- Email: [email protected]
Design principle and performance evaluation of a newtype leak detector
Vacuum/volume
PII: SOO42-207X(96)00014-0
47/numbers 6-6Ipages 531 to 53411996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All riahts reserved 0042-207X/96 $15.00+.00
Design principle and performance evaluation of a newtype leak detector G Horikoshi” and M Kobayashi, Ibaraki-ken 305, Japan
National Laboratory
for High Energy Physics (KEKI, Oho l-l,
Tsukuba-shi,
The design principle of a new-version counter flow leak detector equipped with two turbomolecular pumps is given. In the design, an emphasis was put on the enrichment of the probe gas and the S/N ratio as well as on the response time constant, A general performance evaluation of various kinds of leak-detecting systems was also made in terms of the minimum detectable leak, response time constant and a newly introduced “detectability“. Copyright 0 1996 Elsevier Science Ltd. Key words: Compression time constant.
ratio, counter flow leak detector, leak detection, minimum
detectable leak, response
Introduction
New-version counter-flow leak detector
Since the first proposal of a counter-flow (helium) leak detector (CFLD) by Becker,’ it has been widely used in leak-detecting systems. It contains only a small high vacuum pump having a low ultimate compression ratio but without a fore-pump of the high vacuum pump. In a leak-hunting job, a CFLD is connected to the fore-pump inlet of the vacuum system. Although most of the helium gas entering through a leak reaches the fore-pump and is pumped out, a small part moves to the backside of the high vacuum pump in the CFLD, and goes up through the high vacuum pump by a counter-flow process to the sensor manifold. In a CFLD, the counter-flow to the sensor manifold largely diminishes the helium gas signal. Therefore, it has been incorrectly considered that a CFLD is less sensitive than an ordinarytype LD because of the counter-flow process. In 1987, Reich pointed out the advantages of the CFLD, and proposed a newtype CFLD with two turbomolecular pumps.’ He discussed mainly how to reduce the total working time in the leak-detecting job, and also referred to the possibility of a helium-gas enrichment. One of the authors (Horikoshi et al.) proposed a newversion CFLD and gave an analysis of the minimum detectable leak and response time constant of the signal.3,4 In this paper we take up the new-version CFLD, and evaluate its leak detecting performance in terms of the minimum detectable leak, response time constant and a newly introduced criterion of detectability.
Figure 1 shows schematic diagrams of some leak-detecting systems. Figure I(a) shows a system with an ordinary leak detector, and Figures l(b) and (c) show systems with the new-version CFLD. In the latter two cases, two turbomolecular pumps (TMPA and TMP-B) and a sensor manifold are the main components of the leak detector. Backing lines of these TMPs are connected to a fore-pump with an effective pumping speed of S,.. The pump inlet of TMP-A is connected to a test chamber through a connection tube of conductance C. A part of the helium gas entering through a leak in the test chamber will finally reach the sensor. Along the way to the sensor, helium gas is compressed by TMPA and diluted by TMP-B. If the ultimate compression ratios of these TMPs are suitable, we can enrich the helium and can improve the minimum detectable leak. The key point of the design is to select suitable values of the ultimate compression ratio of both TMPs (R, and &) and the effective pumping speed of the fore-pump (S,). It is also important that these quantities be compromised with the response time constant.
* Home address: Kashiwada 3322-7, Japan (Tel. & Fax: YI-298-72-5944).
Ushiku-shi,
Ibaraki-ken,
30@-12.
