- Email: [email protected]
Optimization of the availability of the core CXRS diagnostics for ITER
Fusion Engineering and Design 86 (2011) 1174–1177
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Optimization of the availability of the core CXRS diagnostics for ITER F. Klinkhamer a,∗ , A. Krimmer b , W. Biel b , N. Hawkes c , G. Kiss b , J. Koning d , Yu. Krasikov b , O. Neubauer b , B. Snijders a a
TNO Science & Industry, partner in ITER-NL, P.O. Box 155, 2600 AD Delft, The Netherlands Institut für Energieforschung - Plasmaphysik, Forschungszentrum Jülich Gmbh, Association EURATOM-FZJ, member of Trilateral Euregio Cluster, 52425 Jülich, Germany Culham Centre for Fusion Energy. Culham Science Centre, OX14 3DB, Abingdon, UK d FOM-Institute for Plasma Physics Rijnhuizen, Association EURATOM-FOM, partner in the Trilateral Euregio Cluster and ITER-NL, P.O. Box 1207, 3430 BE Nieuwegein, The Netherlands b c
a r t i c l e
i n f o
Article history: Available online 31 May 2011 Keywords: Core CXRS Mirror lifetime Mirror cleaning Port plug
a b s t r a c t New optical configurations for the ITER core CXRS system offer the possibility of longer ducts between the first mirror and the plasma. This has led to a renewed optimization of the availability, using a simple model of the degradation of the first mirror that starts with the conditions of (a) the required measurement performance and (b) the geometry of the port plug. It is found that for a fully passive system the design should strive for the longest duct length possible. Given known data, this will result in a diagnostic lifetime still substantially shorter than ITER lifetime. When an option of cleaning the first mirror is introduced (assuming this is a feasible option) the optimum is less straightforward, because the lifetime of the second mirror then also becomes important. The optimum then depends on the ratio between the cleaning interval and the ITER lifetime. Options are presented for various sets of assumptions. Finally practical limitations of supporting subsystems (cleaning system, shutter, calibration system) may influence the final design. Examples of such limitations with their impact are presented. © 2011 Published by Elsevier B.V.
1. Introduction
2. First order mirror degradation model
The core CXRS system in ITER collects radiation from the diagnostic neutral beam (DNB) and uses this for the determination of various plasma parameters via spectroscopy. It consists of a periscope housed in upper port plug 3, a fiber bundle and a set of spectrometers. Previous work has led to a design of the core CXRS port plug in which the duct length between first mirror and plasma was relatively short [1]. This short duct combined with the large aperture towards the plasma probably leads to an unacceptably short lifetime of the first mirror. The relatively short duct was caused by the decision that the optical design should allow for a retractable tube for first mirror maintenance. However, a separate study has revealed that another optical configuration with a longer duct is possible within the envelope given by the port plug, if the retractable tube facility is not included [2]. This information combined with the introduction of a simple first mirror degradation model gives rise to a set of new criteria for the design of the core CXRS port plug.
When developing a first order model of the degradation of the first mirror, a set of assumptions has to be made. In this study they are as follows:
∗ Corresponding author. Tel.: +31 15 2692416. E-mail address: [email protected] (F. Klinkhamer). 0920-3796/$ – see front matter © 2011 Published by Elsevier B.V. doi:10.1016/j.fusengdes.2011.01.064
• A degradation related transmission loss of 15% is estimated to be acceptable. In the first part of the study this full budget is given to the reflection loss of the first mirror • First mirror degradation is deposition dominated (supported by calculations in [3], which concludes no net erosion for duct length/duct diameter > 4). • First mirror deposition is proportional to the solid angle from first mirror to the plasma (Assumption used among others in [4].) • Beryllium deposition is assumed to be the critical contaminant. Carbon deposition will also occur, but cleaning of carbon is less of a challenge than cleaning of beryllium. • Deposition only takes place during a plasma pulse. Possible postpulse deposition and deposition during baking can be minimized by shutter use and proper heating profiles.
