- Email: [email protected]
Sensitive helium leak detection in a deuterium atmosphere using a high-resolution quadrupole mass spectrometer
Vacuum/volume
Pergamon PII: SOO42-207X(96)00063-2
47lnumbers 6-8lpages 767 to 769/1996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All riahts reserved 0042-207X196 $15.00+.00
Sensitive helium leak detection in a deuterium atmosphere using a high-resolution quadrupole mass spectrometer S Hiroki, T Abe and Y Murakami, Japan Atomic Energy Research Institute, Naka Fusion Research Establishment, Naka-machi, Naka-gun, lbaraki 31 l-07, Japan
In fusion machines, realizing a high-purity plasma is a key to improving the plasma parameters. Thus, leak detection is a necessary part of reducing the leak rate to a tolerable level. However, a conventional helium PHel leak detector is useless in fusion machines with a deuterium (0,) plasma, because retained D particles on the first wails release D2 for a long period and the released D2 interferes with the signals from the leaked 4He due to the near identical masses of 4.0026 u tdHe) and 4.0282 u (D2). A high-resolution quadrupole mass spectrometer (HR-QMS) that we have recently developed, can detect a 4He+ population as small as lop4 peak in a DZ atmosphere. Thus, the HR-QMS has been applied to detect 4He leaks. To improve the minimum detectable limit of 4He leak, a differentially pumped HR-OMS analyzer was attached to a chamber of the 4He leak detector. In conclusion, the improved 4He leak detector could detect 4He leaks of the order of lO_” Pa - m3/s in a D2 atmosphere. Copyright 0 1996 Elsevier Science ltd. Key words: Helium leak, deuterium
atmosphere,
high-resolution,
Introduction Current fusion machines such as JT-60 (JAERI Tokamak-60) and future machines like ITER (International Thermonuclear Experimental Reactor) under design as an international collaborative project have large and complicated vacuum vessels. There are hundreds of welds and flange joints on the vessel, and each area should be frequently checked for leaks. Because air leaks into a main plasma lead to a radiative power loss, the realization of an impurity free plasma is essential in improving the plasma parameters. Thus, leak detection is an integral part of maintaining impurity free plasma. The leaked 4He is generally detected with a conventional 4He leak detector that adopts a sector magnet type mass spectrometer. This 4He leak detector cannot, however, detect a small 4He leak in fusion machines using a D, plasma due to its poor resolving power for ‘He (4.0026 u) and Dz (4.0282 u). Since D particles are more easily retained on the first walls than are 4He particles,‘,’ the released Dz background interferes with the detection of the ‘He signals. To reduce the D? pressure to lower than the ‘He pressure in the 4He leak detector, a selective pumping of D, using a Zr-Al getter was proposed,3 although this non-throughput type pump has a limited capacity. The use of argon (Ar) as a tracer gas with a quadrupole mass spectrometer (QMS) was also proposed. However, the minimum detectable leak rate of Ar is far larger than that of ‘He, because of the higher Ar abundance in air and
quadrupole
mass spectrometer.
its larger Ar molecular diameter. Thus, the direct 4He detection in a high D2 background with high sensitivity is a better solution. Recently, we have developed a high-resolution QMS (HRQMS) that adopts a higher stability zone in the Mathieu stability diagram. This HR-QMS can detect at a population of lop3 in a D2 atm0sphere.j In addition, the fringing field characteristics of the HR-QMS were clarified.’ Based on the results, we have applied the HR-QMS to a 4He leak detector for use in a high Dz background. This paper describes initial experimental results on the improved 4He leak detector, where a differentially pumped analyzer was used to lower the minimum detectable limit of the 4He leak rate.