Analysis Gas enrichment factor. We now estimate the ratio of the sensor manifold pressure (PJ to the inlet pressure of TMP-A @) for hydrogen, helium, water vapour and nitrogen, which are the main gases in the analysis. Hydrogen is the lightest and most active gas in the counter-flow process. Water vapour is the main component of desorption gas and nitrogen represents air leakage. Let the pumping speed and ultimate compression ratio of TMP-A and TMP-B be S,, &, RA and RB, respectively. In general, the 531
G Horikoshi
and M Kobayashi:
New-type
leak detector
SB
KpF
- SBp2 = O,
where we have neglected gas generation in the sensor manifold. From eqn (1) we obtain the pressure ratio @Jp), or the enhancement factor (gas enrichment factor), as RASA
P2 p
01
r-4
NCFLD
(b)
= (RASF+
(2)
SJRB’
It is clear that the conditions RA >>RB
and
$& >> SF
(3)
A
NCFLD
(cl Figure 1. Schematic diagrams of some leak-detecting systems. (a) A small system with an ordinary leak detector. (b) A small system with a new version CFLD. (c) A large system with an auxiliary pump and a new version CFLD.
pumping speed of a well-designed TMP is almost equal irrespective of the gas species, while the ultimate compression ratio strongly depends on the molecular weight of the gas. Using the notations given in Figure 1 and Table 1 and letting the pressure at the backside of the TMPs be pF, we obtain the following set of equations:
(
are necessary to obtain a large enhancement factor. It should be kept in mind that if we use an orifice with small hole(s) to limit SF, SF depends on the molecular weight of the gas. This fact is in favour of the new-version CFLD, in suppressing the enrichment factor for hydrogen. We assume an orifice of 0.1 mm thin film with a number of holes of 0.5 mm diameter, which satisfies the molecular flow condition up to 10 Pa. Using the values of RA and Re given in Table 2,3,4 we can estimate the enhancement factor for the main gases as a function of SF. Figure 2 shows the result and indicates that the new-version CFLD enriches helium and hydrogen gases selectively, and that a suitable design will improve the ratio of helium to hydrogen pressure in the sensor manifold without an intolerable increase of the response time.
Table 2. Performance Symbol
and quantities
a
used in the evaluation
b
c
100 0.03
100 100 0.03
ss,WI
2+$+& A
SA( = m [l/s1 SF(for W [I/s1 SL[I/s1 VASl
p,-s,p,=s,p B
)
and
vc[II
C (for N2) Ml Table 1. Definitions Notation
and notations
Definition
used in the analysis
H2 ka l/s]
and explanation
Electrical fluctuation reduced to helium pressure in Pa Response time constant Connection tube conductance from leak detector to test chamber Detectability ( = (L Minimum detectabl%& “))‘) Gas generation rate in test chamber and sensor manifold Ultimate compression ratio of turbomolecular pump A and B Pumping speed of turbomolecular pump A and B Pumping speed of fore-pump in new version CFLD Pumping speed of ordinary leak detector Pumping speed of auxiliary pumping system Volume of inlet part of TMP-A in new version CFLD Volume of test chamber Volume of backside manifold of two TMPs Volume of sensor manifold
pa l/s]
100
0.2 2 0.1 0.2 2
2
VF[II K PI
532
evaluation
He Hz0 N*
0.5 2 1.0x 1.0 x 1.0 x 1.0 x
lomx 10m8 10-j 10-l
1.0x 1.0 x 1.0 x 1.0 x
0.2 100 0.1 0.2 4
L
lo-” lo-” 1om3 lo-’
1.4x 1.0 x 1.4x 1.4x
lo-’ lo-’ IO_’ lo-’
4.0 x lo-’
2.0 x lo-’
2.0 X IO_’
4.0 x lo-.4
2.0 x 10 4
2.0 x lo-”
HZ He H,O N,
650 5500 I .o x IO8 1.0 x 10’”
650 5500 1.0x lo* 1.o x 10’”
HZ He Hz0 Nz
25 74 1.0x IO4 1.0x 10” 1.0 x IO_” 2.24x IO-” 1.30 2.91 x lo-”
25 14 1.0 x lo4 1.0x 105 1.0 x IO_‘? 3.60 x lo-” 1.28 4.61 x lo-“’
HZ He Hz0 N2
1.0x lo-‘? X.06x IO-‘” 0.40 3.22 x 10 ‘”
G Horikoshi
and M Kobayashi:
New-type
leak detector Finally, we classify all sensor outputs into signal S (by helium) and noise N (by gases other than helium), and determine the minimum detectable leak by the condition .S/N = I. Response time constant. The response time constant is defined as a time constant for the helium pressure to build up in the sensor manifold when a constant rate of helium gas generation takes place at t = 0. The pressure build-up phenomenon can be analysed by using differential equations.
