64th IUVSTA Workshop, May 16-19, 2011 Problems of vacuum metrology for industrial applications that call for solutions by rarefied gas dynamics Karl Jousten,

64th IUVSTA Workshop, May 16-19, 2011

Problems of vacuum metrology for industrial applications that call for solutions by rarefied gas dynamics

Karl Jousten, PTB, Berlin

1. Applications of vacuum and leak detection with conclusions for rarefied gas dynamics

2. Present state: Measurement standards for vacuum and low flow rates (leaks)

3. Four problems for rarefied gas dynamics

4. Conclusion

Problems of vacuum metrology for rarefied gas dynamics, IUVSTA WS 64, 2011 2

Applications of vacuum in science

Gravitational wave detectors

Elementary particle physics: Accelerators, KATRIN

Fusion: ITER

Surface physics

Conclusions

In most cases just sufficiently low gas density

In some cases complicated and extensive design calculations of vacuum system (ITER, KATRIN)

Reliable pumping speed values needed!

Applications of vacuum and leak detection

Virgo-Detector, near Pisa with 3 km long vacuum tubes


Pumping speed measurement

p

qS pV

Turbomolecular pump, Pfeiffer Vacuum


From science to industry, standardization:

How and where to measure p?

This (and hence S) is defined by international standards, but is it a physical quantity for design and theoretical calculations?

Conclusions: Physical relevance of standardized quantities needs to be adressed.

pVq p ?


Microelectronic industry


MOCVD: Reactor forferroelectric films

Cluster toolAIS, Dresden

Typical: fast processes


Industrial applications: CD/DVD metallization

Type Cathode Pumps Cycle-time

SINGULUS V Focus 1

Focus Cathode 1 Turbo Molecular Pump

2,7 s

SINGULUS V Focus 2

Focus Cathode 2 Turbo Molecular Pumps

1,5 s

SINGULUS V Smart 1

SMART CATHODE®

1 Turbo Molecular Pump

2,5 s

SINGULUS V Smart 2

SMART CATHODE®

2 Turbo Molecular Pumps

1,9 s




Other examples of fast processes:

Leak tests of rims for cars (mainly aluminum for light wheels)

Coating of bottles (food industry)

1.2 s … 2.5 s

PET- bottle coating Fa. Sidel company

Conclusions: Fast changes of pressure and gas flows need to be calculated for engineering design.


Example of non-detructive leak test: pacemaker and air bag

Required tightness:< 10-7 Pa L/s !

Required tightness:< 10-5 Pa L/s


Source: St. Jude Medical, Sweden


Airconditioning in cars, refrigerators etc. : Environmental issues

Required tightness:< 10-3 Pa L/s (1 g/a)


Conclusions: Leak testing is normally performed with helium, but neededare leak rates for other gases and even liquids.


For a long time until about 1990:Magnetic sectors are used for leak detection and quadrupole mass spectrometers (QMS)for both leak detection and analysis of background residual gas level causing the name Residual gas analyzers.

Nowadays in addition QMS for:Gas purity, in-situ analysis for reagent gasesand low-level components in semiconductorIndustry.• Sputter process control• CVD monitoring, gas abatement analysis• MBE source control• End point detection (etching)• Gas chromatography


Partial pressuremeasurement

Etching: end point detection


Applications of QMS < 1E-2 Pa:Direct installation of QMS to process chamber.

1E-2 Pa:Differential pumping necessary with manifold, conductance, high vacuum pump, and total pressure gauge.Conductance for sputtering pressures PVD ( 0.1 Pa): Orifice, Dual inlet (RGA + process)for MOCVD (up to 100 kPa): Capillary

Conclusions: Mass (gas species) discrimination within the measuring device?

G as

To vacuum pum ps

To Q M SProcess G asC ham ber

Ulvac Co



1. Fast processes: dynamics

2. Physical relevance of standardized quantities: helium leak rate and S

3. QMS as process tool: Discrimination for partial pressure measurement

4. Design of complex vacuum systems

Summary from applications


2. Present state: Measurement standards for vacuum and low flow rates (leaks)


Measurement standards for vacuum and low gas flow


Pressure balance as primary standard

effA

Fp



Static or series expansion system as primary standard

21

112 VV

Vpp

V1,p1V2,p2


V4

100 L

V6

100 L

V3

V2

0,1 L

1 L

1 L

1 L

GAS INLET

UUC

V1

V7

V5

Example forstatic expansion

PTB, SE2




Static expansion system SE2, PTB, Berlin

Accurate pressures from 10-2 Pa to 103 Pa


Continuous expansion system as primary standard

p2 p3

gas flow

C1<< C2

p1

flowmeter2

112 C

Cpp



p=const

t

p

Gas Flow

C

V

Gas Inlet

V1

V2

V3

"Leak"

