Laser photodetachment neutraliser for negative ion beams

Fusion Engineering and Design 85 (2010) 745–751

Contents lists available at ScienceDirect

Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes

Laser photodetachment neutraliser for negative ion beams M. Kovari ∗ , B. Crowley EURATOM/CCFE Fusion Association, Culham Science Centre, Abingdon, Oxon, OX14 3DB, UK

a r t i c l e

i n f o

Article history: Received 3 December 2009 Received in revised form 20 April 2010 Accepted 21 April 2010 Available online 18 May 2010 Keywords: DEMO Neutral beam Neutral injection Negative ion Neutralisation Laser Photodetachment Fusion

a b s t r a c t We outline a speculative design for a photodetachment neutraliser for a negative ion neutral beam system, with neutralisation efficiency of 95% or more. The practical difficulties are enormous. The ion beam must pass through an optical cavity capable of reflecting the light many times. For 500 reflections, the laser optical power output ∼800 kW, giving circulating power ∼400 MW. All sources of light loss combined need to be kept to 0.2% or less per pass. The losses due to photodetachment itself, and due to Thomson scattering in the beam plasma are negligible. A key task is to maintain the reflectance of the mirrors above 99.97% for long periods of operation, protecting all the components from thermal and neutron damage, and from caesium, sputtered atoms and other contamination. A diode-pumped Nd-doped YAG laser can have overall electrical-to-light (“wall-plug”) efficiency up to 25%. A DEMO concept reactor such as the EU Power Plant Conceptual Study (PPCS) Model B requires 270 MW heating power. If this is all provided by neutral beams, then a laser neutraliser might reduce the electrical power consumption for this from 900 MW to 520 MW. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved.

1. Introduction: neutralisation of negative ion beams A major consideration for ITER, DEMO and beyond is the energy efficiency of the heating and current drive systems. Present-day and planned systems have overall wall-plug efficiency in the range 20–30% [1]. In contrast, conceptual studies for power plants assume 60–70%. The biggest source of energy loss in the ITER neutral beam system is the low neutralisation efficiency. The only method currently used to neutralise high power negative ion beams is the simple gas cell. This gives a neutralisation efficiency of 58% for 1 MeV ions [2], and even less for the higher energies that may be required for DEMO. A gas neutraliser also releases a copious flow of gas from each end, requiring enormous cryopumps and imposing high stripping losses in the accelerator. Even for ITER it has not been proven that stripping losses will be acceptable. If a plasma neutraliser is used instead, neutralisation efficiency can reach 80% for 30% ionisation, and the gas required is much less [3,4]. A multi-cusp magnetic trap with microwave ECR heating at the plasma periphery has been proposed as an effective system for production of cold plasma with high ionisation in a large volume. Experimental results suggest that the plasma parameters necessary for ITER can be obtained with a superconducting magnetic system providing the maximum field ∼1 T. The required microwave

∗ Corresponding author. Tel.: +44 0 1235 46 4912. E-mail address: [email protected] (M. Kovari).

power input into the plasma is about 0.5 MW. It is claimed that the problems of beam deflection and divergence can be successfully eliminated. The microwave sources would be gyrotrons, which have demonstrated wall-plug efficiencies ∼45–50%, including waveguide losses [5]. Grisham [2] has proposed a supersonic lithium vapour jet perpendicular to the direction of beam propagation. The maximum neutralisation efficiency in lithium vapour has been measured as 65% for 400 keV H− (equivalent to 800 keV D− ). A photodetachment neutraliser would consist of a laser and an optical cavity through which the negative ion beam would pass. The following reaction takes place, for the example of deuterium, h + D− → D + e. The cross-section for hydrogen is given in [6,7]. We assume that the cross-section is the same for deuterium. 2. Photodetachment neutralisers—review Fink first proposed photodetachment in 1975 [8], and derived the basic parameters of such a neutraliser [9]. He concluded that the damage limit of the mirrors would control the minimum length of the neutraliser. In [10] Fink considered photodetachment in the presence of a background gas. He found that a gas target can improve the net neutral fraction, but only if the photodetachment fraction (the neutral fraction with no gas) is below about 80%. For example, for 60% photodetachment, gas can improve the neutral fraction to at most 75% for 1 MeV D− . This combined system requires only half the power of a pure photoneutraliser, but about

0920-3796/$ – see front matter. Crown Copyright © 2010 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.fusengdes.2010.04.055

746

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751

Fig. 1. The geometry considered.