K~;4
= [P,l-[C&J,(r)].
with an initial condition of b,(O)] = 0. where V, is the volume of the vacuum component corresponding to the ith nodal point. The determination of the response time constant is simple. The pressure build-up phenomenon of the system can almost be expressed by a set of time constants (T,;k = 1,2,....n). The values of si are given by
where Ai is one of the roots of L
0
0.05
0.1
]C:,-(-i)6,,]
purnpiii&ed SF [IS’] Figure 2. Enhancement factor of new version CFLD effective pumping speed of fore-pump (S,.).
(i,.j=
1,7 ,... PI).
(7)
with the C,,s defined by as a function
of
Minimum detectable leak. We estimate the minimum detectable leak of a system by using a general method of network analysis by a matrix.’ At first, we replace the vacuum system with a network of vacuum components with n nodal points. We specify the amounts of gas generation for every nodal point and let them be Q, (i = 1.2,....n). Letting p, be the pressures at the ith nodal point, we obtain the following equation in a matrix form for each gas species:
[Cj,][P/l = @,I> (i,j = 12 ,..,n)
= 0
(4)
where [C,,] is a matrix having n x n elements. An estimation of the minimum detectable leak can be made as follows.‘-4 At first, we solve eqn (4) in terms ofp,, and determine the partial pressures in the sensor manifold for the main gases; we then calculate the sensor outputs by using the relative sensitivities for the gases which are experimentally determined and are given in Table 3.’
Because the system will finally approach a stationary state. the time constants should be real and positive, and the largest T will give the response time constant. Letting the minimum root of eqn (7) and the response time constant be E., and rK, respectively, the response time constant. TV. is given by
Performance evaluation of various leak-detecting systems In Table 2 we summarize a performance evaluation of some leak-detecting systems as well as the assumed parameters in the evaluation. The evaluation was made in terms of the minimum detectable leak, response time constant and newly introduced criterion of detectability, (D), i.e. A=(
minimum
detectable
leak) x (response
time constant).
Figure I gives the constitution of each systetn schematically. result shows the superiority of the new-versron CFLD. Table 3. Relative helium helium
sensor
sensitivities of for gases other than
Gas
Relative sensitivity
H,
5x10 5x10 5x10
H,O
NZ
i i q
( IO) The
Summary and conclusions The concept of a new-version CFLD was introduced. Because the performance of the new-version CFLD depends largely on the gas enrichment for helium, an analysis of the gas enrichment process was carried out. The analysis gives a guideline for selecting a suitable value of the design parameters, such as the ultimate compression ratio and the pumping speed of the pumps. Although the enrichment function acts well not only on helium, 533
G Horikoshi
and M Kobayashi:
New-type
leak detector
but also on hydrogen gas, we can find good design parameters to improve the leak-detecting system. An overall evaluation of some leak-detecting systems was made in terms of the minimum detectable leak, response time constant and newly introduced criterion of detectability, which is defined by eqn (9). The results of the overall evaluation show the superiority of the new version CFLD.
534
References ‘W Becker, Vak Tech, 17, 203 (1968). ‘G Reich, .I Vat Sci Tech&, A5(4), 2641 (1987). ‘G Horikoshi and K Kakihara, Shinku, 38(3), 195 (1995) (in Japanese). 4G Horikoshi, Fizika A, to be published. ‘G Horikoshi. Y Saito and K Kakihara, Vacuum, 41(7-g), 2132 (1990).
