11
Beryllium—a better tokamak plasmafacing material? K. L. Wilson, R. A. Causey, W. L. Hsu, B. E. Mills, M. F. Smith, and J. B. Whitley Citation: Journal of Vacuum Science & Technology A 8, 1750 (1990); doi: 10.1116/1.576843 View online: http://dx.doi.org/10.1116/1.576843 View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/8/3?ver=pdfcov Published by the AVS: Science & Technology of Materials, Interfaces, and Processing Articles you may be interested in Twobody similarity and its violation in tokamak edge plasmas Phys. Plasmas 3, 3191 (1996); 10.1063/1.871624 Detached scrapeoff layer tokamak plasmas Phys. Plasmas 2, 1982 (1995); 10.1063/1.871284 Effect of a nonuniform resistive wall on the stability of tokamak plasmas Phys. Plasmas 1, 2931 (1994); 10.1063/1.870533 Electrontemperaturegradientdriven instability in tokamak boundary plasma Phys. Fluids B 5, 2206 (1993); 10.1063/1.860968 Fluidlike modeling of the plasma–material interaction Phys. Fluids B 5, 1647 (1993); 10.1063/1.860797 Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Beryllium—a better tokamak plasma-facing material?

  • Upload
    k-l

  • View
    214

  • Download
    1

Embed Size (px)

Citation preview

Page 1: Beryllium—a better tokamak plasma-facing material?

Beryllium—a better tokamak plasmafacing material?K. L. Wilson, R. A. Causey, W. L. Hsu, B. E. Mills, M. F. Smith, and J. B. Whitley Citation: Journal of Vacuum Science & Technology A 8, 1750 (1990); doi: 10.1116/1.576843 View online: http://dx.doi.org/10.1116/1.576843 View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/8/3?ver=pdfcov Published by the AVS: Science & Technology of Materials, Interfaces, and Processing Articles you may be interested in Twobody similarity and its violation in tokamak edge plasmas Phys. Plasmas 3, 3191 (1996); 10.1063/1.871624 Detached scrapeoff layer tokamak plasmas Phys. Plasmas 2, 1982 (1995); 10.1063/1.871284 Effect of a nonuniform resistive wall on the stability of tokamak plasmas Phys. Plasmas 1, 2931 (1994); 10.1063/1.870533 Electrontemperaturegradientdriven instability in tokamak boundary plasma Phys. Fluids B 5, 2206 (1993); 10.1063/1.860968 Fluidlike modeling of the plasma–material interaction Phys. Fluids B 5, 1647 (1993); 10.1063/1.860797

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 2: Beryllium—a better tokamak plasma-facing material?

Beryllium-a better tokamak plasma-facing material? K. L. Wilson, R. A. Causey, W. L. Hsu, B. E. Mills, M. F. Smith, and J. B. Whitley Sandia National Laboratorie~; Livermore, California 94551 and Albuquerque, New Mexico 87185

(Received 20 December 1989; accepted 21 January 1990)

The plasma-material interaction and high heat flux properties of beryllium are reviewed to determine its suitability as a plasma-facing component in magnetic fusion energy reactors. Consideration is given to beryllium's outgassing, erosion, and hydrogen retention characteristics. Its r~sponses to normal and off-normal high heat fluxes are compared to graphite in both the as­rece1ved and the neutron-irradiated states, Beryllium's performance in present-day devices is assessed, and its expected behavior in future reactors is summarized. It is concluded that beryllium is potentially a better plasma-facing material than graphite and that more development and testing is warranted.

I. INTRODUCTION

A plasma-interactive component in a magnetic fusion ener­gy reactor must help to extract heat from the plasma, assist in fuel particle recycling, and survive plasma-material inter­action and high heat flux exposure for an economically rea­sonable lifetime. A plasma-interactive component must not contaminate the plasma with eroded wall atoms, nor achieve an unacceptable in-vessel tritium inventory. In many ways, beryllium is the ideal plasma-facing material. The thermal properties of beryllium are excellent for steady-state heat removal. Its mechanical properties are an improvement over graphite. It has a low atomic number, which reduces the impact of eroded first wall atoms on plasma contamination. Beryllium also possesses a low tritium affinity, which could lead to a reduced tritium inventory compared to a graphite­based device. However, beryllium's excellent thermal prop­erties must be balanced against its poor refractory proper­ties, such as a low melting temperature and high vapor pressure. This paper compares beryllium to graphite as a plasma-facing materiaL Consideration is given to plasma­material interaction properties, high heat flux characteris­tics, neutron irradiation response, and present and projected fusion device performance.

II. PHYSICAL AND MECHANICAL PROPERTIES

The physical properties of beryllium are summarized in Table I, which is taken in part from several excellent reviews ~fthe subject. 1,2 In addition to its low atomic number, beryl­hum has several excellent thermal properties that make it well-suited for plasma-facing material applications. The temperature dependence of the thermal conductivity of be­ryllium is compared to Great Lakes H 451 and POCO AXF-5Q graphite in Fig. 1. It can be seen that beryllium's thermal conductivity is comparable to that of these typical nuclear­grade graphites at elevated temperature, and exceeds their capability below 500 K. Additionally, as discussed in Sec. V, the thermal conductivity of graphite is severely degraded by neutron irradiation, while that of beryllium is not affected. The specific heat ofbery11ium exceeds that of graphite by a ~actor 0: 2 over the temperature range of interest for plasma ~nteractlve components. However, beryllium is significantly mferior to graphite in its vapor pressure and melting point.

While graphite does not exhibit significant thermal sublima­tion until temperatures in excess of 2000 K, the vapor pres­sure of beryllium at 1000 K exceeds 10-- 6 Pa. Beryllium's high heat capacity and thermal conductivity can be used to maintain low surface temperatures in plasma-interactive components during normal operation, but its low melting temperature and high vapor pressure present great difficul­ties in the design for survivability from off-normal events such as disruptions and runaway electron impact.

Beryllium is often criticized as being an extremely brittle material, because a typical elongation to failure during room temperature tensile tests is roughly 1 %-5%. I However, the ductility increases to almost 50% elongation to failure at 675 K for unirradiated beryllium. This value is significantly larg­er than those of typical nuclear grade graphites, and it sug­gests that a divertor constructed of beryllium brazed to a copper substrate could be far more robust than a comparable graphite design. 3 Unfortunately, the ductility decreases markedly with neutron irradiation (Sec. V), so the thermo­mechanical advantages of beryllium over graphite may only be valid for low neutron fluence devices. Further details on mechanical properties, such as fatigue and creep rate, and the effects of beryllium fabrication are described in review papers. 1.2

III. PLASMA MATERIALS INTERACTIONS

A. OutgaSSing characteristics

Because of their large specific surface areas ( - 0.5 m2 / g),

nuclear grade graphites and carbon fiber composites absorb

TABLE I. Physical properties of beryllium (Refs. t and 2).

Atomic number Atomic weight Density Crystal structure Melting temperature Heat capacity Latent heat of fusion Latent heat of vaporization Electrical resistivity Thermal expansion coefficient

4 9.013 1.85 g/cm3

hexagonal close-packed 1556 K 2.25 J/g K at 775 K 1.3 kJ/g 24.8 kJ/g 15,un em at 675 K lOX 10- 0 em/em at 300 K

1750 J. Vac. Sci. Technol. A 8 (3), May/Jun 1990 0734-2101/90/031750·10$01.00 @ 1990 American Vacuum SOCiety 1750

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 3: Beryllium—a better tokamak plasma-facing material?