70% as much gas as a gas neutraliser. The benefits of a combined gas and optical neutraliser are therefore modest. In 1983, Fink proposed the supersonic chemical oxygen–iodine laser (COIL) [11], but by 1987 had moved on to the more conventional Nd:YAG (neodymium-doped yttrium aluminium garnet). The laser power required is proportional to the width of the beam, which could be very small if a slit ion source is used. Sources with an array of circular apertures fit the laser profile less efficiently. The LBL self-extracting surface conversion source provided 4 A/m of accelerated ions from a single slit 3 cm wide [12]. Vanek [13] described cavity and laser technology. They proposed a mirror with three layers of cooling channels built-in. They listed mirror materials in order of suitability: (1) (most suitable) silicon (which they rejected because of failures in the development of single crystal heat exchangers); (2) silicon carbide; (3 and 4) tungsten and tungsten carbide; (5) molybdenum; (6) (least suitable) copper. Since a window was required, they proposed one made from two plates of sapphire with organic coolant in between. As well as absorption, scattering and surface reflection, there would be thermal lensing in the coolant. To minimise this they suggested a folded optical path in the vacuum space, reducing the number of times the light passes through the window. Even so it is unlikely that any window can be built with >99% transmission as they proposed. 3. Wavelength, power and efficiency Fig. 1 shows the geometry considered. The laser power required is: hc ln(1 − f ) Power =  



2eVB w . Mmp G

(1)

The symbols have the meanings given in Table 1, together with the values used in this paper except where stated.

Fig. 2. Laser power required as a function of wavelength, for different beam energies.

Eq. (1) can be derived by noting that rate of neutralisations per unit volume = ni np c = −

dni , dt

where ni and np are the ion and photon densities, respectively, the relative velocity is c, and the derivative is taken along the path of an ion. The photon density is assumed to be constant, so attenuation along the photon path is neglected. The neutralised fraction f is f =1−

ni (final) , ni (initial)

and the time taken by the ion to traverse a neutraliser of length L is



L

.

2eVB /Mmp

The gain G is given by G = 1/ε, where ε = fraction of energy lost per pass. Fig. 2 shows the laser power required as a function of wavelength, for different beam energies. Fig. 3 shows the overall efficiency of a neutral beam system (power injected into plasma/electrical power required). The system is based on the ITER beamline with MAMuG (MultiAperture–Multi-Grid) accelerator, but with some anticipated improvements, with parameters in Table 2 [14,15].

Table 1 Symbols and their values.  f  VB M G

w h c mp e

Wavelength Degree of neutralisation (fraction of ions neutralised) Cross-section for photodetachment Acceleration voltage Relative atomic mass Number of times the light passes through the cavity (loosely called “gain”) Width of neutraliser Planck’s constant Speed of light Mass of proton Charge of the electron

1064 nm 0.95 3.375 × 10−21 m2 at 1064 nm 1 MV 2 (deuterium) 500

0.25 m Fig. 3. Energy efficiency of the neutral beam system. — laser neutraliser; efficiency of beamline with gas neutraliser; max theoretical efficiency given perfect neutralisation with no power consumption by the laser.

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751

747

Table 2 Assumed parameters of a beamline system. Gas neutraliser Ion source filling pressure (Pa) Accelerator losses, including particles leaving accelerator (MW)a Accelerated negative ion power (MW) HV power to accelerator (MW) Electrical power to the accelerator (MW) Neutralisation efficiency for the D2 target Halo fraction - lost in beamline Geometrical transmission for non-halo fraction Re-ionisation loss (%) Injected power (MW) Electrical power to the laser (MW) Electrical power to the ion source, correction and compensation coils, water cooling, residual ion dump & cryoplant (MW)b Total electrical power (MW) Overall efficiency

0.2 9.9

Laser neutraliser 0.2 6.6

40

40

49.9 57.0

46.6 53.3

58%

95%

10% 95%

10% 95%

5% 18.8 0 6.0

2% 31.8 3.2 4.4

Fig. 5. The sensitivity of the enhancement ratio to changes in length of the cavity.

