The new PTB standard for dynamic vacuum pressures Karl Jousten, Physikalisch-Technische Bundesanstalt, Berlin, Germany 1. Idea and design of the standard 2. Development 3. Pressure approximation and experiments 4. Conclusions and Discussion Vacuum Metrology for Industry, EMRP IND12 Workshop June 25-27, 2014
What do we need for such a standard? 1. The response time (1/e decrease and increase) of a vacuum gauge shall be measurable. A pressure drop as fast as possible must be generated. 2. A realistic decrease/increase of pressure in a load lock shall be simulated to test, if the gauge can follow the pressure change. The pressure versus time curve must be predictable ( 50% uncertainty at any time); from 100 kpa to 100 Pa within 1 s. 3. Different pressure vs. time curves shall be producable to simulate different vacuum systems and gauges. Also venting load lock shall be simulated. The standard must be flexible. 2
Fast opening valve High pressure volume Low pressure volume p 4 f f2 p10 6.88910 p01 3
Important design goals 1. The exponential pressure decay (p ini - p final ) becomes slow after a few time constants: The volume ratio shall be > 1000 for expansion from 100 kpa to 100 Pa. For clear limits for large volume: minimize small volume! 2. The dead volume between duct and fast opening valve generates an offset of final pressure. This offset must be << 100 Pa and the dead volume as small as possible (is 0.033L). 3. The opening time of the valve must be negligible compared to the time in which there is a significant pressure change (within allowed uncertainty of 50%): < 10 ms. 4. Reduce calculation effort and have equal positions for test gauges: Symmetry around cylindrical axis 4
area in cm² Dynamic vacuum pressure standard Fast opening gate valve by VAT company 4.6 ms 0.6 ms 2.6 ms 3.3 ms 5.5 ms time in ms Opening phase investigated with high speed camera with 10,000 f/s 5
Pressures during expansion: dp1 t Cp t p t V1 1 2 dt t dp2 V2 Cp1t p2 dt t Simplify with C constant (varies by a factor of 2) and use V 2 >> V 1 : p t p p p exp t 1 20 10 20 / 1 p1( t) exp t p10 1 1 V 1 C 7* 1 needed for p 1 /p 10 =0.001 6
In total 4 different conductances available: 1) The full opening 85 L/s 1 0.71 ms (different passive volume) 2) Orifice 1.8 L/s 1 55 ms 3) Laval nozzle 1 1.07 L/s 1 93 ms 4) Laval nozzle 2 2.14 L/s 1 47 ms 7
Pressure during expansion Simplifications for an estimate: constant temperature, choked flow (flow function ), negligible downstream pressure q Kc p pv 1 1 K p1( t) exp p A 4 p 2 1 t p01 1 V1 1 Kc Choked flow was checked by varying down stream pressure (1, 2, and 5 kpa) 1 8
Experiments and results piezo-resistive gauge Capacitance Diaphragm Gauges 9
Pressure step/full opening p 01 *exp(-t/1.7ms) p 01 *exp(-t/1.3ms) Vacuum gauges under test: CDGs by Inficon 130 kpa, 1.3 kpa Signal every 0.7 ms, 21bit Integration time 0.3 ms System: full DN40 opening, 117 mm length 85 L/s, 0.71 ms Valve area opening time 10
Pressure step After 20 ms 100 Pa! 11
Slower expansion through orifice Deviation < 27% everywhere 12
Pressure decay with various gases Should scale with product K and mean thermal velocity c 1 V1 1 Kc 1 Gas (meas.) (choked) Helium 0.368 0.356 Nitrogen 1.000 1.000 Air 1.023 1.018 Argon 1.127 1.127 Krypton 1.616 1.631 Xenon 2.020 2.042 13
Uncertainty of pressure measurement with CDGs as secondary standards (Condition: Orifice with = 78 ms) Factor of influence CDG 1.3 kpa rel. unc. in % 73 Pa < p 1 1.3 kpa CDG 133 kpa rel. unc. in % 1.3 kpa < p 1 <5 kpa CDG 133 kpa rel. unc. in % 5 kpa p 1 <100 kpa Calibration uncertainty 0.30 0.82 0.18 Long-term instability 0.50 0.30 0.30 Zero instability 0.30 0.76 0.15 Instability due to shock 1.12 0.02 0.18 Digitizing error 0.03 0.29 0.06 Dynamic response 0.29 0.29 0.29 Total standard uncertainty 0.72 1.23 0.48 Expanded uncertainty (95 %, k=2) 1.4 2.5 1.0 14
Reverse flow into V 1 (simulating venting of load lock) Faster time constant because of constant upstram pressure (100 kpa) and viscous flow all the time Oscillations due to mechanical shock 15
Conclusions 1 The technical concept of the dynamic vacuum pressure standard worked fine: A well defined pressure step of three decades (100 kpa to 100 Pa) can be realized within 20 ms. To accomplish this, the fastest UHV gate valve of the world was developed. It opens DN40 fully within 4.6 ms. We can generate the fastest well known vacuum possible in the world. Both fast pressure increases and decreases with variable time constants can be generated and many types of gauges for rough and fine vacuum can be tested (a suitable volume, flange and reasonable update time is required). 16
Conclusions 2 Our results let conclude that the pressure is predictable by simulations within the target uncertainty of 50% at any time. But: It makes sense to convert the whole concept of the dynamic standard: Calibrate the CDGs by INFICON in equilibrium, ensure that time constant < 2ms and use them as reference standards for p(t). This reduces the uncertainty to < 3% at any time. Then, theoretical approaches can be tested by agreement with these reference standards. 17
Acknowledgements This work is part of the EMRP IND12 project. The EMRP is jointly funded by the EMRP participating countries within EURAMET and the European Union. PTB acknowledges the excellent cooperation with the industrial partners within EMRP IND12 the VAT company. and my colleagues S. Pantazis, J. Buthig, R. Model. 18
My question to the audience Does this dynamic vacuum pressure standard meet the need of industry? (Time trigger necessary? Extension to lower pressures needed? Is the time scale right? Are you interested in a calibration of time response?) 19