1751 Wilson et sl.: Beryllium-a better tokamak plasma-facing material? 1151

250-,---------------------~

225 I q~ I E I ~ / Berylli'.lm (Normal Pur!ty Block) ~175 /

fl00 I·

'13 ~ lB I

1':1 H45' jl POCO AXF-5Q

2:-r I r ,-T--,---,--,---r--'--T-200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400

Temperature (K)

FIG. 1. Comparison of the unirradiated thermal conductivities ofberyllium with several nuclear grade graphites (Refs. 1 and 2).

significant quantities of gases (primarily water vapor) dur­ing air exposure and retain the gases to elevated vacuum outgassing temperatures.4 In order to release the trapped gas, tokamaks generally are required to undergo bakeouts at > 600 K following atmospheric venting. Water is the pre­dominant gas released from the pore surfaces, and 600 K bakeout for one day is sufficient to remove most of this trapped gas. Bakeout at lower temperatures, such as TFTR's 425 K limit, forces the use of alternate techniques such as disruption discharge cleaning to remove the trapped water prior to the initiation of reproducible tokamak discharges.5

Less is known about the outgassing characteristics of be­ryllium. Arc-cast beryllium would be expected to exhibit far less outgassing than typical graphites because of the lack of porosity. However, arc-cast beryllium has poor mechanical properties, and would not be suitable material for a plasma interactive component. Moreen and Passchier6 have com­pared the outgassing characteristics of two commercial grades of beryllium from the Brush-WeHman Co.: hot­pressed 1-220 grade beryllium and a wrought product ex­truded from S-200 grade hot-pressed beryllium. They found that the hot-pressed beryllium released twenty times the amount of gas compared to the wrought material over the temperature range of 373-873 K. Maximum gas release was observed around 725 K. Water was by far the predominant gas released from the hot-pressed beryllium, while hydrogen was the predominant gas in the wrought beryllium case. Both surface desorption and bulk diffusion appear to con­tribute to the outgassing kinetics. Izhvanov and coworkers 7

have conducted a detailed quantitative study of outgassing for various beryllium powders and hot-pressed compacts. Their measurements also show that hydrogen and water va­por dominate the outgassing from beryllium. The absolute specific gas uptake for their beryllium compacts were typi­cally five to twenty times less than that for POCO AXF-5Q graphite.4

B. Erosion characteristics

Carbon displays a complex sputtering behaviorS when ex­posed to energetic hydrogen ion bombardment. Much re-

J. Vac. Sci. Techno!. A, Vol. 8, No.3, MaylJun 1990

-.,' ..•..... -.-•. , ..•. ',. ....... -.-....•......•.. -.-.:.; .... ·.·,'.-.·.··-·-·-·,·f.·.·.·.·.·.~.·· ... ·.·.·,·.·.· ... ·;-..-,':

search has been conducted on chemical erosion of carbon through the formation of volatile hydrocarbons and on radi­ation-enhanced sublimation (RES) whereby radiation­damage-displaced carbon atoms are able to evaporate at rates far above the equilibrium thermal sublimation rate. Be­ryllium, in contrast, shows a far simpler light ion erosion behavior. Initial measurements by Roth et alY'lO showed that the physical sputtering yield for hydrogen peaks at an incident ion energy of approximately 1000 eV with a maxi­mum yield (eroded Be atoms per incident ion) of approxi­mately 3 X 10- 2 for deuterium and 10- 1 for helium. How­ever the similarity of the beryllium results to those of pure BeO led the authors to conclude that surface oxidation ef­fects may have influenced the beryllium sputtering. As dis­cussed in Sec. III D, beryllium has a strong affinity for oxy­gen and the surface state (beryllium oxide or bare metal) depends on the relative gas phase arrival and sputter remov­al rates in the experiment. Ion beam-induced oxidation has been reported by Langleyll for 2.5 keV deuterium ion im­plantation in a 10.- 6 Pa background pressure of oxygen-con­taining gases. On the other hand, Bastasz12 demonstrated that for pressures < 10 - 7 Pa, surface oxygen removal can be accomplished during 1 keY H -+ bombardment at a fiux of _10 14 cm -2 s -I. Subsequent measurements U

-i5 of the tem­

perature dependence of the light ion sputtering of beryllium clearly demonstrated that the initial room temperature be­ryllium erosion measurements were influenced by the pres­ence of a thin surface oxide layer. As shown in Fig. 2, the deuterium sputter yield at 920 K (where the surface oxide is unstable and beryllium diffuses to the surface) is significant­ly higher than that of the oxide-dominated room tempera­ture case, especially for lower incident energies. For the vacuum and ion flux conditions of the Roth et al. experi­ments, bulk beryllium sputtering behavior was observed for temperatures above 600 K. It should also be noted that, un­like carbon, no other changes in sputtering yield were ob­served with temperature up to the onset of significant ther­mal evaporation.

Self-sputtering rates are critical to modeling of erosion lifetimes and plasma contamination from plasma-facing ma­terials. Atoms sputtered into the plasma edge can become multiply ionized, and they can subsequently be accelerated across the sheath potential back to the plasma-facing sur­face. If the self-sputtering yield exceeds unity at the incident energy of these recycling waH atoms, then an unstable ava­lanche efrect can evolve. Because of the increased erosion of carbon at elevated temperature from RES, it is now general­ly accepted that the self-sputtering yield of carbon can ex­ceed unity above a temperature of 1700 K, especially at glancing angles. The carbon bloom phenomenon, observed in many tokamaks when plasma-facing graphite surfaces reach 1700 K, has been attributed to RES-induced self-sput­tering. 16 The self-sputtering of beryllium has recently been assessed by Roth, Eckstein, and Bohdansky17 using available D, He, Ne and Ar sputtering data together with TRIM-SP computer modeling. They conclude that, unlike previous as­sumptions, the self-sputtering yield of beryllium may also exceed unity for energies above 100 eV and angles of inci­dence (with respect to the surface normal) greater than 45°.

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 4: Beryllium—a better tokamak plasma-facing material?

1752 Wilson et al.: Beryllium-a better tokamak plasma-facing material? 1752

10.1

J C .Q

Ul

E

~ -0 a; ;;: OJ c: .~

'§ 10.3 a.

C/)

0 Be at 925 K

0 Be at 295 K

10.4

101 102 103 104

Ion Energy ( eV)

FIG. 2. Comparison of beryllium's measured deuterium sputter yield at room temperature (0) and 925 (0) K. The high temperature data are believed to be representative of clean beryllium (Ref. 15).

c. Hydrogen retention and release characteristics

The understanding of hydrogen interaction with berylli­um is limited. While beryllium is known to form a hydride (i.e., the covalent hydrogen bridge-bonded BeHz) that can be made through organic chemistry techniques involving the decomposition of organometallic compounds, the crystal­line hydride phase has never been synthesized directly from the elements. IS The hydrogen solubility and diffusivity in pure beryllium are also a matter of controversy. Pemsler et al. 19 inferred a minimum hydrogen diffusivity of 8 X 10- 10

cm2/s at 1025 K and a solubility limit of 9X 1014 H/cm3

(7X 10-9 atom fraction) over the temperature range of 625-1325 K based on proton implantations. A detailed tri­tium charging/desorption experiment by Jones and Gib­sonzo reported that the solubility oftritium in beryllium was 6 X 10 -6 atom fractionl atm l!2, independent of temperature. Al'tovskiy et at. 21 measured gas-driven permeation through beryllium membranes and tubes, and reported large varia­tions in permeability (the product of diffusivity and solubil­ity), depending on the beryllium grain size. More recently, Swansiger22 has determined the tritium solubility in high purity beryllium, using tritium gas charging followed by tri­tium dissolution counting. As shown in Fig. 3, he found an intrinsic lattice solubility which is extremely endothermic, in marked contrast to the virtually athermal solubility re­ported by Jones and Gibson. 20 Swansiger also found evi­dence for strong extrinsic trapping of tritium at the::::: 1-10 ppm level for temperatures below 600 K. The temperature

J. Vac. Sci. Techno/. A, Vol. 8, No.3, May/Jun 1990

Temperature (deg C) 510 440 390 350 310 280

101

• s • E - \ co I

(I)

co 0\ • ~

E 0\ a. a.