PII: SOO42-207X(96)00014-0
47/numbers 6-6Ipages 531 to 53411996 Copyright 0 1996 Elsevier Science Ltd Printed in Great Britain. All riahts reserved 0042-207X/96 $15.00+.00
Design principle and performance evaluation of a newtype leak detector G Horikoshi” and M Kobayashi, Ibaraki-ken 305, Japan
National Laboratory
for High Energy Physics (KEKI, Oho l-l,
Tsukuba-shi,
The design principle of a new-version counter flow leak detector equipped with two turbomolecular pumps is given. In the design, an emphasis was put on the enrichment of the probe gas and the S/N ratio as well as on the response time constant, A general performance evaluation of various kinds of leak-detecting systems was also made in terms of the minimum detectable leak, response time constant and a newly introduced “detectability“. Copyright 0 1996 Elsevier Science Ltd. Key words: Compression time constant.
ratio, counter flow leak detector, leak detection, minimum
detectable leak, response
Introduction
New-version counter-flow leak detector
Since the first proposal of a counter-flow (helium) leak detector (CFLD) by Becker,’ it has been widely used in leak-detecting systems. It contains only a small high vacuum pump having a low ultimate compression ratio but without a fore-pump of the high vacuum pump. In a leak-hunting job, a CFLD is connected to the fore-pump inlet of the vacuum system. Although most of the helium gas entering through a leak reaches the fore-pump and is pumped out, a small part moves to the backside of the high vacuum pump in the CFLD, and goes up through the high vacuum pump by a counter-flow process to the sensor manifold. In a CFLD, the counter-flow to the sensor manifold largely diminishes the helium gas signal. Therefore, it has been incorrectly considered that a CFLD is less sensitive than an ordinarytype LD because of the counter-flow process. In 1987, Reich pointed out the advantages of the CFLD, and proposed a newtype CFLD with two turbomolecular pumps.’ He discussed mainly how to reduce the total working time in the leak-detecting job, and also referred to the possibility of a helium-gas enrichment. One of the authors (Horikoshi et al.) proposed a newversion CFLD and gave an analysis of the minimum detectable leak and response time constant of the signal.3,4 In this paper we take up the new-version CFLD, and evaluate its leak detecting performance in terms of the minimum detectable leak, response time constant and a newly introduced criterion of detectability.
Figure 1 shows schematic diagrams of some leak-detecting systems. Figure I(a) shows a system with an ordinary leak detector, and Figures l(b) and (c) show systems with the new-version CFLD. In the latter two cases, two turbomolecular pumps (TMPA and TMP-B) and a sensor manifold are the main components of the leak detector. Backing lines of these TMPs are connected to a fore-pump with an effective pumping speed of S,.. The pump inlet of TMP-A is connected to a test chamber through a connection tube of conductance C. A part of the helium gas entering through a leak in the test chamber will finally reach the sensor. Along the way to the sensor, helium gas is compressed by TMPA and diluted by TMP-B. If the ultimate compression ratios of these TMPs are suitable, we can enrich the helium and can improve the minimum detectable leak. The key point of the design is to select suitable values of the ultimate compression ratio of both TMPs (R, and &) and the effective pumping speed of the fore-pump (S,). It is also important that these quantities be compromised with the response time constant.
* Home address: Kashiwada 3322-7, Japan (Tel. & Fax: YI-298-72-5944).
Ushiku-shi,
Ibaraki-ken,
30@-12.
