Calibration%20of%20Atmospheric%20Hydrogen

1
Calibration of Atmospheric Hydrogen

• Armin Jordan
• GasLab
• Max-Planck-Institute for Biogeochemistry
• 07701 Jena, Germany
• 2nd HyCare Symposium, Laxenburg 19.12.2005

2
Outline

• Relevance of calibration for studying
  atmospheric trends of H2
• Problems of standard gas stability
• Calibration scale at MPI-BGC
• Method to generate reference mixtures precisely
• Uncertainties related to this method

3
trends H2 mixing ratio records at Cape Grim
Novelli et al. (1999)
Langenfelds et al. (2002)
4
NOAA-CSIRO Flask Air Intercomparison Experiment
Masarie et al. (2001) JGR 106 (D17), 20445
Masarie et al. (2001) Assessment of
atmospheric H2 trends using measurements from
different programs would be difficult to achieve
at present. The offset may be due in part to
the different internal calibration scales used by
both laboratories.
5
No common distributor of H2 reference standards

• Different calibration scales established by
• NOAA Novelli et al.1999
• CSIRO Francey et al. 1996 / Simmonds et al. 2000
  
• MRI Sawa et al. 2004
• NIES Machida 2004
• ALE Rasmussen and Lovelock (1983)
• Bonasoni et al. 1997

6
Drift in Luxfer tanks
H2 increase rates 3 5 ppb/yr
7
Drift in Luxfer tanks
8
H2 scale at MPI-BGC

1. Standard purchased from CSIRO 543 ppb
2. Dilution series of 15 diluted samples generated
   from a ? 800 ppb standard to characterize
   detector reponse
3. Calibration of four cylinders, 40 l stainless
   steel (2), 50 l steel, 50 l aluminium

9
Quality control record
10
H2 calibration residuals
11
H2 residual drift
working standard
12
H2 residual drift 2005
13
Storage in glas flasks
14
Intercomparison record MPI - CSIRO
15
Independent check of H2 calibration

• Uncertainty in stability of standard gas mixtures
  in cylinders demands independent method to check
  for tank drifts.
• This method needs to
• meet or exceed the precision level of analytical
  measurement
• amount of work that's not prohibitive of regular
  repetition (2-6 monthly)
• Approach production of calibration gas by
  precise one step dilution
• n(H2) / n(N2)
• Hydrogen filling of sample loop n(H2) pV/RT
• Dilutant gravimetric determination n (N2) m
  / 28.013 g/mol
• n (air) m / 28.974 g/mol

16
Procedure description (1)

1. Filling of sample loop with ultrapure hydrogen
2. Isolation of sample loop and flushing of valve
   and lines with nitrogen

17
Procedure description (2)

1. Connection to dilution tanks and evacuation of
   lines
2. Transfer of hydrogen to reference gas cylinder

18
Procedure description (3)
19
Results Reproducibility
Reproducibility 1.5 ppb (0.2 ) Deviation
from scale 13 ppb (2 )
20
sources of uncertainty

• limiting factors
• purity of hydrogen
• accuracy of sensors (temperature, pressure)
• volume uncertainty of sample loop
• non-ideal behaviour of hydrogen gas
• purity of dilutant
• loss or production of H2 at surfaces
• accuracy of balance

21
limiting factors (1)

• uncertainty
• purity of hydrogen
• gt 99.9999 lt 0.001
• accuracy of sensors (temperature, pressure)
• sample loop volume
• non-ideal behaviour of hydrogen gas
• purity of dilutant
• loss or production of H2 at surfaces
• accuracy of balance

22
limiting factors (2)

• uncertainty
• purity of hydrogen 99.9999 lt 0.001
• accuracy of sensors (temperature, pressure)
• n pV/RT
• ? T resolution ? 0.1 C, calibration checked
  with ? 0.03 Omega DP251 Precision
  Thermometer
• ? p resolution ? 1 mbar ? 0.1
  calibration with MKS Baratron offset 1.5 mbar
• sample loop volume
• non-ideal behaviour of hydrogen gas
• purity of dilutant
• loss or production of H2 at surfaces

23
limiting factors (3)
sample loop volume determination by
gravimetry filling of sample loop with high
purity water at 22.4 ? 0.1C uncertainty
? balance uncertainty m ? 0.05 mg (resolution
0.01 mg) lt 0.05 reproducibility m ? 0.05
mg lt 0.05 ? r H2O(22.4C) 0.9977
mg/µl lt 0.01 ? internal valve volume
2.46 µl ? 5 ? 0.1 µl dead volume of
Valco fitting ? 0.25 µl ? V ? 0.35 µl ? 0.1