~ u

! 'E 0 ...

..s • >-

i I-;;! CD :::l ..J 0 C/)

10° 12.0 14.0 16.0 18.0 20.0

lIT (deg K) *10.4

FIG. 3. Arrhenius plot of the solubility of tritium in beryllium for three different sample configurations and purities. AI, 99 .8o/e, Be, 0.057 --D.08 mm thick (_); A2, 99.8% Be, 0.10--0.15 mm thick (.); HI, 98.5% Be, 0.057-0.08 mm thick (0) (Ref. 22).

dependence of this extrinsic bulk trap suggests an activation energy for de trapping of ;:::: 1 to 1.5 eV. The best fit to Swan­siger's data gives a solubility S of tritium in beryllium equal to

S = 18.4 exp( - 1.0 eV /kT) atom fraction/atm l!2.

Perhaps surprisingly, the hydrogen solubility in beryllium exceeds that of graphite23 for temperatures above 1000 K. However, beryllium is an endothermic hydrogen occluder, while graphite is exothermic; hence the ratio of hydrogen solubility in beryllium compared to graphite decreases mar­kedly as the temperature decreases.

Of course tritium uptake in a material depends on the kinetics of tritium migration, as well as the intrinsic solubil­ity. Jones20 reported that diffusivity CD) of tritium in berylli­um was:

D= 3x 10-. 7 cxp( - 0.19 eVlkT)cm2/s.

Hydrogen is far more mobile in beryllium than in graphite2.l

for temperatures below beryllium's 1556 K melting point The nature of the trapping of hydrogen isotopes in berylli­

um during energetic ion implantation has been studied by several investigators. After low ftuence implantation, hydro­gen is observed to accumulate in sites near the basal plane. 24

At higher fluences the precipitation of microscopic bubbles and more macroscopic blisters have been reported. 19,25 As in the case of other low atomic number materials such as graphite and TiC, implantation of hydrogen isotopes can lead to significant saturation trapping in the near-surface

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 5: Beryllium—a better tokamak plasma-facing material?

1753 Wilson at sl.: Beryilium-a better tokamak plasma-facing material? 1753

region ofberyUium. Liu et al.26 reported a saturation reten­tion of 4X 1017 D/cm2 following 10 keY D-'- implantation below 425 K. X-ray analysis of the samples after implanta­tion did not show any evidence of crystalline beryllium hy­dride, although they did not rule out the possibility of an amorphous hydride state. Langleyl' observed similar trap­ping of deuterium in a surface layer extending over the range of the implanted D +- ions, although the magnitude of the reported retention was significantly higher than for other investigators. Pontau et al.27 also observed that, unlike the case of nickel, during simultaneous H +- IHe -I implantation, the implanted hydrogen competes with the helium for trap­ping sites in beryllium. This observation is consistent with the stronger hydrogen trapping in beryllium compared to nickel.

The most systematic studies of the retention of implanted hydrogen isotopes in beryllium have been conducted by Wampler28 and Moeller et al. 29 In the experiments per­formed by Wampler,28 nuclear reaction analysis was used to determine the deuterium retention during irradiation and during postirradiation annealing. Deuterium ion energies of 0.5, 1.5, and 3.0 keY were used in the experiments. At low fiuences" virtually aU of the deuterium was retained, but the retention saturated at higher fiuences. At room temperature, the saturation concentration was determined to be 0.31 DI Be. Similar behavior was reported for an experiment per­formed at 473 K, but a lower saturation value was seen. An isochronal annealing program where the temperature was increased every ten minutes with 50 K increments showed the release of hydrogen to be characterized by a relatively narrow detrapping stage at 400 K followed by a broad stage at approximately 675 K. For the saturated samples, most of the release occurred at the lower temperature, but no release was seen at this temperature for the low-dose implants. The release behavior was described as being controlled by ale V trap with a concentration 0.25 atom fraction and a 1.8 eV trap with a concentration of 0.05-0.10 atom fraction. The higher energy traps must be filed prior to occupation of the lower energy ones.

Experiments very similar to those reported by Wampler were also performed by Moeller et al. 29 While greater diver­sity in the implant energies and temperatures were used in this study, the results and conclusions were very similar. Figure 4 compares the data of Moeller et al. for the deuter­ium retention in beryllium as a function of temperature dur­ing postirradiation anneals to that of Wampler. The bom­bardments in both cases were carried to saturation at room temperature. The results have been changed to fractional retentions to account for the difference in implant energies. Also shown in the figure are the data determined by Doyle et al. 3 ] for deuterium retention in room temperature saturated graphite sample. The data demonstrate that agreement has been seen in the release characteristics ofberyHium irradiat­ed with energetic hydrogen isotopes. The data are compared to that for a saturated graphite surface for two reasons. The first reason is to show that the use of beryllium in moderate temperature plasma-facing regions of a fusion reactor could result in lower near-surface hydrogen isotope retention. The second reason is to suggest that similar mechanisms may be

J. Vac. Sci. Technol. A, Vol. S, No.3, May/Jun 1990

.-.~.-., •.•••• ,; ........ :.:.:.:.:.:, •••• .' •••••• ;:.;.:.: •••••••• ' ....... -; ••.• -.:.:.; •••• '7 •••.• ~ •••.• -.~.;.:, •••• ; ...... -.:.~.:.:.:.:.: •• , ••••••• > •••••••••••• .-.: •• -••••••••••• ; •••• -.-••• -.

1.0

0.9 -

0.8 c 0 0.7 "E d.l a; 0.6 c:: (\l 0.5 c 0

.~ 0.4

l.J.. 0.3

:~~ 0

Be

Graphite

Doy~e et ai

I I I I I I I I 100 200 300 400 500 600 700 800

Anneal Temperature ("C)

100e

FIG. 4. Comparison of the thermal annealing of heryllium (Refs. 28, 29) and graphite (Ref. 31) implanted with deuterium at room temperature.

controlling the retention in the two materials. Graphite31-

B

is known to form an amorphous saturated layer with a hy­drogen-to-carbon ratio not significantly different from that of beryllium. For beryllium, the 1 eV trap energy likely cor­responds to amorphous beryllium hydride. 34 Beryllium hy­dride decomposes at 400 K,35 the same temperature as that seen for the sharp release stage for the deuterium-implanted beryllium. The 1.8 eV trap is likely due to trapping at point defect sites. In the work by Swansiger22 on the solubility of tritium in beryllium, evidence of intrinsic trapping at tem­peratures in the range of 500 to 700 K was reported, consis­tent with the higher trap energy seen for the implantation experiments.