Analysis Gas enrichment factor. We now estimate the ratio of the sensor manifold pressure (PJ to the inlet pressure of TMP-A @) for hydrogen, helium, water vapour and nitrogen, which are the main gases in the analysis. Hydrogen is the lightest and most active gas in the counter-flow process. Water vapour is the main component of desorption gas and nitrogen represents air leakage. Let the pumping speed and ultimate compression ratio of TMP-A and TMP-B be S,, &, RA and RB, respectively. In general, the 531
G Horikoshi
and M Kobayashi:
New-type
leak detector
SB
KpF
- SBp2 = O,
where we have neglected gas generation in the sensor manifold. From eqn (1) we obtain the pressure ratio @Jp), or the enhancement factor (gas enrichment factor), as RASA
P2 p
01
r-4
NCFLD
(b)
= (RASF+
(2)
SJRB’
It is clear that the conditions RA >>RB
and
$& >> SF
(3)
A
NCFLD
(cl Figure 1. Schematic diagrams of some leak-detecting systems. (a) A small system with an ordinary leak detector. (b) A small system with a new version CFLD. (c) A large system with an auxiliary pump and a new version CFLD.
pumping speed of a well-designed TMP is almost equal irrespective of the gas species, while the ultimate compression ratio strongly depends on the molecular weight of the gas. Using the notations given in Figure 1 and Table 1 and letting the pressure at the backside of the TMPs be pF, we obtain the following set of equations:
(
are necessary to obtain a large enhancement factor. It should be kept in mind that if we use an orifice with small hole(s) to limit SF, SF depends on the molecular weight of the gas. This fact is in favour of the new-version CFLD, in suppressing the enrichment factor for hydrogen. We assume an orifice of 0.1 mm thin film with a number of holes of 0.5 mm diameter, which satisfies the molecular flow condition up to 10 Pa. Using the values of RA and Re given in Table 2,3,4 we can estimate the enhancement factor for the main gases as a function of SF. Figure 2 shows the result and indicates that the new-version CFLD enriches helium and hydrogen gases selectively, and that a suitable design will improve the ratio of helium to hydrogen pressure in the sensor manifold without an intolerable increase of the response time.
Table 2. Performance Symbol
and quantities
a
used in the evaluation
b
c
100 0.03
100 100 0.03
ss,WI
2+$+& A
SA( = m [l/s1 SF(for W [I/s1 SL[I/s1 VASl
p,-s,p,=s,p B
)
and
vc[II
C (for N2) Ml Table 1. Definitions Notation
and notations
Definition
used in the analysis
H2 ka l/s]
and explanation
Electrical fluctuation reduced to helium pressure in Pa Response time constant Connection tube conductance from leak detector to test chamber Detectability ( = (L Minimum detectabl%& “))‘) Gas generation rate in test chamber and sensor manifold Ultimate compression ratio of turbomolecular pump A and B Pumping speed of turbomolecular pump A and B Pumping speed of fore-pump in new version CFLD Pumping speed of ordinary leak detector Pumping speed of auxiliary pumping system Volume of inlet part of TMP-A in new version CFLD Volume of test chamber Volume of backside manifold of two TMPs Volume of sensor manifold
pa l/s]
100
0.2 2 0.1 0.2 2
2
VF[II K PI
532
evaluation
He Hz0 N*
0.5 2 1.0x 1.0 x 1.0 x 1.0 x
lomx 10m8 10-j 10-l
1.0x 1.0 x 1.0 x 1.0 x
0.2 100 0.1 0.2 4
L
lo-” lo-” 1om3 lo-’
1.4x 1.0 x 1.4x 1.4x
lo-’ lo-’ IO_’ lo-’
4.0 x lo-’
2.0 x lo-’
2.0 X IO_’
4.0 x lo-.4
2.0 x 10 4
2.0 x lo-”
HZ He H,O N,
650 5500 I .o x IO8 1.0 x 10’”
650 5500 1.0x lo* 1.o x 10’”
HZ He Hz0 Nz
25 74 1.0x IO4 1.0x 10” 1.0 x IO_” 2.24x IO-” 1.30 2.91 x lo-”
25 14 1.0 x lo4 1.0x 105 1.0 x IO_‘? 3.60 x lo-” 1.28 4.61 x lo-“’
HZ He Hz0 N2
1.0x lo-‘? X.06x IO-‘” 0.40 3.22 x 10 ‘”
G Horikoshi
and M Kobayashi:
New-type
leak detector Finally, we classify all sensor outputs into signal S (by helium) and noise N (by gases other than helium), and determine the minimum detectable leak by the condition .S/N = I. Response time constant. The response time constant is defined as a time constant for the helium pressure to build up in the sensor manifold when a constant rate of helium gas generation takes place at t = 0. The pressure build-up phenomenon can be analysed by using differential equations.