24
limiting factors (3)
reproducibility with different sample
loops
25
limiting factors (4)

• uncertainty
• purity of hydrogen
• ? 99.9999 lt 0.001
• accuracy of sensors (temperature, pressure)
• n pV/RT ? T ? 0.1 C ? 0.03
• ? pressure sensor resolution1mbar ? 0.1
• sample loop volume (250-400 µl) ? V ? 0.3
  µl ? 0.1
• non-ideal behaviour of hydrogen gas
• P
• van der Waals constants of H2 a 0.2453 bar
  L-2 mol-1
• b 0.02651 L mol-1
• ? deviation of real gas pressure from ideal
  0.06

26
limiting factors (5)

• purity of dilutant
• comparison of two dilutant gases nitrogen and
  air
• no chromatographic blank ? below detection limit
  lt 15 ppb
• Gas purifying cartridges
• Nitrogen Aeronex cartridge 70KFI4R
  specification H2 lt 1 ppb
• air cylinder filled with Sofnocat 423 cartridge
  ? H2 ? 100 ppb
• transferred through 500 cc cartridge filled with
  450 g Sofnocat 514
• (H2 conversion rate gt 99 _at_ residence time of gt
  5 sec))

27
limiting factors (5)
purity of dilutant comparison of two
dilutant gases
Mean offset 3 ppb
28
limiting factors

• uncertainty
• purity of hydrogen
• ? 99.9999 lt 0.001
• accuracy of sensors (temperature, pressure)
• n pV/RT ? T ? 0.1 C ? 0.03
• ? p ? 1 mbar ? 0.1
• sample loop volume (250-400 µl) ? V ? 0.3
  µl ? 0.1
• non-ideal behaviour of hydrogen gas
• ? deviation of real gas pressure from ideal lt
  0.02
• purity of dilutant lt 0.2 - 0.5
• loss or production of H2 at surfaces
• accuracy of balance

29
limiting factors (6) loss or production of
hydrogen

• a) sorption effects on valve rotor polymer
• two different polymers tested
• Valcon E (polyaryletherketone/PTFE composite)
• Valcon M (hydrocarbon) impermeable for light
  gases

30
limiting factors (6) loss or production of
hydrogen

• a) sorption effects on valve rotor polymer
• two different polymers tested
• Valcon E (polyaryletherketone/PTFE composite)
• Valcon M (hydrocarbon) impermeable for light
  gases
• ? no significant difference

31
limiting factors (6) loss or production of
hydrogen
b) hydrogen production within the Luxfer
cylinder storage test of gas after filling in
an evacuated 6 L Luxfer cylinder ? drift rate
insignificant for standard mixing
experiment
drift rate 0.07 ppb/d
32
limiting factors (6) loss or production of
hydrogen
b) hydrogen production by Luxfer cylinders
new cylinders filled with synthetic air _at_ 3
bar
? drift rates _at_ 100 bar 0.2 - 20 ppb/d !
33
Results summary
? std.dev. 1.9 ppb
34
Uncertainty estimate

• potential offset uncertainty
• purity of hydrogen lt - 0.001
• accuracy of sensors
• ? T ? 0.1 C ? 0.03
• ? pressure ? 0.1
• sample loop volume (250-400 µl) ? m ? 0.5
  mg lt 0.02 ? r H2O(T) (H2O) lt 0.01
  
• ? internal valve volume ? 0.05
• ? Dead volume 0.05
• non-ideal behaviour of hydrogen gas
• purity of dilutant lt 0.2 ... lt 0.6
• loss or production of H2 at surfaces lt 0.02
• accuracy of balance 0.2 g / 700 g (balance
  resolution 0.1 g) lt 0.03

35
Uncertainty estimate

• potential offset uncertainty
• purity of hydrogen lt - 0.001
• accuracy of sensors
• ? T ? 0.1 C ? 0.03
• ? pressure ? 0.1
• sample loop volume (250-400 µl) ? m ? 0.5
  mg lt 0.02 ? r H2O(T) (H2O) lt 0.01
  