CDG


t

Vpq pV

Range: 10-8 Pa L/s … 10-1 Pa L/s

Gas flow meter FM3, PTB, Berlin


FM3

FLOW DIVIDERp0,V0

UHV-VESSELp1, V1

XHV-VESSELp2, V2

GAS FLOW

KP2

KP1

C02 C01

C2 C1

Accurate pressures from 10-9 Pa to 10-2 Pa


Primary standard CE3, PTB, Berlin


Relative uncertainties of pressures in primary standards

2E-021E-02

5E-034E-03

2E-03

3E-04

5E-05

2E-05

6E-06 5E-06

1E-06

1E-05

1E-04

1E-03

1E-02

1E-01

1E+00

1E-10 1E-08 1E-06 1E-04 1E-02 1E+00 1E+02 1E+04 1E+06

p in Pa

Re

lati

ve

Un

ce

rta

inti

es

(k=

2)

Mercury Manometer

Continuous expansion Series expansion

p atm



Valve 1

Testleak

Valve 3

Waterbath T=const.

Flowmeter

Valve 2

QMS

Traceability for leak measurements against vacuum



Testleak

Needle

CDG 133 Pa FS

CDG 133 kPa FS T1

T3

T2

T4

V1V2

V3

V = 5.1 cm³

V = 6.1 cm³Thermal insulation

Leak measurements against atmosphere




Conclusions from present state measurement standards

• Accurate vacuum gauge calibration is possible from 10-9 Pa to 105 Pa with uncertainties ranging from 0.001% up to 10%

• Accurate flow rate calibration is possible from 10-8 Pa L/s up to 0.1 Pa L/s against vacuum, 10-4 Pa L/s to 0.1 Pa L/s against atmosphere, with uncertainties ranging from 0.5% to 10%

• Only steady state conditions (constant pressure, equilibrium)

• In some ranges only some gas species (non-adsorbing) and pure gases

• Stable environmental conditions



A comment on traceability

• Whenever experimental results are being compared with a physical model or theory (pressures, mass flow rate, conductance, accommodation coefficient etc.), „true“ values of the experimental results are necessary.

•Characteristic for a true value is the number and the uncertainty of the value. Uncertainty is the interval in which the true value lies with a specified confidence limit (68%, 95%, …) around the given value.

•For true values and the respective uncertainty you need to have traceability to the SI.

•Traceability to the SI is given by complete calibration chain to a national primary standard for the given quantity (vacuum pressure, flow rate etc.)

•National primary standards are regularly checked internationally.



Gap from present state measurement standards to applications

To close this gap there will be a new project IND12 within the EMRP (European metrology research programme) funded by the EU.

Among others the tasks are:

Dynamic vacuum standard

Leak rate conversion from calibrated rate (for gas species, environmental conditions)

Joint research project (JRP) IND12:

Duration 3 years, begins Sept 2011, 2.8 M€


3. Four problems for rarefied gas dynamics

3.1 Dynamic vacuum standard

3.2 Predictable leak (flow) rate from secondary standards

3.3 QMS as process tool: Mass discrimination for partial pressure measurement

3.4 Physical relevance of standardized quantity S


3. Problems for rarefied gas dynamics

3.1 Dynamic vacuum standard


Dynamic vacuum pressures

How fast are vacuum gauges?

Which measurement principle is fast?

Which electronic is needed? Resolution?

Hystereses effects of gauges?

→ Establish well defined dynamic pressures to test gauges




Goal:Pressure reduction from 100 kPato 100 Pa within 1000 ms.Predictable on the time scale of ms and with an of u(p)/p(t) < 50% at all times.Extendible to fast pressure changes down to 0.1 Pa.Tests for optical method possible.

100000

600200

100

80

60

40

20

t in ms

p in

kPa

0.003 kPa



Idea for realization:Expand the gas from a small volume into a large one by a duct or orifice of calculable conductance.Calculate p(t), T(t).Compare with pressure and temperature measurement.

Conductance of fast valve >> conductance of orifice or duct

Fast valve must open within about 10 ms … 50 ms.



Rough estimate for necessary duct or orifice:

teptp 0 S

V

001.0 te

7s1

t ms145

1-Ls1.2L3.0

C

In the case of orifice and molecular flow:

mm8.4d



Problems to be solved:

Conductance must be known for any pressure between 100 kPa and 100 Pa, non-stationary flow.Which is best to calculate conductance for viscous flow: orifice or duct?Shall we generate conditions for choked flow?Temperature change, velocity of sound change, choked flow condition permanent?Can we calculate T(t) on the ms scale and test calculations?Can we calculate p(r) in V1?Is fast opening valve (DN40) available?If not, are their alternatives?Can we extend to lower pressures (< 100 Pa) and include desorption?



3.2 Predictable leak elements


Used for:

Calibrating leak detectors (linearity tests), partial pressure analyzers

Gases and gas mixtures for leaks

Most leak rate measurements are performed with helium, but the tightness for other gases, gas mixtures, even liquids is required.