K~;4
= [P,l-[C&J,(r)].
with an initial condition of b,(O)] = 0. where V, is the volume of the vacuum component corresponding to the ith nodal point. The determination of the response time constant is simple. The pressure build-up phenomenon of the system can almost be expressed by a set of time constants (T,;k = 1,2,....n). The values of si are given by
where Ai is one of the roots of L
0
0.05
0.1
]C:,-(-i)6,,]
purnpiii&ed SF [IS’] Figure 2. Enhancement factor of new version CFLD effective pumping speed of fore-pump (S,.).
(i,.j=
1,7 ,... PI).
(7)
with the C,,s defined by as a function
of
Minimum detectable leak. We estimate the minimum detectable leak of a system by using a general method of network analysis by a matrix.’ At first, we replace the vacuum system with a network of vacuum components with n nodal points. We specify the amounts of gas generation for every nodal point and let them be Q, (i = 1.2,....n). Letting p, be the pressures at the ith nodal point, we obtain the following equation in a matrix form for each gas species:
[Cj,][P/l = @,I> (i,j = 12 ,..,n)
= 0
(4)
where [C,,] is a matrix having n x n elements. An estimation of the minimum detectable leak can be made as follows.‘-4 At first, we solve eqn (4) in terms ofp,, and determine the partial pressures in the sensor manifold for the main gases; we then calculate the sensor outputs by using the relative sensitivities for the gases which are experimentally determined and are given in Table 3.’
Because the system will finally approach a stationary state. the time constants should be real and positive, and the largest T will give the response time constant. Letting the minimum root of eqn (7) and the response time constant be E., and rK, respectively, the response time constant. TV. is given by
Performance evaluation of various leak-detecting systems In Table 2 we summarize a performance evaluation of some leak-detecting systems as well as the assumed parameters in the evaluation. The evaluation was made in terms of the minimum detectable leak, response time constant and newly introduced criterion of detectability, (D), i.e. A=(
minimum
detectable
leak) x (response
time constant).
Figure I gives the constitution of each systetn schematically. result shows the superiority of the new-versron CFLD. Table 3. Relative helium helium
sensor
sensitivities of for gases other than
Gas
Relative sensitivity
H,
5x10 5x10 5x10
H,O
NZ
i i q
( IO) The
Summary and conclusions The concept of a new-version CFLD was introduced. Because the performance of the new-version CFLD depends largely on the gas enrichment for helium, an analysis of the gas enrichment process was carried out. The analysis gives a guideline for selecting a suitable value of the design parameters, such as the ultimate compression ratio and the pumping speed of the pumps. Although the enrichment function acts well not only on helium, 533
G Horikoshi
and M Kobayashi:
New-type
leak detector
but also on hydrogen gas, we can find good design parameters to improve the leak-detecting system. An overall evaluation of some leak-detecting systems was made in terms of the minimum detectable leak, response time constant and newly introduced criterion of detectability, which is defined by eqn (9). The results of the overall evaluation show the superiority of the new version CFLD.
534
References ‘W Becker, Vak Tech, 17, 203 (1968). ‘G Reich, .I Vat Sci Tech&, A5(4), 2641 (1987). ‘G Horikoshi and K Kakihara, Shinku, 38(3), 195 (1995) (in Japanese). 4G Horikoshi, Fizika A, to be published. ‘G Horikoshi. Y Saito and K Kakihara, Vacuum, 41(7-g), 2132 (1990).