Figure 5 shows previously unpublished data by Causey30 on the retention of deuterium and tritium in beryllium ex­posed to a high fluence of 100 e V ions. The measurements by acid dissolution of the samples, followed by liquid scintilla­tion counting of the tritium content. When corrected for implant ranges, the data agree with those reported by MoeHer et at?9 for the temperature dependence of the deu­terium retention in beryllium. The data given by Causey et al. 3b for tritium retention to graphite are also shown in Fig. 5, again to demonstrate the qualitatively similar but quanti­tatively different near-surface retention behaviors of the two materials. The energies and fluences used in the two plasma exposures were the same.

One important difference in the tritium retention charac­teristics of beryllium compared to those of graphite is the behavior of the codeposited layer. Eroded graphite is ob­served to pump, or codeposit with, hydrogen isotopes when it is deposited on nearby plasma-facing surfaces. 37 However, as shown in Fig. 6, eroded beryllium behaves quite different­ly from graphite. In his experiment Hsu30 initiated a Penning plasma discharge in a closed vessel and monitored deuter­ium pressure changes. When graphite electrodes were used, the system deuterium pressure steadily decreased with time due to the codeposition of eroded carbon with deuterium on the chamber walls. Beryllium electrodes, on the other hand, produced no sustained pump-out of deuterium working gas. This lack of pumping indicates that the redeposited beryl1i-

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 6: Beryllium—a better tokamak plasma-facing material?

1754 Wilson et al.: Beryllium-a better tokamak plasma-fac!ng material? 1754

~: ---------------~ -POCO~:hi!e-1

~x 14

E () • 2

~<O~ i 8~ ~ sj ~ 4J

2~

A- BerylliuM

12 x 1020

TJ'CrJ'

/100 eV

L2 5Hwrs

~ li'i O+--<---r--~I---.___r--,___r___r--.--4 200 300 400 500 600 700 800 DOD 1000 1100 12DO

Temperature (K)

FIG. 5_ Tritium retention in beryllium (i'.) and graphite (0) following plasma exposure at various temperatures (Ref. 30).

urn layer is far less chemically active than that of carbon, and it reflects the difficulty in attempts to form beryllium hy­dride directly from the elements.

D. Chemical Effects

Beryllium reacts rapidly with oxygen or oxygen-bearing compounds to form beryllium oxide (BeO), with a large exothermic enthalpy off ormation of -- 144 kcalimoL 38 Be­ryllium oxide is a brittle, refractory material with a melting temperature of 2825 K.l A saturated beryllium oxide layer has been observed to form on clean beryllium surfaces for oxygen exposures of only 10 langmuirs. 39 Langley II report­ed the growth of a beryllium oxide layer during deuterium or helium bombardment under relatively poor vacuum condi­tions. Rates of oxide growth of 0.007 oxygen/D ion and 0.09 oxygenlHe ion respectively were observed, and the growth was attributed to radiation-enhanced diffusion of beryllium through the oxide layer followed by oxidation with vacuum impurities at the surface. Bastasz [2 showed that under UHV conditions the surface BeO layer can be eroded away at a rate

30

25

0 0

§ 20 III • f-.s

ll: :J <f) 1" '" 1'-'

,,1 0-

N 0

51

ot-

r- Plasma On r- Plasma Off

o 0

000000 ~ooooooooooo 00 Ie Beryllium

III

I 500

III III

III • III • • Graphite ••

ee e

r 1000

Time (sec)

1500 2000

FIG. 6. Measurements of the working ga~ pressure changes for a Penning discharge operated in a closed vacuum vessel with either graphite (el or beryllium (0) electrodes (Ref. 30).

J. Vac. Sci. Techno!. A, Vol. 8, No.3, MaylJun 1990

comparable to bulk BeO sputtering. At temperatures ~ 550 K, however, beryllium can also thermally diffuse through the oxide layer, so that surface oxide contamination is eli­minated, and bulk beryllium sputtering is observed. 15 Possi­ble chemical erosion of beryllium by oxygen sputtering has not been investigated. Beryllium oxide has extremely poor resistance to neutron damage, I and hence should not be con­sidered as a plasma-facing material.

The formation of beryllium carbide (Be2C) is also of in­terest since carbon is a major impurity in most tokamaks due to the extensive use of graphite as a plasma-facing material. JET has also employed the deposition of thin beryllium lay­ers onto their graphite inner-bumper limiter tiles by berylli­um evaporators, in order to produce an all-beryllium ma­chine. Beryllium carbide has an enthalpy of formation of - 28 kcaIlmoL 38 Nieh et al. 40 have studied the formation of

beryllium carbide using thin layers of carbon deposited onto beryllium surfaces. They observed that an interfacial reac­tion between carbon and beryllium occurs at 725 K with the growth controlled by beryllium diffusion in the carbide. Mills et al.41 have studied the interdiffusion of thin (-10 t-tm) beryllium layers deposited on graphite substrates in a manner meant to simulate the JET deposition process. Sig­nificant Be2C formation ( > 1 pm conversion) was not ob­served in furnace anneals or in electron beam high-heat fiux tests until temperatures above 1000 K. This higher transfor­mation temperature may have resulted from the technical surfaces used in the Mills measurements. Due to its brittle and hygroscopic nature, beryllium carbide should be viewed as a potentially deleterious by-product of the usc ofberylli­um in a fusion rector, and may not be a suitable plasma­facing material.

IV. HIGH HEAT FLUX PROPERTIES

Although beryllium has higher strength, ductility, and fracture toughness than graphite, beryllium is more suscept­ible to damage resulting from thermal transients (thermal shock or fatigue) due to its higher elastic modulus and high­er thermal expansion.42

.43 Beryllium also has a relatively low

melting temperature (1556 K) which leads to somewhat more restrictive operating temperature limits than graphite and the possibilities of melting occurring during power over­load situations. Beryllium, like other metals, has the advan­tage of being able to absorb large deformations without cata­strophic fracture such as would occur with ceramic type mateIials. This ability to absorb large plastic strains without fracture has led to a large interest in studying the thermal fatigue behavior ofbery1lium for fusion components.

A. Steady~state behavior

Under normal steady-state conditions, a nominal berylli­um tile will sustain a temperature gradient that is roughly 100 K per millimeter of thickness for each 10 MW 1m2 of incident flux. Hence, as a rule of thumb, a 3 mm beryllium tile exposed to a surface heat flux of 10 MW 1m2 would gen­erate a temperature gradient of 300 K across the beryllium. At 20 MW 1m2, the gradient would be doubled, and so forth. Finite element models using temperature-dependent proper-

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 7: Beryllium—a better tokamak plasma-facing material?

1755 Wilson et sl.: Beryllium-a better tokamak plasma-facing material? 1755

ties and detailed coolant heat-transfer models are used to calculate temperature and stress distributions in actual com­ponents. In general, three millimeters is a limiting thickness for a beryllium tile or coating without reaching excessive surface temperatures at the heat fluxes expected in a diver­tor.44

The stress state that exists in the tile is a combination of three factors, namely fabrication stresses, external loads, and thermal stresses. Fabrication stresses can be generated by a variety of methods. Common sources are differential thermal expansion created when materials with different co­efficients of thermal expansion are joined at high tempera­tures and allowed to cooL An example of this effect is braz­ing which is usually carried out at 900-1300 K. A second major source of fabrication stresses is shrinkage during cool­ing from a liquid phase such as in weld joints or from plasma spraying. In some cases, these stresses are large enough to cause component failure even before exposure, and in all cases they must be carefully considered since they are addi­tive to stress generated by other mechanisms.