Experimental arrangement In this experiment, we used two analysers with and without the differential pump unit to clarify the effect of the pump unit. The analysers with and without the pump units were named analyzer1 and -2, respectively. In this section, an experimental arrangement using only analyzer-l is described and the results of using analyzer-2 will be discussed in the next section. The experimental setup, where analyzer-l is attached to an L-N2 trapped chamber is shown in Figure 1. A thin plate with a small orifice of 2.0 mm in diameter separated the ionizer spatially from the quadrupole. The orifice helped to ensure an efficient pumping of the space 767
S Hi&i
et al: Sensitive
helium
leak detection
(a
-
(b) TMP
I set
x104 b+
RP Figure 1. Experimental setup. PI and P,: Calibrated Bayard-Alpert type gauges, 4He-SL: “He standard leak of 4.1 x lo-” Pa+m3/s, SEM: Secondarv electron multiulier, L-N?: Liquid nitrogen. TMP: Magnet suspended turbomolecular pump, R~I Rotary pump: 1 Background
containing the quadrupole and the SEMs6 A 17-stage Cu-Be type SEM was used. An equivolume mixture of 4He/D, was introduced into the chamber to adjust the resolving power of the HR-QMS. A 4He standard leak (4He-SL) of 4.1 x lo-” Pa *m’/s was used to calibrate this 4He leak detector. Two ion gauges (P, and PJ were calibrated in advance by nitrogen with a spinning rotor gauge. The dimensions of the quadrupole in the two analysers were the same as previously used’ and the frequency of the RF voltage applied to the quadrupole was 4.9 MHz in both analyzers. Results and discussion Initially, analyzer-2 was attached to the chamber and its performance was compared with the results obtained later using analyzer-l. Analyzer-2 is a conventional analyzer with an open space between the ionizer and the quadrupole. The equivolume mixture of 4He/D, was introduced into the chamber and the HRQMS was tuned to separate the 4Hef peak from the DC peak resulting in a half-height resolution of about 200. The 4He and D, gases were then, independently effused into the chamber while gradually reducing the rate of the 4He effusion. The 4He’ and D: peaks are shown in Figure 2(a) and (b) at a 4He+/D:ratio of roughly lop4 in linear scales. The small 4He+ peak height in Figure 2(b) corresponds to a lo-* Pa . m’/s order of the 4He leak rate, and the DC peak height corresponds to a 1O-4 Pa . m’js order for the D, effusion rate. The spread of the D, peak width in Figure 2(b) results from a larger time constant of its amplifier range. Besides, the background current of 1.8 x lo-“. A in Figure 2(b) lifts the separated 4He+/D,+peaks, so this background current limits the minimum detectable 4He leak rate. The increase in the background current may be caused by ion feedback phenomena in the SEM due to a higher pressure in the analyzer. To restrict the pressure rise in the SEM volume, a normal ion effusion slit in the ionizer was replaced by a thin plate with the small orifice. The orifice was attached to analyzer-l as shown in Figure 1 and the effective pump speed of its differential pump unit was 0.012 m’js. The SEM was recommended to operate below 5 x 10-j Pa to protect against a breakdown in the SEM volume. When the 4He/Dz gas mixture was slowly introduced 768
current
I set Figure 2. Time series of 4HeC/D: peaks observed using analyzer-2 at (a) IO-’ A/div and (b) lo-” A/div. P, = 3.0 x 10-j Pa. In Figures 24, the applied voltage to the SEM was -2kV.
into the chamber, P, indicated 0.3 Pa whilst P, registered 5 x IO-’ Pa, thus the pressure ratio of P, to P, was 60. Hence the analyzable upper pressure limit can be expanded to 0.3 Pa. A filament of the ionizer will emit electrons in air-based gas mixtures when the HR-QMS is applied to the 4He leak detector. Frequent breaks of tungsten filaments in a higher H,O pressure can be overcome using a more suitable filament such as rhenium.’ In addition, analyzer-l can be directly attached between the turbomolecular pump and the fore pump because of the high maximum analyzable pressure of 0.3 Pa. Thus, the response time and sensitivity of the 4He leak detector can be enhanced,8 especially in large vacuum vessels such as fusion machines. From the 4He-SL, a very small leak rate of 4He was constantly effused into the chamber and then, the D2 gas was effused. To increase the chamber pressure, the pumping speed for the chamber was reduced by regulating the valve V, indicated in Figure 1. As shown in Figure 3 the leaked 4He+ whose peak height corresponds to a leak rate of 4.1 x lo-‘” Pa * m’/s, and the overscale Dzis estimated to be of the order of IO-’ Pa m3/s giving a 4He’/Df peak ratio of 10m3. The deterioration of attainable 4He+/Dzpeak ratios from lop4 to 10eJ in analyzer-l is because of a longer inlet fringe length (i.e. the gap between the orifice plate and the quadrupole), whereas the fringe length in analyzer2 was optimized to 0.6 mm. We have indicated in a previous study5 that a change of the inlet fringe length on the order of 0.1 mm severely affected the sensitivity and resolution of the separated “He’/D: peaks. Therefore in analyzer-l, the “Hew/D: peak ratio of 10-j can be attained by further shortening the inlet fringe length. The background current in Figure 3 is only 2 x lo-” A at P, = 3.7 x lop3 Pa. On the other hand in analyzer-2, the background current reaches 1.8 x lo-” A at P, = 3.0 x 10-j Pa (see
S Hiroki
et al: Sensitive
helium
leak detection
ICi”A 3:
4He+
-
I
I
SW
Background
currenl
using analyzer-l at IO- ”
Figure 3. Time series of 4He/Dz peaks observed A/div and P, = 3.7 x IO- ’ Pa.