Using these assumptions a model can be developed that includes first-order approximations for e.g. wall sputtering and erosion. For each of the contributions quantitative estimates of the effect can
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
M1 Degradation 15.0 % Used value for carbon 20%loss @ layer 20nm, derived from [7]
From table 1 of [5]. In [6] same order (including safety factor 10) Duct length 204 mm
Reflec. loss coeff. 1.0 %/nm TBC
Solid angle 0.096 sr
M1 Be flux 2.6e+17 at./m²/s #### Transm. 1.75 %
Part of the burn time in which the shutter is open
M1 deposition flux 1.3e+17 at./m²/s
M1 net Be flux 9.2e+16 at./m²/s
M1 deposition rate 6.7e-04 nm/s
Sticking coeff 0.5 Estimate, value of 1 would be conservative
Geom. transm. 1.52 % Wall sputtering 115.0 %
Aperture 72 mm
From table 1 of [5].
M1 life (no clean) 2.3E+04 s 56 ITER burns 0.2 % ITER life
M1 allowed dep. 15.0 nm
FW Be flux. 1.5e+19 at./m²/s
1175
For Be, from study of effect on baffling in [5]
Shutter transm. 100.0 %
FW at. flux. 2.6e+20 at./m²/s
M1 H+D flux 3.9e+18 at./m²/s #### Transm. 1.52 %
M1 erosion flux 3.9e+16 at./m²/s Sputter yield 1.0 % TBC
From [5], conservative since it is yield for Mo and Rh for 1keV D particles
Fig. 1. Visual representation of the mirror degradation model. The quantities that appear in the boxes are taken from published designs [5–7].
be combined. A graphical representation of such a model is given in Fig. 1. In the opinion of the authors this model is incomplete. However, studies that aim at improved prediction are generally very time-consuming and the results of these studies may be too late to influence the design of the diagnostics, given scheduling targets. This model on the other hand may be used to give coarse predictions and modify the design when needed. 3. Application to port plug design When the model is applied to the design reported earlier in [1], it is found that the mirror lifetime is 56 ITER pulses, equivalent to 0.2% of ITER lifetime or slightly more than one day of ITER operation. When less conservative estimates are used, this is increased to about 20 days of operation. The model also indicates in which areas improvement could be obtained: the most effective route will be to increase the duct length for the first mirror to the aperture in the plasma, combined with a minimization of the aperture diameter. Note that in this analysis the extension of mirror lifetime by sacrificing observation time (closing the shutter during part of a pulse) is not included; this option may be considered as a last resort. 3.1. Maximizing duct length In a parallel study described in [2], alternative optical configurations were developed that would increase the duct length to the maximum possible. Two alternatives were identified, one in which the aperture still resides in the blanket that is mounted on the port plug itself and the other where the aperture is moved to the blanket below the port plug. These options are called the large M1 option and the 11C blanket option respectively. The duct length achievable for the large M1 option can be up to 600 mm, while that for the 11C blanket option is 900 mm. 3.2. Minimizing aperture diameter The other parameter is the aperture diameter. This has been re-assessed using the requirement on the transmission × etendue
(effective etendue) per spatial channel and removing the margin in etendue (about a factor 2.5) that was applied in the past. For the first three mirrors the requirement for the effective etendue is 3.125e−7 m2 sr for one (radial) spatial channel of a/30, where a is the minor radius of the ITER tokamak. (The requirement for the full system up to the detector is an effective etendue of 5e−8 m2 sr [8]. The estimated transmission of the optical system after the first three mirrors is 16%. Since etendue is preserved in the optics behind the first three mirror, this leads to 5e−8 m2 sr/0.16 = 3.125e−7 m2 sr for the first three mirrors). In previous studies single crystal molybdenum has been chosen for the first mirror because of its good resistance to erosion. Here, however, with a long duct, erosion is not a significant threat to the mirror and we have therefore selected a rhodium coated substrate for the first mirror, benefiting from its higher reflectivity. (The second and third mirrors remain rhodium coated, as in previous designs.) This choice leads to a total transmission of the first three mirrors of 36% including degradation (using an estimated 75% reflection per mirror at 460 nm [9], and 15% transmission loss due to deposition). So the etendue offered by the system has to be 3.125e−7 m2 sr/0.36 = 8.7e−7 m2 sr for one spatial channel. Using margins for separation of the spatial channels in the radial direction of the field and using the 1/e diameter of the DNB in the toroidal direction the maximum size of the available field of view is 40 mm × 400 mm for one spatial channel. This means the solid angle of the aperture as seen from the object (the DNB) has to be 8.7e−7 m2 sr/(0.04 m × 0.4 m) = 5.4e−5 sr. The solid angle ˝ created by an aperture with radius r at a distance R can be approximated by ˝ = r2 /R2 . This gives – for a distance between DNB and aperture of 4.5 m – an aperture radius of slightly less than 19 mm. 3.3. Lifetime for alternative designs Using the value for the duct length and the aperture diameter as given in the previous sections, and applying the model of Section 2 gives the following lifetimes for the first mirror. Since the design lifetime of ITER shots is 30,000 shots, it follows that there is no solution available that guarantees first mirror sur-