Stresses generated by external loads are usually straight­forward to calculate and include forces like gravity loads and coolant pressure. In some cases, there may be electromag­netic loads or vibration loads that may also be significant even under normal pulse conditions. The thermal stress state is generated by the large temperature gradients that are pres­ent during the plasma discharge. Depending on the compo­nents constraints, these stresses can be quite large and may exceed the materials yield stress at the surface. Under these conditions, the surface materials will yield upon heating, and a residual stress state will be generated due to the permanent deformation of the surface during cooling. This behavior can lead to large, cyclic swings in the plastic strain and to failure by low cycle fatigue. One method by which these thermal stresses can be relieved is by cutting grooves at appropriate spacing to allow room for thermal expansion of the surface layers. An example of this type of situation is shown in Fig. 7.45 In this test, a pair of beryllium tiles were exposed to a total of 5000 thermal cycles at a surface heat flux of 4-4.5 MW 1m2

• One tile was grooved with a groove spacing of 1 em while the second tile was tested without grooves. After expo­sure to the cyclic heat flux, the ungrooved tile had developed a large surface crack while the grooved tile was only slightly affected by the testing.

A related problem that can develop under thermal cycling is that of thermal ratcheting. If two materials with widely different thermal expansion coefficients and different me­chanical properties are joined. then it is possible to develop a situation under thermal cycling where the materials will un­dergo a plastic strain cycle that will ratchet in one direction on each cycle, leading to a growth of the component and subsequent failure. This effect must be avoided by analyzing and testing components to avoid this ratcheting regime.

e, Off~normal thermal response

The thermal energy deposited during either plasma dis­ruption or runaway electron events can cause melting andl or vaporization of the plasma-facing surface. Typical disrup­tion conditions are an energy deposition of 500-1000 J/m2 in

J. \lac. Sci. Techno!. A, Vo!. 8, No.3, May/JIJI11990

~.-.-,>.'.'.'.~.'.'.<.,:": .... :.:.;.,. •••• ;>., .. ;-;.:.:.:.:.:.; •••••••• ~ ••• > •••• .::.:.:.:.: ••••• , •••••••••••••••• ~, •• -o: •••••••• '.. ,; •••••• -0'.-........... .

FIG. 7. Photograph of two beryllium tiles exposed to 10 s pulses with 4000 cycles at 4 MW/m2 and 1000 cycles at 4.5 MW/m2 (Ref. 45).

times of 0.1 to 1.0 ms. A detailed energy balance, including surface cooling by evaporation and radiation, has been per­formed for beryllium under these disruption conditions. The computer code SOAST is used to solve this two moving boundaries problem (i.e., the liquid-vapor and the solid­liquid interfaces) and predict the surface temperature rise, the melt layer depth, and the amount of material lost by vaporization.46

.47 Since vapor shielding is not included in

these calculations, the results will be conservative. Figure 8 compares the predicted melt layer thickness for a

thermal quench time of 0.1 ms for beryllium and tungsten .44

For a canonical disruption energy density of 1000 J/cm2, we

see that only 10 pm ofberyUium are melted, as compared to 30 Itm of tungsten. This result occurs even though beryllium melts at 1556 K, whereas the melting point of tungsten is 3683 K. The explanation for this result is that, under very intense energy deposition, a nearly instantaneous thermal balance is established between the energy deposited by the plasma and cooling by vaporization of beryllium. The vapor­ization temperature of beryllium is variously reported as 2750 to 3240 K, as compared to over 5900 K for tungsten. Similarly, the latent heats of melting and vaporization of beryllium (2.8 and 53.55 kcaI/mol, respectively) are also lower than the corresponding tungsten values (9.6 and 190 kcallmol, respectively). This explanation is consistent with the results in Fig. 9, which shows the amount of material that is vaporized for a thermal quench time of 0.1 ms.44 At an energy density of 1000 J/cm2

, we see that the thickness of

~ 100~----------------------------~ r: Thermal quench time = 0.1 ms o t 80 Tungsten ! CTo" 150Q °C)

~ 60 ~ U) (!) c ~ 40 1§ _ 20 OJ ::

O~----r---~----~--~~----~ o 200 400 600 800

DepOSited Energy (J/cm2) 1000

FJG. 8. Calculated amount of material melted during a 0.1 ms plasma dis­ruption for beryllium and tungsten (1;, is the initial material temperature prior to the disruption) (Ref. 44).

- •• -.... < ••••• -.-.~.--. ••••• ~ ••• ~ • ..". •• -.-•• -.-••••• ··········-.-.·.·'!'.'.·.·.· • ..-.··:·.·.·.·.·.v.·.·,..·.· Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 8: Beryllium—a better tokamak plasma-facing material?

1756 Wilson et al.: Berymum-a better tokamak plasma-facing material? 1756

vaporized materia! is lOO,umforberyl1ium, 75,um for graph­ite, and 70,um for tungsten. Thus, there is more vaporization of beryllium than either graphite or tungsten. It is important to note, however, that even at this high energy density, dif­ferences in vaporization of the three materials are relatively small.

Another area of possible concern is the small surface cracks that form when molten metals resolidify. These reso­lidification cracks could serve as thermal fatigue crack ini­tiation sites and hence accelerate this type of damage. While this effect has not been extensively studied due to the diffi­culty of simulating disruptions in the lab, it may not be a critical issue since thermal fatigue cracks form after a few hundred cycles in most materials and they only grow to depths where the thermal stress level is above the yield stress.45

In the ISX-B experiment,48 deliberate extensive melting of the beryllium limiter surface was done in order to study the effect of oxygen gettering by the beryllium on plasma perfor­mance. Even with this severely melted and irregular berylli­um surface, good plasma performance was achieved. The intentional melting of the beryllium limiter surface in ISX-B was done in a manner such that the local temperature re­mained above the melting point long enough to produce a deep melt layer and consequent flow of the molten material. For the much thinner and more transient melt layers pre­dicted by the disruption calculations (e. g., ~ 10 11m), the melt layer is expected to be much more stable. If the melt layer resolidifies without significant movement, then the problem of material loss from disruption heat loads is mana­geable. Even if the molten beryllium is redistributed by eddy current forces before it solidifies, the total erosion is of the same magnitude as that expected for graphite.

Compared to disruptions, the thermal effects from runaway electrons are confined to a much smaller area, but the localized damage is expected to be more severe. Deposit­ed energy densities for runaway electrons have been estimat­ed to be as high as 80 MJ/cm2 over areas of a few square centimeters.49 Energy densities of this magnitude will cause severe melting/vaporization in virtually all materials and can lead to surface spallation. These events have been ob­served to cause severe damage to graphite tiles in present day tokamaks. While the beryllium in the strike region will prob-

Ul c 100.-~--~~--~~--~--------~ E Thermal quench time = 0.1 ms I 80 ~ initial temperature = 700°C

f/)

~ 60 c

.!<!

~ 40 t--g 20 N .;::

~ O+-~~~--~~--~--r-----~----~ .g;. 0 200 400 600 800

Deposited Energy (J/cm2)

1000

FIG. 9. Calculated amount of material vaporized during a 0.1 ms plasma disruption for beryllium, graphite, and tungsten (Ref. 44).

