1. Measuring the specific surface area of snow using methane adsorption and IR reflectance -- difficulties and some progress -- Florent Domine Laurent Arnaud Carlo Carmagnola Nicolas Champollion Anne Dufour Frédéric Flin Jean-Charles Gallet Bernard Lesaffre Samuel Morin Ghislain Picard Takuvik joint international Laboratory, Québec Glaciology Laboratory, Grenoble Météo France, Grenoble Glaciology Laboratory, Grenoble Météo France, Grenoble Norwegian Polar Institute, Tromsø Météo France, Grenoble Glaciology Laboratory, Grenoble

2. Specific Surface Area (SSA)  Central physical and chemical variable Determines : Currently known range of values: light scattering and albedo radiation e-folding depth amounts of adsorbed gases rate of surface chemical reactions 0 50 100150200 250 m 2 kg -1 Fresh snow Decomposing particles Fine-grain snow Faceted crystals Depth hoar Refrozen snow

3. Specific Surface Area (SSA) Thousands of values have been measured using: CH 4 adsorption NIR or SWIR reflectance X-Ray tomography Stereology … Claimed Precision of a given method is 5-12% Claimed accuracy a given method is 10-20% Do all methods agree with each other ? Not always What are the various sources of error ?The method itself and snow sampling

4. Objectives of this talk Detailed presentation of the CH 4 adsorption technique Reliability of the method and comparison with X-Ray tomography Comparison between the DUFISSS and ASSSAP instruments Conclusions on possible sources of errors and artifacts How to improve the reliability of SSA measurements Detailed presentation of the SWIR reflectance method using the DUFISSS instrument  Focus of this presentation is on current difficulties

5. Specific Surface Area (SSA) Currently known range of values: 50 – 223 m 2 kg -1 for dry fresh snow 15 – 90 m 2 kg -1 for decomposing grains 10 - 40 m 2 kg -1 for fine-grained snow 8 - 45 m 2 kg -1 for faceted crystals 7 - 22 m 2 kg -1 for depth hoar 2 – 25 m 2 kg -1 for rerfrozen snow Early measurements: 200 - 1300 m 2 kg -1, fresh snow, Adamson 1967 7770 m 2 kg -1 Jellinek 1967 57 m 2 kg -1 fresh snow, Chaix et al. 1996 200 - 1300 m 2 kg -1, fresh snow, Adamson 1967 7770 m 2 kg -1, Jellinek 1967 57 m 2 kg -1, fresh snow, Chaix 1996 N 2 adsorption at 77 K CH 4 adsorption at 77 K  Measuring snow SSA using gas adsorption is delicate, and CH 4 is better than N 2

6. Principle of gas adsorption method

7. Can any transformation take place in the sample during those operations ? 77 K l-N 2 260 K First step : snow sampling

8. Before expansion: n 1 moles of gaz in V intro n 1  P 1 N ads = n 1 - n 2 n 2  P 2 After expansion: n 2 moles of gaz in V intro + V exp  adsorption isotherm : N ads = f (P 2 ) Liquid N 2 snow TcTc ThTh V intro V expansion P ThTh TcTc expansion ThTh TcTc ThTh TcTc P Second step : Measuring the isotherm

9. Liquid N 2 snow TcTc ThTh V intro V mc P V mh N ads = n 1 - n 2 After n increments of adding CH 4 in V intro :  Obtain adsorption isotherm N ads =f(P 2 )  accuracy of volume measurement is critical

10. BET hypotheses (Brunauer, Emmet, Teller, 1938) First layer adsorbs with enthalpy Q a Subsequent layers adsorb with enthapy Q L = enthalpy of condensation Principle of gas adsorption method QaQa QLQL

11. B.E.T. Equation, linear section 0.07 - 0.22 P/P 0 SSA of sample, m 2 kg -1 Analysis of adsorption isotherm: BET transform Surface area of 1 molecule Surface area of sample, m 2 mass of sample, kg (3 h of work) CH 4 : 19.18 Å 2 (Chaix et al., 1996)

12. n 1  P 1 n 2  P 2 Liquid N 2 snow TcTc ThTh V intr o V expansion P ThTh TcTc expansion ThTh TcTc ThTh TcTc P Why CH 4 works better than N 2 Accuracy depends on (P 1 - P 2 )/P 1 N 2 : P 0 =P sat  1000 hPa  (P 1 - P 2 )/P 1 low CH 4 : P 0 =P sat  13 hPa  (P 1 - P 2 )/P 1 high Kr : P 0 =P sat  1 hPa  (P 1 - P 2 )/P 1 very high Precision: 6 % Accuracy: 12 % Constraints on pressure sensor and field use