1176
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
Table 1 First mirror lifetimes for new options, according to first order model. Option
Conservative assessment
Nominal assessment
Large M1 11C blanket
1720 shots (36 op. days) 3860 shots (81 op. days)
27,500 shots (570 op. days) 61,700 shots (1280 op. days)
vival over this full lifetime, given a conservative assessment. For a nominal assessment (using Be particle fluxes as derived in [6] without applying the recommended safety factor of 10) the 11C blanket option does give sufficient lifetime including a good margin. In this approach the large M1 option also would survive almost the full ITER lifetime. 4. Introduction to mirror cleaning Given the large uncertainties reported in particle flux modeling [6], we, as well as other groups, have devised several options to further mitigate mirror degradation [10,11]. The search for mitigation schemes is based on the assumption that Be deposition is the dominant degradation mechanism. Combining the results published in [10] with knowledge gathered in experiments on removal of deposition on extreme UV wafer scanner systems [12], the preferred mechanism to remove the Be deposition is physical plasma sputtering. Research and prototyping on this cleaning mechanism has been started. An important factor in the feasibility of the cleaning with physical sputtering is whether or not the magnetic field of ITER is on while the cleaning is in progress. The preferred scheme is that the magnetic field would be off during the cleaning process (doubts exist whether the cleaning process would be at all possible in the presence of the magnetic field). A decision as to whether the magnetic field will be off during each or some of the regular short term maintenance periods (each fortnight, as defined in [13]) is pending. If it is assumed that cleaning is indeed feasible, then Table 1 makes clear that such a cleaning should take place each third short term maintenance period for the case of the conservative estimate of the large M1 option (given 11 operational days between each short term maintenance). If this or one of the variants (less conservative lifetime combined with less frequent cleaning) is possible then the degradation of the first mirror is no longer the limiting factor on the availability of the core CXRS system. What may become an issue is the degradation of the next optical element.
Normally the fluxes towards the second mirror are an order of magnitude lower than the flux towards the first mirror. Therefore little attention is given to this degradation. However, as explained above, it is proposed to maintain remove the depositions on the first mirror by cleaning. In this case the degradation of the second mirror may become an issue. There are three effects which may cause a material flux to the second mirror: (a) Be and/or C atoms impinging on the first mirror do not stick, but are reflected diffusely and thus may reach, and be deposited on, the second mirror, (b) high-energy H and D atoms erode material from the first mirror sputter, part of this eroded material can also be deposited on the second mirror, (c) when cleaning the first mirror, part of the material sputtered off this mirror may be deposited on the second mirror. For the deposition on the second mirror a first order analytical model is developed. Its result is the following relation for the lifetime of the second mirror: 1 · F · f (ϑ) · G
L − L 2 0 1 L1
6. Addition of supporting systems The core CXRS port plug consist of more than an optical system alone. Several systems have to be included to allow the proper performance of the whole diagnostic system. These systems must also be housed in the port plug and may have impact on the optical configuration. These systems are: • A shutter, for protection of the optics between shots and possibly during part of the shots. • A calibration system, required to get the correct scientific results. • The cleaning system. • Possibly a retractable tube, allowing more frequent maintenance of components with limited lifetime. A separate publication is dedicated to a study of the issues related to integrating these systems in the core CXRS port plug [14].