+ Test conditions are different than the calibration conditions.

→ Establish procedures to convert the quantities.

Leak elements for industry


Secondary standard for leaks:

Permeation leak (see figure), temperature dependent, gas specific

Capillaries (less temperature dependent), crimped capillaries,

Porous plugs (sintered material)

Permeation leak type

Crimped capillary leak type

Gas flow



permeation porous capillary crimped

Temperature dependence

strong weak weak weak

Stability under rough cond

medium bad medium good

Operability under rough condition

good medium medium bad

Gas species flexibility

no yes yes yes

Rate easily changeable (T, p)

no yes yes yes

Predictable (gas, T, p)

medium medium yes yes

size large medium medium-large small



Choice for industry:

Permeation leak: stable, but not flexible, strong T-dependence, slow

crimped capillaries: small, large flexibility, work (good result) or do not work, fast, geometry is poorly defined



Possible equations of gas flow rate or conductance for capillaries (Tison, 1993)

2 Knudsen equation

1 Slip-flow equation

3 Linearized Boltzmann equationLoyalka, 19904 Guthrie, 1949, Steckelmacher, 1951:


Empirical, shows minimum

EmpiricalKnudsen successivemonotonic


Stuart Tison, Vacuum 44 (1993), 1171-1175F.Sharipov, Handbuch Vakuumtechnik, Bild 5.25



Stuart Tison, Vacuum 44 (1993), 1171-1175

p: 20 kPa … 2 MPa

Crimped capillary

Agreement slip-flow and exp fortuitous?

Residuals from Slip Flow Model



Stuart Tison, Vacuum 44 (1993), 1171-1175

Regularcapillary


1

018.1478.01

ln04.01

3

88.0

7.0tubepG

p: 20 kPa … 2 MPa

In Future?Sharipov, 2010:


To extend calculations from regular capillary to crimped capillary:

Geometry has to be well determined. Regular capillary: Uniformity of diameter? Advantage of crimped capillary may be that crimped part will dominate the result for mass flow.If geometry not well defined:Prediction from He calibration, pc, for other species, p? or …

Courtesy M. Bergoglio, INRIM

600 µm



alternatively nano „holes“ made by focused ion beams:


FIB hole L2 80 nm

y = 3,364E-14x - 8,749E-14

0,0E+00

5,0E-12

1,0E-11

1,5E-11

2,0E-11

2,5E-11

3,0E-11

3,5E-11

0 100 200 300 400 500 600 700 800 900 1000p in mbar

q in

mo

l/s



3.3 Predict mass discrimination of „high pressure“ quadrupole mass spectrometers


Mass discrimination in inlet stages for QMS

G as

To vacuum pum ps

To Q M SProcess G asC ham ber



3.4 Improve standards for pump speed measurements


Physical relevance of standardized S

Infinite volume

Equilibrium not disturbed by in- or

outflow

Vacuum pump

Vqt

VS

p

qS pV

Mechanical pumps

All vacuum pumps, but

Avn

Avn

qN 44

mkTAS 2

p is not isotropic in molecular range

The concept of pumping speed

intrinsic pumping speed

Infinite volume cannot be realized, reflected particles disturb equilibrium.



Feng Yu-guo and Xu Ting-wei, The appropriate test domes for pumping speed measurement, Vacuum 30 (1980), 377…382.

Concept of tubular test dome with diameter equal to pump inlet flange.

Ideal gauge position: Simulates infinitely large dome

Appropriate test dome: Ideal gauge position is independent of .

„Play“ with d/D and L/D to find appropriate test dome.



Deficencies of calculation of Yu-guo:

Only lower chamber simulated

Transmission probability calculated with 0.1% with a few 10 000 particles only

Optimum L/D could only be calculated with uncertainty of 10 %

Conclusion however:

L/D=1.5 is not optimal, lower values preferable.


Improving standard for pumping speed measurement


Repeat MC calculations to find better L/D and/or gauge position (d/D is less critical)

Extend simulations to transitional flow

Consider non-uniform across pump inlet area D.

Consider disturbed angular distribution of reflected particles inside pump


Conclusion

• Primary standards for pressure and low flow rates are available world wide

• These standards give traceability to SI. This traceability is needed whenever theory/simulation of rarefied gas flow is compared with experimental result

• Industry quite often needs conditions other than ideal

• Rarefied gas dynamics can help in designing new standards and predict behaviour of standards under industrial conditions: standard for fast changing pressure, predict leak elements, predict mass discrimination, design standard for physical S.


Dome of the Reichstag in Berlin

Documents

64th IUVSTA Workshop, May 16-19, 2011 Problems of vacuum metrology for industrial applications that call for solutions by rarefied gas dynamics Karl Jousten,