63.0 30%

60.9 52%

Enhancement ratio =

a The losses in the accelerator have been assumed to be proportional to the gas pressure [16]. b The losses in the accelerator and residual ion dump are reduced due to the lower gas pressure in the laser case. For this reason we have assumed that the cryopumps are the same size in both cases.

The biggest losses are in the accelerator, due to the acceleration of extracted electrons, stripping of partially accelerated ions, and acceleration of stripped electrons and secondary electrons. 4. The optical cavity We propose a single cavity concept, in which the laser gain medium is inside the same cavity as the particle beam. The laser does not have an output coupling mirror as is conventional. This ensures that the intensity in the neutralisation region is the same as in the laser medium. There is, however, a concept in which the intensity in the neutralisation region is actually higher than in the laser medium. This is discussed in the next section. 4.1. Enhancement cavity Chaibi [17] has proposed the use of an enhancement cavity (Fabry–Perot cavity, Fig. 4). This is used for the resonant enhancement of optical power or intensity: if the incident light is resonant with the cavity, the intracavity power can be far above the incident power [18,19]. The benefit would be that the circulating power in the laser material could be much less than that in the neutraliser. This reduces losses and thermal lensing in the laser. Note this scheme has an advantage only when losses due to the laser medium dominate.

Fig. 4. Enhancement cavity.

The performance of an enhancement cavity with a perfectly reflecting end mirror is given in [19] (Eq. (4)). If there are no losses in the enhancement cavity, this reduces to 1−R  , √ 1 + R + 2 R cos 4 L 

(2)

where Enhancement ratio = intensity in enhancement cavity/intensity in laser cavity; R = intensity reflection coefficient of the coupling mirror; L = length of enhancement cavity. The length of the laser cavity enters only implicitly through the wavelength. Note that this equation is the same as the one that would apply to a single cavity with radiation entering from free space. This is because the gain in the laser means the bandwidth of the laser cavity is zero (except for quantum effects). Fig. 5 shows the sensitivity of the enhancement ratio to nanometre changes in length of the cavity (for fixed wavelength). Fig. 6 shows the sensitivity to femtometre wavelength changes. The bandwidth is given by





ı ≈ ln(R) ·

c , 2L

(3)

giving a value ∼1 MHz, equivalent to 4 fm. By using special stabilizing and feedback techniques, bandwidths as low as 0.1 Hz can be realized in specialised laboratories. Each mode in the laser (axial and transverse) must be exactly matched to a mode in the neutraliser cavity. Of course the coupling mirror presents the same problems with cooling and wavefront distortion as a window, as described in Section 2. Furthermore the coupling mirror will have reflections from both faces—creating a third parasitic resonator. This would normally be solved by angling one face, but this could not be permitted here because the unwanted reflection would then be lost. Because of the extreme sensitivity of the enhancement cavity to variations in length, wavelength and refractive index, and the

Fig. 6. The sensitivity of the enhancement ratio to changes in wavelength.

748

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751 Table 3 Speculative design targets for loss per pass.

Fig. 7. An example geometry. The laser medium is hatched.

difficulty of cooling the coupling mirror, we will not discuss this option further.

Component

Loss

End mirror Laser end mirror Diffraction loss Misalignment loss Beam plasma Laser medium

0.04% 0.04% 0.04% 0.04% Negligible 0.04%

Total Gain

0.2% 500

5. Cavity concept design

5.2. Losses

5.1. Optical design, diffraction losses, misalignment sensitivity

Table 3 illustrates the design targets that would be needed for the cavity concept. Loss of light in the beam plasma is negligible—this is the only target in Table 3 that is almost certain to be achieved. We have estimated this loss in two ways—first by calculating the Thomson scattering from electrons, and second by using a standard opacity database. The total (incoherent) Thomson scattering cross-section of an electron is

The use of curved mirrors leads to lower diffraction losses than do plane mirrors, and they also have less stringent alignment tolerances. It is essential, however, to ensure that the beam does not form a waist inside the neutralisation volume, or near any optical component. The best way to ensure this is to place a virtual focal point (waist) outside the cavity, as illustrated in Fig. 7. A possible geometry is as follows. The formulae used in this section are taken from [20]. Mirror 1 radius R1 = −30 m convex Mirror 2 radius R2 = 30 m concave Cavity length L = 10 m The position of the beam waist relative to mirror 1 is (1 − g )g