J. Vac. Sci. Technol. A, Vol. 8, No.3, May/Jun 1990

ably be severely melted, the most critical issue for runaway electron damage is that of coolant line damage. Due to the deep penetration and large spatial dispersion of the high­energy electrons, a thick armor may be required to avoid overheating of the coolant channels with subsequent coolant leakage. Since thicker armor implies higher surface tempera­tures, then the only possible solution may be local regions that are either un cooled or with thick armor that receives low heat flux during normal operations.

V. NEUTRON IRRADIATION

A. Helium Effects

The properties of neutron irradiated beryllium have re­cently been reviewed for both plasma-facing and tritium­breeding blanket applications. 1.2 Unlike graphite which un­dergoes a complex change in lattice microstructure from atom displacement during neutron irradiation, neutron-in­duced deformation in beryllium is dominated by the nuclea­tion and growth of helium bubbles. Helium is produced in beryllium by the (n,2n) and (n,a) transmutation reactions. In a fusion environment the (n,2n) reaction dominates, pro­ducing nearly 90% of the helium generation.50

•51

Helium production in beryllium is calculated to be ~ 3700 at. ppm per MW-y/m2 neutron ftuence. 52 Helium is initially trapped within the beryllium lattice in submicroscopic clus­ters. At higher neutron fiuences massive helium-bubble-in­duced swelling occurs, especially at elevated irradiation or postanneal temperatures.51.S:l.54 Figure 10 summarizes the neutron ftuence lifetime as a function of irradiation tempera­ture for both beryllium and nuclear grade graphite, based on data from the FED/INTOR study. I Lifetime is defined as the onset of helium bubble nucleation for the case ofberylli­urn, and the beginning of positive expansion in the case of graphite. While rapid swelling occurs above 1000 K in beryl­lium, the onset fiuence for swelling is extended considerably for temperatures below 800 K where helium mobility is pre­sumably limited. It can be seen that beryllium's irradiation lifetime is comparable to that for nuclear grade graphites 1

Temperature (K)

FIG. 10. Neutron irradiation lifetimes for beryllium and graphite, based on data in the FED/INTOR study (Ref. 1).

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 9: Beryllium—a better tokamak plasma-facing material?

1757 Wilson et al: Beryliium-a better tokamak plasma-facing material? 1757

over the typical temperature range of actively cooled plas­ma-facing materials,

Helium generation has significant effects on the mechani­cal properties of materials. Beryllium undergoes hardening from dislocation pinning and grain boundary decohesion from the helium bubble nucleation at interfaces. The yield strength generally increases while thc ductility diminishes foHowing neutron irradiation. However, since deformation is dominated by microscopic helium bubbles rather than iso­latcd point defects, the thermal conductivity ofberyHium is not diminished by neutron irradiation. I This is in marked contrast to graphites, which suffer from major decreases in thermal conductivity with irradiation fluence. 1 The atomis­tic nature of the helium bubble nucleation and growth sug­gests that porous beryllium microstructures, such as from powder metallurgy or plasma spray technologies, will not be effective in releasing significant amounts of helium 1,2 under fusion reactor conditions. Measurements by Wilson et ai."s have shown that the helium release kinetics during helium ion implantation are not significantly different between 85 % and 100% dense plasma-sprayed beryllium microstructures.

So Tritium Effects

Tritium is generated in beryllium by neutron irradiation, primarily through the direct "Be (n/H) 7Li reaction, al­though the (n,a) reaction can also contribute. 2 While the tritium production is less than one percent of the helium generation in beryllium, the buildup of tritium in a beryllium plasma-facing or blanket component could still be a signifi­cant safety concern for a device such as ITER.

Causey et al.56 have recently measured the tritium content of a beryllium sample irradiated at the Advanced Test Reac­tor (INEL) to a fast neutron fiuence of 5 X 1022 nl em 2 at 350 K. They found that essentialy all of the tritium generated by transmutation was still in the sample, resulting in a tritium concentration of 0.2 at. %. Scaling these results to fusion reactor conditions, 60 Mg of beryllium in a blanket/first wall exposed to 3 MWy/m2 fluence would generate as much as 6.6 kg tritium inventory.

VI. FUSION DEVICE PERFORMANCE

A. Present~day devices

Only three tokamaks have operated with beryllium as the limiter or the first-wall material. The first experiments were performed by UNITOR,57 which were then followed by ISX_B.5s

,59 Both tokamaks investigated the effects of small beryllium limiters on plasma behavior CUNITOR had side limiters at two toroidal locations and ISX-B had one top

. limiter) in support of the more ambitious beryllium experi­ment in JET. At the time of this writing, JET has just begun operation with toroidal belt limiters made out of bulk beryl­lium and the first waH coated with a thin layer of evaporated beryllium.60

Both UNITOR and ISX-B showed that once beryllium is evaporated from the limiter and coated over a large segment of the first wall, gettering leads to significant reduction of impurities. Figure 11 shows the emission intensity from 0 VI and Be IV in ISX-B. When the heat load on the beryllium

J. 'lac. Sci. Technol. ft., Vol. 8, No.3, May/Jam 1990

limiter was increased to the point of evaporating beryllium, the oxygen concentration was decreased dramatically (solid line before the evaporation and dotted line after the evapora­tion). Although the concentration of beryllium in the plas­ma was increased, its contribution to Zcff was more than compensated by the reduction of oxygen, carbon, and metal impurities. 58

•61 The plasma Ze!T was observed to be reduced

from ~ 2.4 to near unity when beryllium gettering was in effect. There were no direct measurements of beryllium con­centration in the plasma central core in either UNITOR or ISX-B. There were, however, edge measurements of Be I and Be II in UNITOR using laser-induced fluorescence and Be IV using emission spectroscopy in ISX-B. Using impurity transport models, the core concentration was then estimat­ed. A similar calculation was performed for the plasma pa­rameters in ASDEX,,2 with an assumed beryllium divertor plate. The inferred beryllium density in the core in all of these estimations ranged from 1 % to 6% of electron density. There was a negative aspect associated with beryllium oper­ation during the ISX-B campaign, The reduction in plasma impurities was not observed until the limiter surface was partially melted causing beryllium to be evaporated and coated on the first wall. Once melting did occur, the droplets made subsequent evaporation more likely but hard to con­trol. The consequent strong reduction in plasma impurities associated with gettering then made discharge reproducibil­ity hard to obtain. If a much larger plasma contact area is already covered with beryllium, however, one does not need to rely on limiter melting to obtain the beneficial effect of beryllium. This effect can be achieved by using large area

········155 kA 400

a:: 300 o VI

~ 1032A z 2 200 V) IJ)

~ UJ 100

~~ ..... ~.~: 0

40 [

a: Be IV ~ 30 .... «o ...... ~

Cl

20 ~ 7sA D~· ~ . . :z . . 0 .~,: : IJ) .' en ::E : '. w 10 ~ :~

" .'

...... ~ ..... ~ ..... ~ ...... 0

"' .......

0 100 200 300 400

TIME (ms)

FIG. II, 0 VI and Be IV radiation for ungettered (solid) and beryllium· gettered (dashed) discharges in ISX-B (Ref. 58).

....•.... ~ .. ~ .. -: ... -....•. -.-: ..... . Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 10: Beryllium—a better tokamak plasma-facing material?

1758 Wilson et al.: Beryllium-a better tokamak plasma-facing material? 1758

beryllium limiters, or coating the inside wall with beryllium which is the approach taken by JET.