13. Can we check the validity of the CH 4 adsorption method ? Kerbrat et al. (2008) ACP 8, 1261. Compare SSA from CH 4 adsorption and from X-Ray tomography Fig. 5. The correlation between adsorption measurements and tomography was found to be SSA μCT =1.03(±0.03) SSA BET. Swiss comparison Both methods agree within 3%

14. Possible artifacts Formation of amorphous ice with super high SSA 77 K 260 K H2OH2O Adsorption of CH 4 onto container surfaces Evidenced and corrected in: Domine et al. (2007) JGR, F02031. Correction up to 20% in low SSA samples Jellinek and Ibrahim (1967) SSA= 7770 m 2 kg -1

15. Comparison of various SSA methods at La Grave, French Alps, 3100 m a.s.l., 1 st and 17 th April 2009 Experiment coordination: Ghislain Picard

16. Sample preparation for X-Ray tomography: Frédéric Flin

18. Sampling vials for CH 4 adsorption

19. Comparison CH 4 adsorption - X-Ray Microtomography in Grenoble X-Ray tomography values are 10-25% lower than CH 4 adsorption values Lab measurements of SSA from samples taken in pits at La Grave French comparison

20.  From our point of view, it is today difficult to resolve the discrepancy between both methods Differences can be due to : The methods themselves X-Ray Tomography CH 4 adsorption Image treatment protocolIntrinsic approximations The sampling protocol X-Ray Tomography CH 4 adsorption Thermal cycling Making the sampleTaking the sample

21. Domine et al., CRST 2006 CalculationsMeasurements Another illustration of SSA measurement difficulties: Measuring the SSA of snow using IR reflectance

22. Snow DUal Frequency Integating Sphere for SSA measurement Laser diode 1310 nm Integrating sphere InGaAs Photodiode Signal, mV Reflectance Standards Gallet et al. (2009) The Cryosphere, 3, 167-182

23. DUFISSS : SSA - reflectance calibration at 1310 nm Measure (1) SSA by CH 4 adsorption and (2) Reflectance with the sphere

24. La Grave intercomparison, 1 st and 17 th April 2009

26. Conclusion on CH 4 – DUFISSS intercomparison Essentially good agreement Expected since DUFISSS has been calibrated with CH 4 adsorption

27. Profilers of specific surface area (POSSSUM & ASSSAP) InGAs photodiodes Si photodiode Laser 635 nm Laser 1310 nm Calibrated with CH 4, DUFISSS and radiative transfer modeling

28. POSSSUM Distance measurement Electronics with pre-amplifier and laser power supply Optical module Bore and its motor Winch (25m-long cable) Profiles down to 20 m depths

29. La Grave intercomparison, 1 st and 17 th April 2009

31. 1 - POSSSUM bore hole2 - POSSSUM measurements La Grave intercomparison, 1 st and 17 th April 2009 On 17 th April 3 – Dig pit right on borehole 4 – DUFISSS measurements

32. La Grave intercomparison, 1 st and 17 th April 2009

33. Conclusion on CH 4 – DUFISSS - POSSSUM intercomparison Essentially good agreement POSSSUM has been calibrated with CH 4 adsorption and DUFISSS But 0-15% differences observed Differences could be due to : Lateral variations in snow properties Modifications of snow properties after surface has been cut

34. ASSSAP : Alpine Snow Specific surface Area Profiler Smaller version of POSSSUMSSA profiles down to depths of 2 m Used with DUFISSS at Summit, Greenland

37. 10 May stratigraphy

41. T= - 25°C T= - 45°C H20H20 Small ice crystal condensation ??

45. Conclusion on DUFISSS – ASSSAP comparison at Summit Preparing a sample for SSA measurements perturbs the snow Values obtained with 2 intercalibrated methods can differ by up to 70 % Minimizing the time between sample preparation and measurement may also be critical At Summit, being able to examine the sample proved critical  a thoroughly tested sampling protocol is critical

46. General conclusion Many methods exist to measure snow SSA These methods can be compared and intercalibrated in the laboratory In the field, sample preparation may be the main cause of differences between methods  efforts at comparing sample preparation methods appear essential