L01 = L 

1 2 , g1 + g2 − 2g1 g2 

(4)

where g1 = 1 −

L R1

g2 = 1 −

L , R2

(5)

and the condition for a stable resonator is 0 < g1 g2 < 1. In this example g1 g2 = 0.889. Reflection from the laser surface should be taken into account. In Fig. 7 we have drawn the laser medium with equal radius of curvature on both sides. This will ensure that the parameters for the cavity formed by the rear surface are approximately the same as those for the cavity formed by reflection from the front surface. The laser medium itself forms a sub-cavity which is marginally stable (g1 g2 = 1 almost exactly). The radii of the fundamental (“Gaussian” or TEM00 ) mode on each mirror, w1 and w2 , are given by w1 2 =

L 



g2 g1 (1 − g1 g2 )

w2 2 =

L 



g1 . g2 (1 − g1 g2 )

(6)

w1 = 2.68 mm, w2 = 3.79 mm. The actual beam size is of course much larger, indicating that many transverse modes are excited—about 2000 for a beam 50 mm wide. Calculating the diffraction losses in multimode operation is not trivial. Only those modes whose radii are close to the aperture radius exhibit noticeable diffraction losses. As the mirror aperture is increased the loss of the highest order mode will decrease until the next mode starts oscillating and the loss increases again. With increasing aperture radius the loss will approach zero. Detailed analysis is required—we have not established whether diffraction loss is acceptable. The sensitivity to misalignment of the mirrors depends on how close the cavity is to the edge of the stability criterion g1 g2 < 1. In this example the maximum misalignment ∼20 ␮rad for 0.03% loss per pass. The mirrors would certainly need active steering.

T = (8/3)re2 = 6.652 × 10−29 m2 . For electron density ne , we use the value derived in [21] for the ITER neutraliser–maximum 4 × 1014 m−3 . Of course this is an overestimate, as the ITER neutraliser has a much larger gas density. We take beam height, H = 1.5 m. The total (incoherent) Thomson optical depth is  T ne H = 4 × 10−14 , which is negligible. Coherent Thomson scattering is always less than incoherent. In order to take into account a wider range of species and processes in the beam plasma, we use the TOPS on-line opacity database [22]. This does not include molecular species, and assumes local thermodynamic equilibrium. Neither of these conditions apply, but we can show that even pessimistic assumptions give negligible opacity by assuming a hot gas/plasma, and including impurities, as detailed in Table 4 Nd:YAG and Yb:YAG (ytterbium-doped YAG) have a loss coefficient at 1064 nm of 0.003 cm−1 [23]. A loss of 0.04% per pass requires a laser length ≤ 1 mm. Thin-disc lasers are much thinner than this—typically 100–200 ␮m—so loss in the laser material is in principle acceptable. The loss power due to photodetachment itself, for a 40 A ion beam, using 1064 nm photons (1.2 eV), is 48 W. For an 800 kW laser, the fractional loss is 6 × 10−5 . Since the light traverses the cavity ∼500 times, this corresponds to a fractional loss of only 10−7 per pass. 5.3. Mirrors A key task is to maintain the reflectance of the mirrors above 99.96% for long periods of operation, protecting all the components from caesium from the negative ion source, sputtered atoms from beam interaction with acceleration grids and the neutraliser body, and other contamination. Mirrors are commercially available with Table 4 Assumptions and results for TOPS calculation. Number density (D2 ) Temperature Corresponding pressure at 300 K Iron Copper Results Opacity (absorption + scattering) Optical depth for path length 1.5 m

2.4 × 1018 m−3 5 eV 0.01 Pa 0.1% by number 0.1% by number 6.5 m2 /kg 1 × 10−7

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751

very high reflectivity (>99.99% at 1064 nm). Other mirrors are available with very high damage thresholds—1010 W/m2 at 1064 nm in continuous wave (CW) operation [24]. It is not clear if these two specifications could be combined in one mirror. The fraction of light diffusely reflected from a surface, S, at normal incidence, is given [25] by S = R0