Figure 2(b)). Thus, the use of the differential pump unit with the orifice suppressed the background current to 1/lo. Lastly, we checked the background current characteristics of analyzers-l and -2 by effusing the 4He/D, gas mixture to each analyzer. In Figure 4 the background current change is plotted
as a function of pressure in the SEM volume at a SEM voltage of -2 kV. The background current in analyzer-l is roughly an order of magnitude larger than that in analyzer-2 at the same pressure, this may be due to a drop in a multiply coefficient in analyzer-2 over a longer period operation. From the curve obtained for analyzer-l (see Figure 4), we can estimate the suppression ratio of the background current obtained with differential pumping to that obtained without pumping. At a chamber pressure of 3.7 x 10 -’ Pa (see Figure 3). the background current was reduced to 2 x lo-” A using differential pumping. On the other hand without differential pumping, the chamber pressure of 3.7 x lW1 Pa produces a high background current of 2.5 x lo- ‘I’A, giving a suppression ratio of l/125. If analyzer-l isused todetect the SL-“He of4.1 x IO-“‘Pa . m’;s without the differential pump unit, the small ‘He. peak at 6 x ION ” A as shown in Figure 3 will disappear in the large background current of 2.5 x IO-“’ A. Therefore, the use of analyzer-l for the ‘He leak detector is very effective in enhancing the sensitivity. The lo-“’ Pa’ m3/s order ‘He leak rate achieved in this improved 4He leak detector satisfies the ITER-CDA (Conceptual although a further improvement Design Activity) requirement,’ such as suppressing the radioactive gas induced background noise’ is desirable. Conclusion The HR-QMS was applied to the ‘He leak detector for use in a DZ atmosphere. The lower detectable ‘He leak rate was limited by the background current, and the reduction in the background current using the differential pumped analyzer was estimated to be 1: 125. Therefore, the use of the differential pumped analyzer is a key to improving the sensitivity of the “He leak detector.
1 SEM -2 kV
./
.’
/
Analyzer-l ./’
.*
/ .’
Acknowledgements
.,~ .,,... LA”
The authors wish to thank Dr M Yoshikawa moto, both in JAERI, for encouragement.
.K“ Analyzer-2
and Dr S Shima-
References ‘B L Doyle, W R Wampler. D K Brice and S T Picraux. .I Nut, Mater. 93/94 551 (1980). ‘S Hiroki, Y Hasegawa, K Kaneko, T Abe and Y Murakami. J .VICC Matcw,
IO-14
’
’ L’
” 1 o-4
1
I
IO-'
10-2
Pressure in SEM Volume (Pa) Figure 4. Background current changes SEM volume of analyzers-l and -2.
as a function
of pressure
in the
224, 293 (1995).
‘W R Blanchard, R B Krawchuk and H F Dylla. J VNCSci T&nm/, 20, II62 (1982). ‘S Hiroki. T Abe and Y Murakaml, Rer Sci hrrunz. 65, I9 I2 ( 1994). ‘S Hiroki, T Abe and Y Murakami, fn/ J MUXSSpeclorrz forr Processc.v. 136, 85 (1994). ‘J von Seggern, S Berger, M Erdweg and W 0 Hofer. / Vrrc,.Y(,i T~c,hnol. A2, 1516 (1984). ‘W H Kohl. in Matrrids and Twhniqu~.~ for E/wtw~~Tzhcs.ReinholdKinokuniya (1960). *D T Santeler, VCICUU~.13, I02 (1963). ‘ITER Documentation Series, No 31 (IAEA. Vienna, 1991) p 33.