5. Lifetime of second mirror
NM2 =
ticle flux impinging on the first mirror, Note that with successful cleaning this factor will be 1. f(): angular distribution of material reflected/sputtered of the first mirror, normalized such that integral over a hemisphere is 1. Used below is a Lambertian distribution with f() = cos()/. G: correction factor introduced because calculating the size of the first mirror based on FOV and distances gives a too small area. This factor can be in the order of 2–5. L0 : the distance from the aperture to the second mirror [m]. For the core CXRS this is normally about 1 m, the 11C blanket option is a special case, since the folding mirror in front of the first curved mirror is removed. L1 : the distance from the aperture to the first mirror [m]. Note that L0 –L1 represents the distance between the first and the second mirror. dDNB p0 : the distance from the DNB to the aperture (p0 because it is the zeroth instance of the pupil) [m]. FOVr : the linear field of view of the core CXRS port plug towards the DNB in radial direction (determines size of first mirror) [m]. FOVt : the linear field of view of the core CXRS port plug towards the DNB in toroidal direction (determines size of first mirror) [m]. NM1 : lifetime of first mirror without cleaning [number of ITER shots]. Thus it is found that the lifetime of the second mirror is proportional with the square of the distance between the first and the second mirror (L0 –L1 ). Using the relation it can also be estimated that the lifetime of the second mirror is about 23,000 ITER shots for the large mirror option, when the full 15% reflection loss is given to second mirror. This result will be used as an input for a trade-off between second mirror lifetime and required cleaning period for the first mirror.
2
·
dDNB p0 · NM1 FOVR · FOVt
(1)
with NM2 : lifetime of second mirror [number of ITER shots]. F: ratio of the particle flux emitted by the first mirror and the par-
7. Discussion and conclusions An approach to maximize the availability of the optical system of the core CXRS port plug is presented for the case that the lifetime of the first mirror is dominated by loss of reflection due to deposition of CX atoms. Calculations in [3] support the assumption that this mechanism is dominant over other degradation mechanisms. More than one concept for the core CXRS port plug has been presented, and within the concepts variations are still possible. In principle a decision can only be made when (a) it is known in more detail what the boundary conditions of ITER are and (b) better modeling of the mirror degradation is available. It is not foreseen that either will be available in the near future, so choices will have to be made based on the present state of knowledge. One of the critical decisions is the choice between the large M1 option and the 11C blanket option. However, the introduction of an aperture in blankets other than the blanket shield module of the diagnostic port plug may be ruled out by engineering constraints on the first-wall design [15].
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
The cleaning system can probably only be operated during a shut-down of the superconducting toroidal field system. The scheduling of such shutdowns is presently under discussion. In order to reach the best availability for the system the first mirror has to be located in a certain distance from the second mirror where the reduced deposition on the second mirror and the frequency of the cleaning actions for the first mirror are balanced. The full availability is predicted to be reached with a relatively high frequency of cleaning actions (around once each month). However, if the magnetic field is shut down less often, there is no backup left to reach 100% availability. In this scenario the present analysis shows that it is best to maximize the duct length of the first mirror to the plasma and then decrease the observation time with the shutter. The reduction of observation time to be chosen should then be based on the observed rate degradation of the reflectivity. Thus the design will offer the highest possible availability, but physics will determine what the availability actually is. Acknowledgments Optimization of a design can only be done when all information is available. Our thanks go to Dan Thomas, Robin Barnsley, Chris
1177
Walker, Spencer Pitcher, George Vayakis, Didier van Houtte of ITER, Glen Counsell, Christian Ingesson of F4E for their discussions, allowing to slowly grasp the complexity of the ITER system. Furthermore thanks to Andrei Litnovsky for his comment on the concept proposals. And of course I have to thank Manfred von Hellemann for introducing me in the world of CXRS diagnostics. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]
S. Sadakov, et al., Fusion Sci. Technol. 56 (2009) 134–138. F. Klinkhamer, et al., SOFT Conference 2010, 2010, pp. 3–151. V. Kotov, et al., SOFT Conference 2010, 2010, pp. 2–117. V. Konovalov, et al., Fusion Eng. Des. 56–57 (2001) 923–927. J. Brooks, J. Allain, Nucl. Fusion 48 (2008) 045003. V. Kotov, et al., J. Nucl. Mater. 390–391 (2009) 528. IAEA2006/ITERP1. M. von Hellermann, Review of Beam Aided Diagnostics for ITER, Paper, 21st IAEA Fusion Energy Conference. L. Marot., ITPA 12 Meeting, Princeton March 28, 2007. E. Muhkin et al., Progress in Development of Deposition Prevention and Cleaning Techniques of In-vessel Optics in ITER, IT/P6-24. A. Litnovsky, et al., Nucl. Fusion 47 (2007) 833. N. Koster, EUVL Symposium 2008, Lake Tahoe, 2008. D. van Houtte, ITER Machine Operation Availability, IDM 34GC2G. A. Krimmer, et al., SOFT Conference 2010, 2010, pp. 2–113. C.I. Walker, S. Pitcher, Private communications.