 4 2 

(7)

where R0 = reflectance in the absence of roughness;  = root mean square roughness;  = wavelength. The applicability of this formula to multi-layer mirrors is not clear, but using it as a guide, we find that for  = 1064 nm and S = 0.0005,  should be 2 nm. While this is well within the state of the art for mirror manufacture, it will be essential to maintain this smoothness for long periods, despite radiation damage and possible erosion from or deposition onto the surface. Orlovskiy and Vukolov [26] exposed multi-layer dielectric mirrors to heat and neutrons. The reflectance remained ≥ 99% after irradiation to 1023 n/m2 , but they did not measure to greater precision than this. After heating to 250 ◦ C in vacuum some mirrors retained their reflectance and others were destroyed. Of five types of dielectric mirrors irradiated to 1.1 × 1023 n/m2 (with temperature cycling between 60 ◦ C and 270 ◦ C), three showed flaking or blistering of the coating and two showed little mechanical damage [27]. Reflectivity changes were not stated. Experience with particle beam systems using copper components shows that other surfaces often become coated with a visible layer of copper, while windows become virtually opaque. Since sputtered atoms travel in straight lines, it is possible to reduce but perhaps not eliminate surface contamination by recessing the windows or providing baffles. Under experimental conditions a gold layer 5 nm thick reduces the external transmission of silica at 1064 nm from ∼93% to about 60% [28]. The effect of a sputtered layer of stainless steel was smaller: a 70 nm layer reduced transmission to 88% [29]. For mirrors we can speculate that to maintain 99.97% reflectance one would need to ensure that any metallic impurity layer is no more than a fraction of a nanometre thick. The synergistic effects of high power laser beams, radiation damage and deposition of thin impurity layers may be important. When silica is implanted with 5 × 1020 H+ ions/m2 , the external transmission at 1064 nm drops to about 80% [30]. Since the ions are implanted in the surface, there will be a similar effect in a mirror. For comparison, the total number of accelerated ions for a 40 A beam (such as in ITER) is 8 × 1027 per year of continuous operation. 5.4. Radiation dose Radiation damage will be a major issue for many components. While detailed neutron fluxes are not available for DEMO, we can make a rough estimate by scaling values calculated for the original large ITER design [31] in proportion to the average neutron load at the first wall predicted for Model B of the PPCS [32]. The fast (>0.1 MeV) neutron flux just inside the wall of the neutral beam vacuum vessel in the neutraliser region is then about 6 × 1020 neutrons/m2 per full power year. By using mirrors and shielding it would be possible to place the laser itself at a lower flux if required. 6. The laser Inoue et al. [33] proposed arrays of semiconductor lasers. These have area ∼0.05 × 0.05 mm, and large divergences of 10–35◦ . We cannot see any way in which such lasers could be incorporated into a single optical cavity as proposed in this paper. The same problem applies to fibre lasers, which also have large divergence.

749

Table 5 Summary of laser parameters. Total laser output power Type Pump diode-laser output Electricity consumption Number of lasers Laser dimensions Output power per unit area Pump power absorbed per unit area Pump power absorbed per unit volume Power density on each mirror Power density (including light travelling in both directions) Heat dissipation in laser material Cost of diodes @ $10/W

800 kW Nd:YAG or Yb:YAG, thin-disc 1600 kW 3.2 MW 100 5 cm × 5 cm × 0.2 mm 300 W/cm2 600 W/cm2 30 kW/cm3 80 kW/cm2 160 kW/cm2 300 W/cm2 8 kW per laser $16 million

