J. Caldwell, E. T. Arakawa, and T. A. Callcott, Appl. Opt.20, 3047s1981d.

12S. A. Flodstrom and R. Z. Bachrach, Rev. Sci. Instrum.47, 1464s1976d.13N. M. Ceglio, A. M. Hawryluk, D. G. Stearns, M. Kühne, and P. Müller,

Opt. Lett. 13, 267 s1988d.14G. R. Morrison, inX-Ray Science and Technology, edited by G. Michette

and C. J. BuckleysIOP, Bristol, 1993d, p. 319.15L. Kipp, M. Skibowski, R. L. Johnson, R. Berndt, R. Adelung, S. Harm,

and R. Seemann, NaturesLondond 414, 184 s2001d.16K. Kincade, Laser Focus World40, 34 s2004d.17Jon M. Bendickson, Elias N. Glytsis, and Thomas K. Gaylord, J. Opt. Soc.

Am. A 16, 113 s1999d.18B. L. Henke, E. M. Gullikson, J. Kerner, A. L. Oren, and R. L. Blake, J.

X-Ray Sci. Technol.2, 17 s1990d.19S. P. Regan, K. B. Fournier, M. J. May, V. Soukhanovskii, M. Finkenthal,

and H. W. Moos, Rev. Sci. Instrum.68, 1002s1997d.20D. Stutman, M. Finkenthal, V. Soukhanovskii, M. J. May, H. W. Moos,

and R. Kaita, Rev. Sci. Instrum.70, 572 s1999d.21V. A. Soukhanovskiiet al., Rev. Sci. Instrum.72, 737 s2001d.22D. Stutmanet al., Rev. Sci. Instrum.76, 013508s2005d.23L. A. Shmaenoket al., Rev. Sci. Instrum.72, 1411s2001d.24D. Stutman, M. Finkenthal, M. Iovea, V. Soukhanovskii, M. J. May, and

H. W. Moos, Rev. Sci. Instrum.72, 732 s2001d.25S. Tugarinovet al., Rev. Sci. Instrum.74, 2075s2003d.26V. A. Soukhanovskii, D. Stutman, M. Finkenthal, H. W. Moos, R. Kaita,

and R. Majeski, Rev. Sci. Instrum.72, 3270s2001d.27S. P. Reganet al., Rev. Sci. Instrum.68, 757 s1997d.28S. Yamamotoet al., J. Nucl. Mater.283, 60 s2000d.29F. H. Seguin, R. D. Petrasso, and C.-K. Li, Rev. Sci. Instrum.68, 753

s1997d.30L. F. Delgado-Aparicio, D. Stutman, K. Tritz, M. Finkenthal, R. Kaita, L.

FIG. 8. Cross-sectional structure of hollow core photonic crystal opticalfiber fcourtesy of R. Kristiansen, Crystal-Fibre. Inc.ssee Ref. 50dg. Up to<98% of the light energy is guided though the hollow core, due to a pho-tonic “band gap” effectssee Refs. 47 and 48d.

023505-8 Stutman et al. Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions

Roquemore, D. Johnson, and R. Majeski, Rev. Sci. Instrum.75, 4020s2004d.

31B. Zurro, C. Burgos, A. Ibarra, and K. J. McCarthy, Fusion Eng. Des.34/35, 353 s1997d.

32S. P. Regan, L. K. Huang, M. J. May, H. W. Moos, D. Stutman, S.Kovnovich, and M. Finkenthal, Appl. Opt.33, 3595s1994d.

33V. A. Soukhanovskii, S. P. Regan, M. J. May, M. Finkenthal, and H. W.Moos, Rev. Sci. Instrum.74, 4331s2003d.

34See, e.g., www.esrf.fr/Conferences/UsersMeeting2003/Detectors/Speakers/rogerIdurst.pdf.

35K. Yu. Vukolov and B. A. Levin, Fusion Eng. Des.66–68, 861 s2003d.36B. Brichardet al., J. Nucl. Mater.329–333, 1456s2004d.37I. H. Coffey, R. Barnsley, and JET EFDA Contributors, Rev. Sci. Instrum.

75, 3737s2004d.38A. S. Tremsin, J. F. Pearson, A. P. Nichols, A. Owens, A. N. Brunton, and

G. W. Fraser, Nucl. Instrum. Methods Phys. Res. A459, 543 s2001d.39K. W. Hill et al., Rev. Sci. Instrum.66, 913 s1995d.40A. Ramsey, Rev. Sci. Instrum.66, 873 s1995d.41J. L. Wiza, Nucl. Instrum. Methods162, 587 s1979d.

42A. Tremsin, J. V. Vallerga, O. H. W. Siegmund, C. P. Beetz, and R. W.Boerstler, Proc. SPIE4854, 215 s2002d.

43C. P. Beetz, R. Boerstler, J. Steinbeck, B. Lemieux, and D. R. Winn, Nucl.Instrum. Methods Phys. Res. A442, 443 s2000d.

44D. Mörmann, M. Balcerzyk, A. Breskin, R. Chechik, B. K. Singh, and A.Buzulutskov, Nucl. Instrum. Methods Phys. Res. A504, 93 s2003d.

45F. Sauli, Nucl. Instrum. Methods Phys. Res. A386, 531 s1997d.46C. Iacobaeus, A. Breskin, M. Danielsson, T. Francke, D. Mörmann, J.

Ostling, and V. Peskov, Nucl. Instrum. Methods Phys. Res. A525, 42s2004d.

47C. M. Smith, N. Venkataraman, M. T. Gallagher, D. Müller, J. A. West, N.F. Borrelli, D. C. Allen, and K. W. Koch, NaturesLondond 424, 657s2003d.

48J. D. Joannopoulos, P. R. Villenueve, and S. Fan, NaturesLondond 386,143 s1997d.

49M. Ibanescu, Y. Fink, S. Fan, E. L. Thomas, and J. D. Joannopoulos,Science289, 415 s2000d.

50R. Kristiansen, Crystal-Fibre Inc.sprivate communicationd.

023505-9 Burning plasma spectroscopic diagnostic Rev. Sci. Instrum. 76, 023505 ~2005!

Downloaded 12 Apr 2013 to 147.188.128.74. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://rsi.aip.org/about/rights_and_permissions