JET began beryllium operation60 in 1989 with the evapo­ration of thin beryllium layers (-100 A average thickness per deposition) onto the plasma-facing surfaces of the de­vice. Significant changes were observed in plasma perfor­mance. Oxygen impurity content in the plasma was reduced tenfold and the fraction of radiated power decreased. Large wall pumping of deuterium from the plasma discharge was also reported. A maximum QOT equivalent value ap­proached 0.5. Most recently the JET graphite belt limiter was replaced by a beryllium tile design. Energy deposition of 180 MJ was achieved on the belt limiter, a value that is twice that for the graphite limiter. It is planned that JET will rely on beryllium even more as a plasma-facing material.

One should not overlook certain potential benefits from operating a reactor with graphite. In the ITER design stud­ies, it has been noted that the erosion lifetime of divertor plates is a major concern. It has been suggested that a possi­ble solution to this problem is to intentionally inject impuri­ties that are of the same element as the wall material into the edge plasma. The enhanced radiation loss not only spreads the heat load over a larger surface area to reduce the peak heat flux, but also reduces the energy ions impinge onto the wall. If radiation loss is dominated by Bremsstrahlung, which goes as the square of the atomic number Z 2 then (61 4)2 = 2.25 more Be is needed to accompish the same radi­ation loss as C. Similarly, this same amount of Be will result in 6/4 times more fuel dilution. Despite this benefit, a carbon wall will more than likely result in higher erosion than a Be wall because of the well-known effects from chemical ero­sion and radiation-enhanced sublimation. 63

B. Future reactors

BerylIium has been adopted as a plasma-facing material for a number of magnetic fusion energy reactor designs. The MARS tandem mirror reactor64,65 used beryllium as a pro­tective tile material in its halo recycler design. Both ST AR­FIRE66 and FED/INTOR 1 adopted beryllium as their base­line material for pump limiter and divertor designs, In the expected edge temperature range of 100-400 eV, high atom­ic number materials were not considered viable because of self-sputtering considerations. Metals like tungsten or mo­lybdenum do not become attractive until edge temperatures are reduced below 50 eV to eliminate the possibility of pI as­ma contamination by sputtered atoms. Beryllium was select­ed over graphite as the low atomic number material option because of graphite's poor neutron irradiation response and severe chemical erosion. The disruption melting was consid­ered as the main drawback to beryllium, but was not serious enough to eliminate beryllium from selection.

Using essentially the same database for graphite and be­ryllium, the designers of TIBER n67,68 and ITER69 have selected graphite over beryllium as the baseline plasma-fac­ing material. The TIBER II and ITER designs are driven more by disruption survivability than previous design stud­ies, so graphite is given the edge in the competition. How­ever, the reactor designers have not yet seriously considered

J. Vac. Sci. Technol. A, Vol. 8, No.3, May/Jun 1990

a significant advantage of beryllium over graphite: the po­tential for in situ repair.

c. In situ repair

It is clear that portions of the plasma chamber surfaces in advanced fusion devices will experience significant erosion and localized damage. Therefore, technology must be devel­oped to repair or replace such surfaces. Plasma spray is a high deposition rate coating process which offers the poten­tial for ill situ repair of eroded or damaged surfaces in fusion devices. In the plasma spray process, a powder ofthe materi­al to be deposited is fed into a small arc-driven plasma jet, and the resulting molten droplets are sprayed onto the target surface. Upon impact, the droplets flow out and quickly so­lidify to form the coating. Because graphite does not have a stable molten phase, it cannot be plasma spray deposited. However, beryllium can be plasma sprayed and, with recent process improvements, high quality beryllium coatings ranging up to morethalll em in thickness have been success­fully produced. Using specially developed spray technology, beryllium deposition rates up to 450 g/h have been demon­strated with 98% of theoretical density in the as-deposited material. Therefore, localized areas that have been damaged by disruptions or other off-normal events could, in principle, be repaired in a few hours. For larger repairs, it may be cost effective to use mUltiple plasma spray guns to simultaneous­ly cover different areas within the fusion device. For exam­ple, it has been estimated that the heaviest erosion on the divertor surface of the proposed ITER device may extend over a total surface area of approximately 6.8 m2

• Four plas­ma spray guns operating simultaneously could deposit 1 mm of beryllium over this entire surface area in less than eight hours.

VII. CONCLUSIONS

Beryllium offers many advantages over graphite as a plas­ma-facing material. Beryllium possesses superior thermal and mechanical properties that make it easier to design to steady-state heat removal. The outgassing characteristics of beryllium will lead to reduced conditioning time, and its oxygen affinity and gettering properties lead to reduced plas­ma contamination compared to graphite. Beryllium has a low hydrogen affinity, however, which will lead to reduced in-vessel tritium inventory due to codeposition processes. The neutron-induced property degradation, such as thermal conductivity, is far less severe for beryllium than for graph­ite. Beryllium's main liability, its low melting temperature and resulting poor disruption response, is mitigated by the ability to repair beryllium plasma interactive components by ill situ plasma spray. Beryllium should therefore be consid­ered as a strong alternative to graphite in future reactor de­signs such as ITER, and a more vigorous research and devel­opment program should be undertaken.

ACKNOWLEDGMENT

This work was supported by the Department of Energy under Contract No. DE-AC04-76DP00789.

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34

Page 11: Beryllium—a better tokamak plasma-facing material?

1759 Wilson et sl.: Beryllium-a better tokamak plasma-facing material? 1759

'w. M. Stacey. M. A. Abdou, D. B. Montgomery. J. M. Rawls, J. A. Schmidt, T. E. Shannon, R. Thome, and B. F. Cranfill, U. S. FED-IN­TOR Activity and U. S. Contribution to the International Tokamak Reac­tor Phasc-2A Workshop, V. 1. USA FED-INTOR/82-1, Oct. 19S2, p. 293.

'T. J, McCarville, D. H. Berwald, W. Wolfer, F. J. Fulton, J. D. Lee, R. C. Maninger, R. W. Moir, J. M. Beeston, and L. G. Miller, VCID 20319. Jan. 1985.

JR. T. McGrath, J. A. Koski, and R. D. Watson et al., Sandia National Laboratories, SAND89-·0901, July 1989.

4A. E. Pontau and D. H. Morse, J. Nue!. Mater., 14-t-143, 124 (1986). 'H. F. Dylla and the TFTR Team, J. Nucl. Mater., 145-147,48 (1987 l. ('H. A. Moreen and A. Passchier, in AIAA Conference 84-0951 Oil Struc-tures, Structural DYllamics, and Materials, 1'tiay 1984 {AIAA, Palm Springs, CAl p. 305.

71. A. Izhvanov, V. V. Kromonozhkin, B. G. Drozdov, and V. G. Vikhins­kii, Poroshkovaya Metatlurgiya 1 (1985),56, English translation: SOy.

Powder Metal!. Met. Ceram. 24, 49 (1985). RJ. Roth, J. Bohdansky, and K. L. Wilson, J. Nuel. Mater., 111/112, 775 (1982).

9J. Roth, J. Bohdansky, and W. Ottenberger, Max-Planek-Institut fur Plas­maphysik Report No. IPP 9/26,1979.

,oJ. Roth, J. Bohdansky, R. S. Blewer, W. Ottcllberger, and 1. Borders, J. NucL Mater. 85/86, 1077 (1979).