The oxygen–iodine chemical laser has been proposed [11], but we have not yet been able find the wall-plug efficiency in the literature. A caesium vapour laser pumped with a cw laser-diode array has been demonstrated [34], with output power 10 W at 895 nm, a suitable wavelength. The overall optical-to-optical efficiency was 62%. If the diode pumps are 50% efficient, the overall wall-plug efficiency would be ∼30%. Both these lasers of course require windows to contain the vapour. Thin-disc solid-state lasers can provide high output power, high efficiency and good beam quality [35]. The disc can be edgepumped using arrays of laser-diodes [36]. More than 5 kW CW has been obtained from a single disc, with optical efficiency of 65%. The electrical efficiency of the total laser system is more than 25%. Scaling laws suggest that more than 40 kW is possible from a single disc. The disc is cooled through one of its large flat faces, which is also used as the end mirror of the laser cavity. This minimises temperature variations, so thermal lensing is much less than in a comparable rod-laser. Even so, finite element analysis [35] shows that thermal effects will create a lens of focal length ∼1 m, which must be corrected. Since the cavity is several metres long, and lenses are highly undesirable, the only surfaces that can be used for correction are the faces of the laser itself, which must, at a minimum, be machined to the correct shape. Both Nd:YAG and Yb:YAG have been used for thin-disc lasers, and both have suitable wavelengths (1064 nm for Nd and 1030 or 1050 nm for Yb). Ytterbium has the advantage of a smaller quantum defect (9.6% compared to 32%), which gives high power efficiency in principle and low thermal effects. It has the disadvantage that it must be pumped at high power densities ∼10 kW/cm3 . (The lower laser level is only 0.076 eV above the ground state, so an appreciable population in that level occurs in thermal equilibrium. An atom in this level can absorb laser light, so the level must be emptied by very strong pumping.) Table 5 gives a set of order-of-magnitude parameters for the laser system as a whole and for each laser module, assuming 50% optical efficiency and 100% absorption of pump light. It turns out that optical non-linearity in the laser is insignificant. The non-linear refractive index n2 for YAG is 3 × 10−13 esu = 7 × 10−20 m2 /W [37,38]. With the irradiance (power density or optical intensity) from Table 5, the change in refractive index is 1 × 10−10 , which is negligible. 7. Pulsed operation The DEMO design target is for steady state continuous operation of the reactor, and, as the beams have to drive the reactor current continuously, pulsed operation is not an option. Thus pulsed laser

750

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751

Fig. 9. Folded cavity.

spaced circular apertures (see Fig. 10, which also shows the ITER layout for comparison). It is in any case desirable to structure the beam into several “beam groups”, as this allows radiation shielding to be placed inside the beamline [33]. Fig. 11 illustrates a possible layout in a fusion reactor. The lasers themselves are located in a shielded region far from the reactor, using a single folding mirror for each laser. The folding mirrors allow the lasers to be separated from each other as much as is required for pump diodes and cooling systems. Only one beam column and its neutraliser are shown. Several beam columns would be required. The lasers are inside the primary tritium barrier of the fusion reactor. 9. Discussion Fig. 8. Light at 45◦ to particle beam (top), and normal to it (bottom).

operation, which would save laser energy, is not an option for a working system. 8. Layout At first sight it seems attractive to angle the laser beam relative to the ion beam, as this increases the path length. This does reduce the power required. However, the space occupied by the laser beam along the axis of the neutraliser is then greater than the length of the laser material. Fig. 8 shows an example of two layouts producing the √ same degree of neutralisation. The laser at 45◦ has 2 times the path length. To compensate, the laser normal to the ion beam produces √ √ 2 times the power and therefore needs to have 2 times the area. Nevertheless, the neutraliser with angled laser is much longer. (The additional length is not altogether wasted, as it provides space for gas pumping.) This layout is not an efficient use of space. Fig. 9 shows a folded cavity, as proposed in [13]. This can reduce losses, if scattering and absorption in the laser dominate. If there are Nf folding mirrors, and the light traverses the entire cavity Nc times, the “gain” G = Nc (Nf + 1). The loss in the laser medium is only proportional to Nc . The ratio of the gain to this loss is therefore improved by a factor (Nf + 1). However, the power density in the laser is likely to be a limiting factor. If this is the case, the laser area will be a fixed quantity. For given laser area the folded design is (Nf /2) + 1 times as long (in the direction of the particle beam) as the unfolded design. This is probably unacceptable. To keep laser power to a minimum, and allow the use of small modular optics and lasers, the accelerator should produce narrow distinct beam columns, using slit apertures, or arrays of closely

The most powerful CW laser of which we are aware is the Boeing Airborne Laser, which is described as a “megawatt class” chemical oxygen–iodine laser. The most powerful CW solid-state laser seems to be a Northrop Grumman weapon prototype which has achieved a power level of 100 kW for 5 min, with electro-optical efficiency of 19%. In contrast, the laser discussed here is eight times as powerful, and must run 24 h a day. In addition, it must be reflected several hundred times before being attenuated or lost, and operate in a radiation environment. On the other hand, it does not need to be portable.