Contents lists available at ScienceDirect
Fusion Engineering and Design journal homepage: www.elsevier.com/locate/fusengdes
Optimization of the availability of the core CXRS diagnostics for ITER F. Klinkhamer a,∗ , A. Krimmer b , W. Biel b , N. Hawkes c , G. Kiss b , J. Koning d , Yu. Krasikov b , O. Neubauer b , B. Snijders a a
TNO Science & Industry, partner in ITER-NL, P.O. Box 155, 2600 AD Delft, The Netherlands Institut für Energieforschung - Plasmaphysik, Forschungszentrum Jülich Gmbh, Association EURATOM-FZJ, member of Trilateral Euregio Cluster, 52425 Jülich, Germany Culham Centre for Fusion Energy. Culham Science Centre, OX14 3DB, Abingdon, UK d FOM-Institute for Plasma Physics Rijnhuizen, Association EURATOM-FOM, partner in the Trilateral Euregio Cluster and ITER-NL, P.O. Box 1207, 3430 BE Nieuwegein, The Netherlands b c
a r t i c l e
i n f o
Article history: Available online 31 May 2011 Keywords: Core CXRS Mirror lifetime Mirror cleaning Port plug
a b s t r a c t New optical configurations for the ITER core CXRS system offer the possibility of longer ducts between the first mirror and the plasma. This has led to a renewed optimization of the availability, using a simple model of the degradation of the first mirror that starts with the conditions of (a) the required measurement performance and (b) the geometry of the port plug. It is found that for a fully passive system the design should strive for the longest duct length possible. Given known data, this will result in a diagnostic lifetime still substantially shorter than ITER lifetime. When an option of cleaning the first mirror is introduced (assuming this is a feasible option) the optimum is less straightforward, because the lifetime of the second mirror then also becomes important. The optimum then depends on the ratio between the cleaning interval and the ITER lifetime. Options are presented for various sets of assumptions. Finally practical limitations of supporting subsystems (cleaning system, shutter, calibration system) may influence the final design. Examples of such limitations with their impact are presented. © 2011 Published by Elsevier B.V.
1. Introduction
2. First order mirror degradation model
The core CXRS system in ITER collects radiation from the diagnostic neutral beam (DNB) and uses this for the determination of various plasma parameters via spectroscopy. It consists of a periscope housed in upper port plug 3, a fiber bundle and a set of spectrometers. Previous work has led to a design of the core CXRS port plug in which the duct length between first mirror and plasma was relatively short [1]. This short duct combined with the large aperture towards the plasma probably leads to an unacceptably short lifetime of the first mirror. The relatively short duct was caused by the decision that the optical design should allow for a retractable tube for first mirror maintenance. However, a separate study has revealed that another optical configuration with a longer duct is possible within the envelope given by the port plug, if the retractable tube facility is not included [2]. This information combined with the introduction of a simple first mirror degradation model gives rise to a set of new criteria for the design of the core CXRS port plug.
When developing a first order model of the degradation of the first mirror, a set of assumptions has to be made. In this study they are as follows:
∗ Corresponding author. Tel.: +31 15 2692416. E-mail address: [email protected] (F. Klinkhamer). 0920-3796/$ – see front matter © 2011 Published by Elsevier B.V. doi:10.1016/j.fusengdes.2011.01.064
• A degradation related transmission loss of 15% is estimated to be acceptable. In the first part of the study this full budget is given to the reflection loss of the first mirror • First mirror degradation is deposition dominated (supported by calculations in [3], which concludes no net erosion for duct length/duct diameter > 4). • First mirror deposition is proportional to the solid angle from first mirror to the plasma (Assumption used among others in [4].) • Beryllium deposition is assumed to be the critical contaminant. Carbon deposition will also occur, but cleaning of carbon is less of a challenge than cleaning of beryllium. • Deposition only takes place during a plasma pulse. Possible postpulse deposition and deposition during baking can be minimized by shutter use and proper heating profiles.
Using these assumptions a model can be developed that includes first-order approximations for e.g. wall sputtering and erosion. For each of the contributions quantitative estimates of the effect can
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
M1 Degradation 15.0 % Used value for carbon 20%loss @ layer 20nm, derived from [7]
From table 1 of [5]. In [6] same order (including safety factor 10) Duct length 204 mm
Reflec. loss coeff. 1.0 %/nm TBC
Solid angle 0.096 sr
M1 Be flux 2.6e+17 at./m²/s #### Transm. 1.75 %
Part of the burn time in which the shutter is open
M1 deposition flux 1.3e+17 at./m²/s
M1 net Be flux 9.2e+16 at./m²/s
M1 deposition rate 6.7e-04 nm/s
Sticking coeff 0.5 Estimate, value of 1 would be conservative
Geom. transm. 1.52 % Wall sputtering 115.0 %
Aperture 72 mm
From table 1 of [5].