769
Pergamon PII: SOO42-207X(96)00063-2
47lnumbers 6-8lpages 767 to 769/1996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All riahts reserved 0042-207X196 $15.00+.00
Sensitive helium leak detection in a deuterium atmosphere using a high-resolution quadrupole mass spectrometer S Hiroki, T Abe and Y Murakami, Japan Atomic Energy Research Institute, Naka Fusion Research Establishment, Naka-machi, Naka-gun, lbaraki 31 l-07, Japan
In fusion machines, realizing a high-purity plasma is a key to improving the plasma parameters. Thus, leak detection is a necessary part of reducing the leak rate to a tolerable level. However, a conventional helium PHel leak detector is useless in fusion machines with a deuterium (0,) plasma, because retained D particles on the first wails release D2 for a long period and the released D2 interferes with the signals from the leaked 4He due to the near identical masses of 4.0026 u tdHe) and 4.0282 u (D2). A high-resolution quadrupole mass spectrometer (HR-QMS) that we have recently developed, can detect a 4He+ population as small as lop4 peak in a DZ atmosphere. Thus, the HR-QMS has been applied to detect 4He leaks. To improve the minimum detectable limit of 4He leak, a differentially pumped HR-OMS analyzer was attached to a chamber of the 4He leak detector. In conclusion, the improved 4He leak detector could detect 4He leaks of the order of lO_” Pa - m3/s in a D2 atmosphere. Copyright 0 1996 Elsevier Science ltd. Key words: Helium leak, deuterium
atmosphere,
high-resolution,
Introduction Current fusion machines such as JT-60 (JAERI Tokamak-60) and future machines like ITER (International Thermonuclear Experimental Reactor) under design as an international collaborative project have large and complicated vacuum vessels. There are hundreds of welds and flange joints on the vessel, and each area should be frequently checked for leaks. Because air leaks into a main plasma lead to a radiative power loss, the realization of an impurity free plasma is essential in improving the plasma parameters. Thus, leak detection is an integral part of maintaining impurity free plasma. The leaked 4He is generally detected with a conventional 4He leak detector that adopts a sector magnet type mass spectrometer. This 4He leak detector cannot, however, detect a small 4He leak in fusion machines using a D, plasma due to its poor resolving power for ‘He (4.0026 u) and Dz (4.0282 u). Since D particles are more easily retained on the first walls than are 4He particles,‘,’ the released Dz background interferes with the detection of the ‘He signals. To reduce the D? pressure to lower than the ‘He pressure in the 4He leak detector, a selective pumping of D, using a Zr-Al getter was proposed,3 although this non-throughput type pump has a limited capacity. The use of argon (Ar) as a tracer gas with a quadrupole mass spectrometer (QMS) was also proposed. However, the minimum detectable leak rate of Ar is far larger than that of ‘He, because of the higher Ar abundance in air and
quadrupole
mass spectrometer.
its larger Ar molecular diameter. Thus, the direct 4He detection in a high D2 background with high sensitivity is a better solution. Recently, we have developed a high-resolution QMS (HRQMS) that adopts a higher stability zone in the Mathieu stability diagram. This HR-QMS can detect at a population of lop3 in a D2 atm0sphere.j In addition, the fringing field characteristics of the HR-QMS were clarified.’ Based on the results, we have applied the HR-QMS to a 4He leak detector for use in a high Dz background. This paper describes initial experimental results on the improved 4He leak detector, where a differentially pumped analyzer was used to lower the minimum detectable limit of the 4He leak rate.