Fig. 10. Beamlet layouts. Not to scale.

M. Kovari, B. Crowley / Fusion Engineering and Design 85 (2010) 745–751

Fig. 11. Schematic of a neutraliser unit for one slit beam.

The potential efficiency gains in the fusion reactor’s Neutral Beam system, and the consequent large decrease in the necessary recirculating power in the reactor’s on-site infrastructure, would appear to justify an experimental R&D programme. This might consist of a 100 W thin-disc laser and optical cavity. Using folding mirrors the optical cavity could have a realistic length. It would be possible to test components after irradiation, and after coating with thin impurity layers; to optimise the doping level; and to evaluate thermal lensing effects. Acknowledgments We would like to thank G. Naylor, M. Walsh, E. Surrey, D. Ciric, T.T.C. Jones, D. Stork, D. Ward and E. Hodgson for valuable discussions and comments. This work was funded by the United Kingdom Engineering and Physical Sciences Research Council and the European Communities under the contract of Association between EURATOM and CCFE. The views and opinions expressed herein do not necessarily reflect those of the European Commission. This work was carried out within the framework of the European Fusion Development Agreement. References [1] J. Pamela, A. Bécoulet, D. Borba, J.-L. Boutard, L. Horton, D. Maisonnier, Efficiency and availability driven R&D issues for DEMO, Fusion Eng. Des. 84 (2009) 194–204. [2] L. Grisham, Lithium jet neutralizer to improve negative ion neutral beam performance, in: E. Surrey, A. Simonin (Eds.), Negative Ions Beams and Sources:1st International Symposium, AIP Conference Proceedings 1097, 2009, pp. 364–373. [3] M. Hanada, M. Kashiwagi, T. Inoue, K. Watanabe, T. Imai, Experimental comparison between plasma and gas neutralization of high-energy negative ion beams, Rev. Sci. Instrum. 75 (2004) 1813. [4] ITER Design Description Document 5.3 (2001) https://user.iter.org/?uid= 22H4Q8&action=get document.