M1 life (no clean) 2.3E+04 s 56 ITER burns 0.2 % ITER life
M1 allowed dep. 15.0 nm
FW Be flux. 1.5e+19 at./m²/s
1175
For Be, from study of effect on baffling in [5]
Shutter transm. 100.0 %
FW at. flux. 2.6e+20 at./m²/s
M1 H+D flux 3.9e+18 at./m²/s #### Transm. 1.52 %
M1 erosion flux 3.9e+16 at./m²/s Sputter yield 1.0 % TBC
From [5], conservative since it is yield for Mo and Rh for 1keV D particles
Fig. 1. Visual representation of the mirror degradation model. The quantities that appear in the boxes are taken from published designs [5–7].
be combined. A graphical representation of such a model is given in Fig. 1. In the opinion of the authors this model is incomplete. However, studies that aim at improved prediction are generally very time-consuming and the results of these studies may be too late to influence the design of the diagnostics, given scheduling targets. This model on the other hand may be used to give coarse predictions and modify the design when needed. 3. Application to port plug design When the model is applied to the design reported earlier in [1], it is found that the mirror lifetime is 56 ITER pulses, equivalent to 0.2% of ITER lifetime or slightly more than one day of ITER operation. When less conservative estimates are used, this is increased to about 20 days of operation. The model also indicates in which areas improvement could be obtained: the most effective route will be to increase the duct length for the first mirror to the aperture in the plasma, combined with a minimization of the aperture diameter. Note that in this analysis the extension of mirror lifetime by sacrificing observation time (closing the shutter during part of a pulse) is not included; this option may be considered as a last resort. 3.1. Maximizing duct length In a parallel study described in [2], alternative optical configurations were developed that would increase the duct length to the maximum possible. Two alternatives were identified, one in which the aperture still resides in the blanket that is mounted on the port plug itself and the other where the aperture is moved to the blanket below the port plug. These options are called the large M1 option and the 11C blanket option respectively. The duct length achievable for the large M1 option can be up to 600 mm, while that for the 11C blanket option is 900 mm. 3.2. Minimizing aperture diameter The other parameter is the aperture diameter. This has been re-assessed using the requirement on the transmission × etendue
(effective etendue) per spatial channel and removing the margin in etendue (about a factor 2.5) that was applied in the past. For the first three mirrors the requirement for the effective etendue is 3.125e−7 m2 sr for one (radial) spatial channel of a/30, where a is the minor radius of the ITER tokamak. (The requirement for the full system up to the detector is an effective etendue of 5e−8 m2 sr [8]. The estimated transmission of the optical system after the first three mirrors is 16%. Since etendue is preserved in the optics behind the first three mirror, this leads to 5e−8 m2 sr/0.16 = 3.125e−7 m2 sr for the first three mirrors). In previous studies single crystal molybdenum has been chosen for the first mirror because of its good resistance to erosion. Here, however, with a long duct, erosion is not a significant threat to the mirror and we have therefore selected a rhodium coated substrate for the first mirror, benefiting from its higher reflectivity. (The second and third mirrors remain rhodium coated, as in previous designs.) This choice leads to a total transmission of the first three mirrors of 36% including degradation (using an estimated 75% reflection per mirror at 460 nm [9], and 15% transmission loss due to deposition). So the etendue offered by the system has to be 3.125e−7 m2 sr/0.36 = 8.7e−7 m2 sr for one spatial channel. Using margins for separation of the spatial channels in the radial direction of the field and using the 1/e diameter of the DNB in the toroidal direction the maximum size of the available field of view is 40 mm × 400 mm for one spatial channel. This means the solid angle of the aperture as seen from the object (the DNB) has to be 8.7e−7 m2 sr/(0.04 m × 0.4 m) = 5.4e−5 sr. The solid angle ˝ created by an aperture with radius r at a distance R can be approximated by ˝ = r2 /R2 . This gives – for a distance between DNB and aperture of 4.5 m – an aperture radius of slightly less than 19 mm. 3.3. Lifetime for alternative designs Using the value for the duct length and the aperture diameter as given in the previous sections, and applying the model of Section 2 gives the following lifetimes for the first mirror. Since the design lifetime of ITER shots is 30,000 shots, it follows that there is no solution available that guarantees first mirror sur-
1176
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
Table 1 First mirror lifetimes for new options, according to first order model. Option
Conservative assessment
Nominal assessment
Large M1 11C blanket
1720 shots (36 op. days) 3860 shots (81 op. days)
27,500 shots (570 op. days) 61,700 shots (1280 op. days)