• ? internal valve volume ? 0.05
• ? dead volume 0.05
• non-ideal behaviour of hydrogen gas
• purity of dilutant lt 0.2 ... 0.6
• loss or production of H2 at surfaces lt 0.02
• accuracy of balance 0.2 g / 700 g (balance
  resolution 0.1 g) lt 0.03

36
Conclusions

• there is a need to improve atmospheric hydrogen
  data
• need to make different data sets comparable
• intercomparisons are extremely important
  especially
• because there is no international calibration
  scale
• prerequisite for setting up a scale is choice of
  adequate
• containers
• dilution method may provide a fixed calibration
  point that . allows to detect
  standard drifts
• could provide absolute mixing ratios
• relatively simple set-up, but very precise if
  critical . . parameters are
  thoroughly determined / controlled
• once set-up not very labour-intensive

37
NOAA-CSIRO Flask Air Intercomparison Experiment
Masarie et al. (2001) JGR 106 (D17), 20445

• Masarie et al. (2001)
• Assessment of atmospheric H2 trends using
  measurements from different programs would be
  difficult to achieve at present. The offset may
  be due in part to the different internal
  calibration scales used by both laboratories.
• Schmidt and Wetter (2002) ..the differences in
  absolute H2 levels of ... 20-40 ppbv, might also
  be attributed to possible differences in absolute
  calibration scales
• Simmonds et al (2000) we might expect
  significant differences in absolute calibration
  between the early and most recent hydrogen
  measurements..

38
Why care about atmospheric hydrogen?

• Hydrogen technology as clean energy source is in
  development.
• A change to hydrogen economy would significantly
  increase the atmospheric H2 burden. Consequences
  for the Earth's radiation balance and atmospheric
  chemistry could result from
• a reduction of atmospheric oxidative capacity ?
  increase of CH4
• lifetime (indirect greenhouse gas)
• conversion to water vapour in the stratosphere
  ? modifcation of the . HOx cycle ?
  depletion of stratospheric ozone
• Model scenarios predict mixing ratios of 700
  2300 ppb depending on different assumptions on
  economic change and leak rates.
• Trends of atmospheric hydrogen will also be
  driven by global climate change that affect
  source and sink processes.
• Prediction of these future trends requires a
  thorough understanding of the current atmospheric
  budget of hydrogen.

39
Sensors

• Temperature determination
• PT 1000 (resolution 0.1 C) calibrated with
  Omega DP251 Precision Thermometer
• Pressure determination 1 mbar
• Leybold LHXXX (resolution 1 mbar) calibrated
  with
• Balance
• CP8201-0CE (resolution 0.1 g)

40
present knowledge of global H2 budget (1)

• mixing ratios
• (Langenfelds et al.(2002), Simmonds et al.(2000),
  Novelli et al.(1999), Francey et al. (1998),
  Khalil Rasmussen (1990)
• budget
• Sanderson et al. 2003, HauglustaineEhhalt 2002,
  Novelli et al. 1999, Ehhalt 1999, Warneck 1988,
  SeilerConrad1987

41
sorption/desorption effects

• sample loop in touch with Valvo valve rotor
• H2 adsorbed to polymer desorbed again while
  transferred to 6 L cylinder
• two different polymers tested
• Valcon E (polyaryletherketone/PTFE
  composite)
• Valcon M (hydrocarbon) impermeable for light
  gases

42
test of Valco rotor polymer
43
Diluent check N2 or purified air
44
factors limiting accuracy of the method

• accuracy of sensors (temperature, pressure)
  and balance accuracy
• sample loop volume
• purity of dilutant
• no loss or production at surfaces

45
Drift in 6 l Scott Luxfer cylinder (synth. air,
3 bar)
? drift rates _at_ 100 bar 0.2 - 20 ppb/d !
46
sample loop volume
47
dilutant gas check

• nitrogen
• no chromatographic blank
• Aeronex cartridge H2 lt 1 ppb
• air
• cylinder filled with Sofnocat 423 cartridge
• transferred through cartridge filled with 450
  g Sofnocat 514

48
Determination of sample loop volume
2.46 µL

• volume determination
• filling with ultrapure water, 22.4 0.1, ?
  0.9977 mg/µL
• balance uncertainty 0.05 mg, reproducibility
  0.05 mg
• additional volume of Valco valve
• rotor path and valve bores 2.46 µL (0.65 -
  1 of loop volume)
• uncertainty 0.3 µL ( 0.1 rel.)