751

[5] K. Sakamoto, A. Kasugai, K. Kajiwara, K. Takahashi, Y. Oda, K. Hayashi, et al., Progress of high power 170 GHz gyrotron in JAEA, Nucl. Fusion 49 (2009) 095019. [6] C.F. Barnett, J.A. Ray, E. Ricci, M.I. Wilker, E.W. McDaniel, E.W. Thomas, et al., Atomic Data for Controlled Fusion Research Volume I, ORNL-5206 (1977). [7] H.P. Saha, Phys. Rev. A 38 (1988), 4546. [8] J.H. Fink, A.M. Frank, Photodetachment of electrons from negative ions in a 200 keV deuterium beam source, Lawrence Livermore Natl. Lab. (1975), UCRL16844. [9] J.H. Fink, Photodetachment now, in: 12th Symposium on Fusion Engineering, Monterey, CA, IEEE, New York, 1987. [10] J.H. Fink, Photodetachment of negative ion beams in the presence of a background gas, Fusion Technol. 11 (1987) 416–419. [11] J.H. Fink, Photodetachment technology, in: Production and Neutralization of Negative Ions and Beams: 3rd Int. Symposium, Brookhaven 1983, AIP, New York, 1984, pp. 547–560. [12] K.N. Leung, K.W. Ehlers, Self-extraction negative ion source, Rev. Sci. Instrum. 53 (1982) 803–809. [13] V. Vanek, Technology of a laser resonator for the photodetachment neutralizer, in: Production and Neutralization of Negative Ions and Beams: 3rd Int. Symposium, Brookhaven 1983, AIP, New York, 1984, pp. 568–584. [14] R. Hemsworth, H. Decamps, J. Graceffa, B. Schunke, M. Tanaka, M. Dremel, et al., Status of the ITER heating neutral beam system, Nucl. Fusion 49 (2009) 045006. [15] R. Hemsworth, ITER Organisation, private communication. [16] A. Krylov, R.S. Hemsworth, Gas flow and related losses in the ITER neutral beam injector, Fusion Eng. Des. 81 (2006) 2239–2248 (The mean pressure was derived from Fig. 3). [17] W. Chaibi, C. Blondel, L. Cabaret, C. Delsart, C. Drag, A. Simonin, Photoneutralization of negative ion beam for future fusion reactor, in: E. Surrey, A. Simonin (Eds.), Negative Ions Beams and Sources:1st International Symposium, AIP Conference Proceedings 1097, 2009, pp. 385–394. [18] http://www.rp-photonics.com/enhancement cavities.html. [19] Y.C. See, S. Guha, J. Falk, Coupled optical resonators for the enhancement of laser intracavity power, Appl. Opt. 20 (1981) 1211–1217. [20] N. Hodgson, H. Weber, Laser Resonators and Beam Propagation: Fundamentals, in: Advanced Concepts and Applications, Springer, 2005, We have calculated the misalignment sensitivity from eq. 5.88, together with eq. 5.90 (which diverges for the plane-plane cavity) or Fig. 5.51.(which does not). [21] E. Surrey, Gas heating in the neutralizer of the ITER neutral beam injection systems, Nucl. Fusion 46 (2006) S360–S368. [22] http://www.t4.lanl.gov. See also N.H. Magee, Jr., J. Abdallah, Jr., R.E.H. Clark, J.S. Cohen, L.A. Collins, G. Csanak, Atomic Structure Calculations and New Los Alamos Astrophysical Opacities, Astronomical Society of the Pacific Conference Series (Astrophysical Applications of Powerful New Databases, S.J. Adelman, W.L. Wiese (eds.)) 78 (1995) 51. [23] http://www.redoptronics.com/Nd-YAG-crystal.html. [24] www.cvilaser.com. [25] H. Haihong, Measurements of light scattering from glass substrates by total integrated scattering, Appl. Opt. 44 (2005) 6163–6166. [26] I.I. Orlovskiy, K.Yu. Vukolov, Thermal and neutron tests of multilayered dielectric mirrors, Fusion Eng. Des. 74 (2005) 865–869. [27] E.H. Farnum, F.W. Clinard Jr., S.P. Regan, B. Schunke, Neutron damage to diagnostic mirrors, J. Nucl. Mater. 219 (1995) 224–229. ˜ E.R. Hodgson, Surface degradation effects on laser damage [28] P. Martin, A. Morono, in KU1 quartz glass windows for LIDAR applications, J. Nucl. Mater. 307–311 (2002) 1260–1263. ˜ E.R. Hodgson, Laser induced damage enhancement due [29] P. Martin, A. Morono, to stainless steel deposition on KS-4 V and KU1 quartz glasses, J. Nucl. Mater. 329–333 (2004) 1442–1445. ˜ E.R. Hodgson, Optical and electrical degradation of [30] S.M. Gonzàlez, A. Morono, H+ implanted KS-4V quartz glass, Fusion Eng. Des. 74 (2005) 831–834. [31] T. Inoue, Neutral beam heating & current drive system, Design Description Document, appendix H, Nucl. Anal. (1998). [32] D. Maisonnier, D. Campbell, I. Cook, L. Di Pace, L. Giancarli, J. Hayward, et al., Power plant conceptual studies in Europe, Nucl. Fusion 47 (2007) 1524–1532. [33] T. Inoue, M. Hanada, M. Kashiwagi, S. Nishio, K. Sakamoto, M. Sato, et al., Design study of a neutral beam injector for fusion DEMO plant at JAERI, Fusion Eng. Des. 81 (2006) 1291–1297. [34] B. Zhdanov, R.J. Knize, Diode-pumped 10 W continuous wave cesium laser, Opt. Lett. 32 (2007) 2167–2169. [35] A. Giesen, J. Speiser, Fifteen years of work on thin-disk lasers: results and scaling laws, IEEE J. Selected Top. Quantum Electron. 13 (2007) 598–609. [36] J. Vetrovec, R. Shah, T. Endo, A. Koumvakalis, K. Masters, W. Wooster, et al., Progress in the development of solid-state disk laser, Proc. SPIE 5332 (2004) 235. [37] VLOC, subsidiary of II-VI Incorporated, http://www.vloc.com/PDFs/ YAGBrochure.pdf. [38] Encyclopedia of Laser Physics and Technology, http://www.rp-photonics.com/ nonlinear index.html.