Experimental arrangement In this experiment, we used two analysers with and without the differential pump unit to clarify the effect of the pump unit. The analysers with and without the pump units were named analyzer1 and -2, respectively. In this section, an experimental arrangement using only analyzer-l is described and the results of using analyzer-2 will be discussed in the next section. The experimental setup, where analyzer-l is attached to an L-N2 trapped chamber is shown in Figure 1. A thin plate with a small orifice of 2.0 mm in diameter separated the ionizer spatially from the quadrupole. The orifice helped to ensure an efficient pumping of the space 767
S Hi&i
et al: Sensitive
helium
leak detection
(a
-
(b) TMP
I set
x104 b+
RP Figure 1. Experimental setup. PI and P,: Calibrated Bayard-Alpert type gauges, 4He-SL: “He standard leak of 4.1 x lo-” Pa+m3/s, SEM: Secondarv electron multiulier, L-N?: Liquid nitrogen. TMP: Magnet suspended turbomolecular pump, R~I Rotary pump: 1 Background
containing the quadrupole and the SEMs6 A 17-stage Cu-Be type SEM was used. An equivolume mixture of 4He/D, was introduced into the chamber to adjust the resolving power of the HR-QMS. A 4He standard leak (4He-SL) of 4.1 x lo-” Pa *m’/s was used to calibrate this 4He leak detector. Two ion gauges (P, and PJ were calibrated in advance by nitrogen with a spinning rotor gauge. The dimensions of the quadrupole in the two analysers were the same as previously used’ and the frequency of the RF voltage applied to the quadrupole was 4.9 MHz in both analyzers. Results and discussion Initially, analyzer-2 was attached to the chamber and its performance was compared with the results obtained later using analyzer-l. Analyzer-2 is a conventional analyzer with an open space between the ionizer and the quadrupole. The equivolume mixture of 4He/D, was introduced into the chamber and the HRQMS was tuned to separate the 4Hef peak from the DC peak resulting in a half-height resolution of about 200. The 4He and D, gases were then, independently effused into the chamber while gradually reducing the rate of the 4He effusion. The 4He’ and D: peaks are shown in Figure 2(a) and (b) at a 4He+/D:ratio of roughly lop4 in linear scales. The small 4He+ peak height in Figure 2(b) corresponds to a lo-* Pa . m’/s order of the 4He leak rate, and the DC peak height corresponds to a 1O-4 Pa . m’js order for the D, effusion rate. The spread of the D, peak width in Figure 2(b) results from a larger time constant of its amplifier range. Besides, the background current of 1.8 x lo-“. A in Figure 2(b) lifts the separated 4He+/D,+peaks, so this background current limits the minimum detectable 4He leak rate. The increase in the background current may be caused by ion feedback phenomena in the SEM due to a higher pressure in the analyzer. To restrict the pressure rise in the SEM volume, a normal ion effusion slit in the ionizer was replaced by a thin plate with the small orifice. The orifice was attached to analyzer-l as shown in Figure 1 and the effective pump speed of its differential pump unit was 0.012 m’js. The SEM was recommended to operate below 5 x 10-j Pa to protect against a breakdown in the SEM volume. When the 4He/Dz gas mixture was slowly introduced 768
current
I set Figure 2. Time series of 4HeC/D: peaks observed using analyzer-2 at (a) IO-’ A/div and (b) lo-” A/div. P, = 3.0 x 10-j Pa. In Figures 24, the applied voltage to the SEM was -2kV.
into the chamber, P, indicated 0.3 Pa whilst P, registered 5 x IO-’ Pa, thus the pressure ratio of P, to P, was 60. Hence the analyzable upper pressure limit can be expanded to 0.3 Pa. A filament of the ionizer will emit electrons in air-based gas mixtures when the HR-QMS is applied to the 4He leak detector. Frequent breaks of tungsten filaments in a higher H,O pressure can be overcome using a more suitable filament such as rhenium.’ In addition, analyzer-l can be directly attached between the turbomolecular pump and the fore pump because of the high maximum analyzable pressure of 0.3 Pa. Thus, the response time and sensitivity of the 4He leak detector can be enhanced,8 especially in large vacuum vessels such as fusion machines. From the 4He-SL, a very small leak rate of 4He was constantly effused into the chamber and then, the D2 gas was effused. To increase the chamber pressure, the pumping speed for the chamber was reduced by regulating the valve V, indicated in Figure 1. As shown in Figure 3 the leaked 4He+ whose peak height corresponds to a leak rate of 4.1 x lo-‘” Pa * m’/s, and the overscale Dzis estimated to be of the order of IO-’ Pa m3/s giving a 4He’/Df peak ratio of 10m3. The deterioration of attainable 4He+/Dzpeak ratios from lop4 to 10eJ in analyzer-l is because of a longer inlet fringe length (i.e. the gap between the orifice plate and the quadrupole), whereas the fringe length in analyzer2 was optimized to 0.6 mm. We have indicated in a previous study5 that a change of the inlet fringe length on the order of 0.1 mm severely affected the sensitivity and resolution of the separated “He’/D: peaks. Therefore in analyzer-l, the “Hew/D: peak ratio of 10-j can be attained by further shortening the inlet fringe length. The background current in Figure 3 is only 2 x lo-” A at P, = 3.7 x lop3 Pa. On the other hand in analyzer-2, the background current reaches 1.8 x lo-” A at P, = 3.0 x 10-j Pa (see
S Hiroki
et al: Sensitive
helium
leak detection
ICi”A 3:
4He+
-
I
I
SW
Background
currenl
using analyzer-l at IO- ”
Figure 3. Time series of 4He/Dz peaks observed A/div and P, = 3.7 x IO- ’ Pa.