vival over this full lifetime, given a conservative assessment. For a nominal assessment (using Be particle fluxes as derived in [6] without applying the recommended safety factor of 10) the 11C blanket option does give sufficient lifetime including a good margin. In this approach the large M1 option also would survive almost the full ITER lifetime. 4. Introduction to mirror cleaning Given the large uncertainties reported in particle flux modeling [6], we, as well as other groups, have devised several options to further mitigate mirror degradation [10,11]. The search for mitigation schemes is based on the assumption that Be deposition is the dominant degradation mechanism. Combining the results published in [10] with knowledge gathered in experiments on removal of deposition on extreme UV wafer scanner systems [12], the preferred mechanism to remove the Be deposition is physical plasma sputtering. Research and prototyping on this cleaning mechanism has been started. An important factor in the feasibility of the cleaning with physical sputtering is whether or not the magnetic field of ITER is on while the cleaning is in progress. The preferred scheme is that the magnetic field would be off during the cleaning process (doubts exist whether the cleaning process would be at all possible in the presence of the magnetic field). A decision as to whether the magnetic field will be off during each or some of the regular short term maintenance periods (each fortnight, as defined in [13]) is pending. If it is assumed that cleaning is indeed feasible, then Table 1 makes clear that such a cleaning should take place each third short term maintenance period for the case of the conservative estimate of the large M1 option (given 11 operational days between each short term maintenance). If this or one of the variants (less conservative lifetime combined with less frequent cleaning) is possible then the degradation of the first mirror is no longer the limiting factor on the availability of the core CXRS system. What may become an issue is the degradation of the next optical element.
Normally the fluxes towards the second mirror are an order of magnitude lower than the flux towards the first mirror. Therefore little attention is given to this degradation. However, as explained above, it is proposed to maintain remove the depositions on the first mirror by cleaning. In this case the degradation of the second mirror may become an issue. There are three effects which may cause a material flux to the second mirror: (a) Be and/or C atoms impinging on the first mirror do not stick, but are reflected diffusely and thus may reach, and be deposited on, the second mirror, (b) high-energy H and D atoms erode material from the first mirror sputter, part of this eroded material can also be deposited on the second mirror, (c) when cleaning the first mirror, part of the material sputtered off this mirror may be deposited on the second mirror. For the deposition on the second mirror a first order analytical model is developed. Its result is the following relation for the lifetime of the second mirror: 1 · F · f (ϑ) · G
L − L 2 0 1 L1
6. Addition of supporting systems The core CXRS port plug consist of more than an optical system alone. Several systems have to be included to allow the proper performance of the whole diagnostic system. These systems must also be housed in the port plug and may have impact on the optical configuration. These systems are: • A shutter, for protection of the optics between shots and possibly during part of the shots. • A calibration system, required to get the correct scientific results. • The cleaning system. • Possibly a retractable tube, allowing more frequent maintenance of components with limited lifetime. A separate publication is dedicated to a study of the issues related to integrating these systems in the core CXRS port plug [14].
5. Lifetime of second mirror
NM2 =
ticle flux impinging on the first mirror, Note that with successful cleaning this factor will be 1. f(): angular distribution of material reflected/sputtered of the first mirror, normalized such that integral over a hemisphere is 1. Used below is a Lambertian distribution with f() = cos()/. G: correction factor introduced because calculating the size of the first mirror based on FOV and distances gives a too small area. This factor can be in the order of 2–5. L0 : the distance from the aperture to the second mirror [m]. For the core CXRS this is normally about 1 m, the 11C blanket option is a special case, since the folding mirror in front of the first curved mirror is removed. L1 : the distance from the aperture to the first mirror [m]. Note that L0 –L1 represents the distance between the first and the second mirror. dDNB p0 : the distance from the DNB to the aperture (p0 because it is the zeroth instance of the pupil) [m]. FOVr : the linear field of view of the core CXRS port plug towards the DNB in radial direction (determines size of first mirror) [m]. FOVt : the linear field of view of the core CXRS port plug towards the DNB in toroidal direction (determines size of first mirror) [m]. NM1 : lifetime of first mirror without cleaning [number of ITER shots]. Thus it is found that the lifetime of the second mirror is proportional with the square of the distance between the first and the second mirror (L0 –L1 ). Using the relation it can also be estimated that the lifetime of the second mirror is about 23,000 ITER shots for the large mirror option, when the full 15% reflection loss is given to second mirror. This result will be used as an input for a trade-off between second mirror lifetime and required cleaning period for the first mirror.