"R. A. Langley, J. Nne!. Mater. 85/86, 1123 (1979). 12R. Bastasz, Thin Solid Films 121, 127 (1984). uJ. Hohdansky, J. Roth, J. Nue!. Mater. 1221123, 1417 (1984). '4Workshop on Beryllium for Plasma-Side Applications. Reactor Technol-

ogies Branch, USDOE, Germantown, MD, July 6, 1983 (unpuhlished). "I. Roth, J. Nucl. Mater. 145-147, 87 (1'187). "'Sandia National Laboratories Report No. SAND89-3012 (in press). '7J. Roth, W. Eckstein,andJ. Bohdansky,J. Nuc!. Mater. 165 199 (1989). '"w. M. Mueller, J. P. Biacklcge, and G. G. Libowitz, l'l1etal Hydrides

(Academic, New York, 1968). 19J. P. Pemsler and E. J. Rapperport, Trans. Metall. Soc. AIME 230, 90

(1964 ). 20p. M. S. Jones and R. Gibson, J. Nuc!. Mater. 21, 353 (1967) and AWRE

Report No. 0-2/67. 21R. M. AI'tovskiy, A. A. Ercmin, et al., Russ. Metall . .3 (1981). !OW. A. Swansiger, 1. Vac. Sci. Techno!. A 4, 1216 (1986). 2'R. A. Causey, J. Nnc!' Mater., 162-164, 151 (1989). C'

4 R. Viandcn, E. N. Kaufmann, and J. W. Rodgers, Phys. Rev. 22 63 (l980).

!'H. Verbeek and W. Eckstein, in Applications of Ion Beams ta j\fetal~, edited by S. T. Picraux, E. P. EcrNisse and F. L. Vook (Plenum, New York, [974) p. 607.

26M. B. Liu, L Sheft, and D. M. Gruen, J. Nucl. Mater. 79 267 ([979). 27A. E. Pontau, W. Bauer, and R. W. Conn, J. Nue\. Mater. 93/94,564

(1980). 2"W. R. Wampler, J. Nue!. Mater. 122/123,1598 (1984). lOW. Miiller, B. M. U. Scherzer, and J. Bohdansky, IPP···JET Report No. 26

(1986) . . lOR. A. Causey and W. L. Hsu (to be published). liB. L. Doyle, W. R. Wampler, and D. K. Brice, J. NueL Mater. 103/104,

513 (1981). 3'K. Niwasi, M. Sugimoto, T. Tanabe, and F. E. Fujita, J. Nucl. Mater.

155-157, 303 (1988). "K. Ashida, K. Kanamori, K. Ichimura, M. Matsuyama, and K. Watan­

abe, J. Nue!. Mater. 137, 288 (19ll6).

J, 'lac. Sci. Tee/mol. A, '101.8, No.3, May/Jun 1990

'4K. Sonnenberg (JET Joint Undertaking) (privaie communication). "Handbook a/Chemistry and Physics, edited by Robert C. Weast (Chemi­

cal Rubber, Cleveland, 1974) p. B-n. 'OR. A. Causey, M. r. Baskes, and K. L. Wilson, J. Vae. Sci. Techno!. A 4,

1189 (1986). 37W .. L Usn and R. A. Causey, J. Vae. Sci. Techno\., A 5, 2768 (1987). "Handbook afChemistry alld Physics, 67th ed. (Chemical Rubber, Cleve-

land, 1987). wJ. T. Hurd and R. O. Adams, J. Vae. Sci. TechnoL 6, 229 (1969). "'T. G. Nieh, J. Wadsworth, A. Joshi, Scri. Metal!. 20,87 (1986). 4'B. Mills, R. A. Causey, and C. D. Croessmann, Sandia Naiional Labora­

torics (unpublished data). 4'M. F. Smith, R. D. Watson, J. B. Whitley, and J. M. McDonald, Fusion

Techno!. 8,1174 (1985). 4'J. n. Whitley, K. L. Wilson, and D. A. Buchcnauer, 1. Nuc!. Mater. 155-

157,82 (1988). 44Sandia National Laboratories, Report No. SAND89-0901. 4sR. D. Watson and J. B. Whitley, Nucl. Eng. Des./Fusion 4, 49 (1986 l. 4('A. M. Hassanein et at., Nuc. Eng. Des./Fusion 1, 307 ( 1984). 47c. D. Crocssmann et al., J. Nue\. Mater. 128/129,816 (1984). "Final Report, DOE Project No. ER D-83-339, Oak Ridge National Labo­

ratory, Oak Ridge, TN, July, 1986 . . I9J. 13. Whitley, J. A. Koski, and R. Aymar, "Engineering Considerations

for the Tore Supra Pump Limiter," in Proceedings ofthe 14th Symposium on Fusion Technology, Avignon, France, 1986 (unpublished).

5()W. G. Wolfer and T. J. McCarville, Fusion Techno!. 8, 1157 (1985). s 'c. E. Ells and E. C. W. Perryman, J. N lle!. Mater. 1. 73 (1959). '"Lawrence Livermore National Laboratory, 1982, Report No. UCID-

19327. 53J. M. Beeston, M. R. Martin, C. R. Brinkman, G. E. Korth, and W. C.

Francis, in Symp. on Materials Performance in Operating Nuclear Sys­tems, CONF-73081, Ames Laboratory, Ames, lA, Aug. 1978. Nue!. Me­tall. 19, 59.

S"Y. Mishima, S. Ishina, and S. Shiazawa, "The Electron Microscopic Ob­servation of Beryllium and Its Alloys Irradiated at Elevated Tempera­tures," Paper 25 ill Beryllium 1977.

"K. L. Wilscm, G. J. Thmnas, and W. Bauer, ANS Trans. 27, 272 (1977). S0R. A. Causey, 1. G. Miller. and G. R. Longhurst (to be published). s·!J. Hackmann and J. Uhlenbusch, J. Nuel. Mater. 128/129, 41 (1984 l. '"I'. K. Mioduszcwski et al., Nne!. Fusion 26, 1171 (1986). 59 Joint JEl:ISX-B Beryllium Limiter Experiment, Final Report, Oak

Ridge National Laboratory, July 1986. 60K. L Dietz, M. A. Pick. A. T. Peacock. K. Sonnenberg, J. Ehrenberg, G.

Saibene, and R. Sartori, in 13th Symposium of Fusioll Engineering, Kllox­ville, Tennessee, October 1989 (to be published).

'''M. Bcssenrodt-Weberpals, J. Hackmann, C. Nicswand, and J. UIlIen­busch, J. Nud. Mater. 162·-163, 435 (1989).

62J. Roth, J. Nue!. Mater. 145-147, 87 (1987). (':'M. Hugon, 1'. 1'. Lallia, and P. H. Rebut, Joint European Torus Report,

JET-R(89) 14. 64G. L. Kulcinski, J. Nucl. Mater., 1221123,29 (1984). bSM. I. Baskes, A. E. Pontau, K. L. Wilson, and W. L. Barr, J. Nue!. Mater.

122/123, 1511 (1984). h('STARFIRE: A Commercial Tokamak Fusion Power Plant Study, ANLI

PPP-80-1, 19110, Vol. J, p. 2. "7K. R. Schultz et aI., Fusion Eng. Des. 9, 15 ( 1989). ""K. R. Schultz et al., "TIBER II First Wall and Divertor Engineering,"

GA·CI9065. Jan. 1988. ""ITER Concept Definition, IAEA, Vienna, 1989, IAEA/ITER/DS/3.

Redistribution subject to AVS license or copyright; see http://scitation.aip.org/termsconditions. Download to IP: 130.88.90.110 On: Fri, 19 Dec 2014 05:26:34