Figure 2(b)). Thus, the use of the differential pump unit with the orifice suppressed the background current to 1/lo. Lastly, we checked the background current characteristics of analyzers-l and -2 by effusing the 4He/D, gas mixture to each analyzer. In Figure 4 the background current change is plotted
as a function of pressure in the SEM volume at a SEM voltage of -2 kV. The background current in analyzer-l is roughly an order of magnitude larger than that in analyzer-2 at the same pressure, this may be due to a drop in a multiply coefficient in analyzer-2 over a longer period operation. From the curve obtained for analyzer-l (see Figure 4), we can estimate the suppression ratio of the background current obtained with differential pumping to that obtained without pumping. At a chamber pressure of 3.7 x 10 -’ Pa (see Figure 3). the background current was reduced to 2 x lo-” A using differential pumping. On the other hand without differential pumping, the chamber pressure of 3.7 x lW1 Pa produces a high background current of 2.5 x lo- ‘I’A, giving a suppression ratio of l/125. If analyzer-l isused todetect the SL-“He of4.1 x IO-“‘Pa . m’;s without the differential pump unit, the small ‘He. peak at 6 x ION ” A as shown in Figure 3 will disappear in the large background current of 2.5 x IO-“’ A. Therefore, the use of analyzer-l for the ‘He leak detector is very effective in enhancing the sensitivity. The lo-“’ Pa’ m3/s order ‘He leak rate achieved in this improved 4He leak detector satisfies the ITER-CDA (Conceptual although a further improvement Design Activity) requirement,’ such as suppressing the radioactive gas induced background noise’ is desirable. Conclusion The HR-QMS was applied to the ‘He leak detector for use in a DZ atmosphere. The lower detectable ‘He leak rate was limited by the background current, and the reduction in the background current using the differential pumped analyzer was estimated to be 1: 125. Therefore, the use of the differential pumped analyzer is a key to improving the sensitivity of the “He leak detector.
1 SEM -2 kV
./
.’
/
Analyzer-l ./’
.*
/ .’
Acknowledgements
.,~ .,,... LA”
The authors wish to thank Dr M Yoshikawa moto, both in JAERI, for encouragement.
.K“ Analyzer-2
and Dr S Shima-
References ‘B L Doyle, W R Wampler. D K Brice and S T Picraux. .I Nut, Mater. 93/94 551 (1980). ‘S Hiroki, Y Hasegawa, K Kaneko, T Abe and Y Murakami. J .VICC Matcw,
IO-14
’
’ L’
” 1 o-4
1
I
IO-'
10-2
Pressure in SEM Volume (Pa) Figure 4. Background current changes SEM volume of analyzers-l and -2.
as a function
of pressure
in the
224, 293 (1995).
‘W R Blanchard, R B Krawchuk and H F Dylla. J VNCSci T&nm/, 20, II62 (1982). ‘S Hiroki. T Abe and Y Murakaml, Rer Sci hrrunz. 65, I9 I2 ( 1994). ‘S Hiroki, T Abe and Y Murakami, fn/ J MUXSSpeclorrz forr Processc.v. 136, 85 (1994). ‘J von Seggern, S Berger, M Erdweg and W 0 Hofer. / Vrrc,.Y(,i T~c,hnol. A2, 1516 (1984). ‘W H Kohl. in Matrrids and Twhniqu~.~ for E/wtw~~Tzhcs.ReinholdKinokuniya (1960). *D T Santeler, VCICUU~.13, I02 (1963). ‘ITER Documentation Series, No 31 (IAEA. Vienna, 1991) p 33.
769