2
·
dDNB p0 · NM1 FOVR · FOVt
(1)
with NM2 : lifetime of second mirror [number of ITER shots]. F: ratio of the particle flux emitted by the first mirror and the par-
7. Discussion and conclusions An approach to maximize the availability of the optical system of the core CXRS port plug is presented for the case that the lifetime of the first mirror is dominated by loss of reflection due to deposition of CX atoms. Calculations in [3] support the assumption that this mechanism is dominant over other degradation mechanisms. More than one concept for the core CXRS port plug has been presented, and within the concepts variations are still possible. In principle a decision can only be made when (a) it is known in more detail what the boundary conditions of ITER are and (b) better modeling of the mirror degradation is available. It is not foreseen that either will be available in the near future, so choices will have to be made based on the present state of knowledge. One of the critical decisions is the choice between the large M1 option and the 11C blanket option. However, the introduction of an aperture in blankets other than the blanket shield module of the diagnostic port plug may be ruled out by engineering constraints on the first-wall design [15].
F. Klinkhamer et al. / Fusion Engineering and Design 86 (2011) 1174–1177
The cleaning system can probably only be operated during a shut-down of the superconducting toroidal field system. The scheduling of such shutdowns is presently under discussion. In order to reach the best availability for the system the first mirror has to be located in a certain distance from the second mirror where the reduced deposition on the second mirror and the frequency of the cleaning actions for the first mirror are balanced. The full availability is predicted to be reached with a relatively high frequency of cleaning actions (around once each month). However, if the magnetic field is shut down less often, there is no backup left to reach 100% availability. In this scenario the present analysis shows that it is best to maximize the duct length of the first mirror to the plasma and then decrease the observation time with the shutter. The reduction of observation time to be chosen should then be based on the observed rate degradation of the reflectivity. Thus the design will offer the highest possible availability, but physics will determine what the availability actually is. Acknowledgments Optimization of a design can only be done when all information is available. Our thanks go to Dan Thomas, Robin Barnsley, Chris
1177
Walker, Spencer Pitcher, George Vayakis, Didier van Houtte of ITER, Glen Counsell, Christian Ingesson of F4E for their discussions, allowing to slowly grasp the complexity of the ITER system. Furthermore thanks to Andrei Litnovsky for his comment on the concept proposals. And of course I have to thank Manfred von Hellemann for introducing me in the world of CXRS diagnostics. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15]
S. Sadakov, et al., Fusion Sci. Technol. 56 (2009) 134–138. F. Klinkhamer, et al., SOFT Conference 2010, 2010, pp. 3–151. V. Kotov, et al., SOFT Conference 2010, 2010, pp. 2–117. V. Konovalov, et al., Fusion Eng. Des. 56–57 (2001) 923–927. J. Brooks, J. Allain, Nucl. Fusion 48 (2008) 045003. V. Kotov, et al., J. Nucl. Mater. 390–391 (2009) 528. IAEA2006/ITERP1. M. von Hellermann, Review of Beam Aided Diagnostics for ITER, Paper, 21st IAEA Fusion Energy Conference. L. Marot., ITPA 12 Meeting, Princeton March 28, 2007. E. Muhkin et al., Progress in Development of Deposition Prevention and Cleaning Techniques of In-vessel Optics in ITER, IT/P6-24. A. Litnovsky, et al., Nucl. Fusion 47 (2007) 833. N. Koster, EUVL Symposium 2008, Lake Tahoe, 2008. D. van Houtte, ITER Machine Operation Availability, IDM 34GC2G. A. Krimmer, et al., SOFT Conference 2010, 2010, pp. 2–113. C.I. Walker, S. Pitcher, Private communications.