1. 1 Session 2 – Analytical Issues Mercury Speciation Workshop 330 Nantucket Blvd. Toronto, Canada M1P 2P4 Rev 1.10 Nov 2003

2. 2 Issues in Session Two 11Do 1130 & 1135 work in the Arctic ? 1 22What Species do denuders measure ? 2 33What does a Model 2537A measure ? 3 44Precision & Accuracy of method 4 55Effects of sodalime trap 5 66Calibration for RGM & HgP 6 77Denuder coating techniques 7 88RPF refill techniques 8

3. 3 Do the Model 1130 & 1135 Work in the Arctic?

4. 4 Caveats  The following is not based upon any actual arctic measurements  There is currently no reliable RGM calibration source that would work under those conditions  Tekran doesn’t go up there often  We don’t have an environmental test chamber

5. 5 Tests at Tekran  Tests were continuously run during the year it took to deliver first Model 1130-P prototypes (May 1998)  Initial tests used packed cartridges  Subsequent tests used thermal denuder  Tests were run using outdoor air in Toronto  Summer:+30° C, (typical moist summer air)  Winter: -20° C, (typical dry winter conditions)  Method worked well under full seasonal range

6. 6 Temp Dependent Denuder Variables  Diffusion coefficient of HgCl 2  Low temps could reduce capture efficiency  Actual gas volume of sample (p,v)  Affects residence time in denuder  Low temps increase residence time  Gas is pre-heated by impactor and denuder inlet  Approx inlet volume is 80 ml  Heating residence time is ~0.45 sec. (at 10 l/m)  Actual gas temp at denuder inlet will not be close to -40° C

7. 7 Main Difference with Arctic Air  The major difference between arctic air and temperate air is the moisture content of the air  After heating, the air will be very dry  Does the 1130 capture RGM under low humidity conditions?

8. 8 Tests by Frontier Geosciences  Two prototype 1130’s purchased by Florida DEP were extensively tested by Eric Prestbo in 1998  Contract funded by Tom Atkeson of FL DEP  Formal report never issued  Results were presented at several conferences and incorporated into Landis et. al. (ES&T, 2002)

9. 9 Relevant FL DEP Tests  Tests were done using:  Nitrogen from a dewar  Ambient air  Sample gas created from a dewar was extremely dry  No significant differences in capture efficiency were detected between very dry gas sample and ambient samples

10. 10 Arctic Tests  Alert: Bill Schroeder ran two different systems in parallel  Arctic Pyrolyzer  No inlet filter  Large pyrolyzer, 900 °C with lengthy residence time  Feeds into a Model 2537A  Expected to yield total atmospheric mercury (TAM) (both gaseous and particulate forms)  Model 1130/1135/2537A

11. 11 Results  During non-depletions  Fairly good agreement between the methods  Hg 0 (GEM) slightly lower in 1130/35 system  During MDEs  Some differences, typically 20-30% with pyrolyzer being higher (personal communication S. Steffen, B. Schroeder)  Shows that there is no gross failure of the 1130/1135 method in the Arctic, even during MDEs

12. 12 When No RGM or TPM Present Pyro TAM ~0.2 ng/m 3 higher than GEM measured through 1130/35  There may be slight contamination in pyrolyzer system  Material in pyro chamber  Downstream heated line  There may be scavenging in 1130/35  RPF, downstream filter or internal lines  Heated line or fittings around PM

13. 13 Possible Reasons - GEM  Should review Arctic QA/QC data to determine:  Do TAM values decrease after cleaning of the pyrolyzer and downstream heated line & fittings?  Have zero checks and manual injection tests of entire pyrolyzer system revealed any problems?

14. 14 Possible Tests - GEM  Could perform external zero and Hg 0 addition system test on 1130/35  Requires 10 l/m zero air source  Manual injection source & syringe (large volume syringe: 100-250 µl)  Must first perform accurate flow rate tests on both 2537A and 1130 pump!  Required to determine what fraction of injected Hg will disappear through PM  Tricky test !

15. 15 Other Reasons for Differences  Two devices are measuring slightly different things:  Pyro measures total particulate loading  1130/35 measures fine fraction particulates (< 2.5 µm)  Difference could be legitimate  Some mercury may be in coarse particulate fraction  Could also be losses of RGM on inlet surfaces  Dirty impactor surfaces  Insufficient heating

16. 16 What Species Do Thermally Regenerated KCl Denuders Measure?

17. 17 Mercury Chloride - HgCl 2  Compound most often used as a surrogate for “RGM”  Reasonable choice since it’s believed to be created by many industrial sources  Believed to be the bulk of RGM loadings

18. 18 Mercury Chloride - HgCl 2  Extensively tested by Tekran  Major pain to work with  Extremely “sticky”  Regenerable KCl media had >98% capture efficiency  Initial work with KCl coated quartz chips  Subsequently validated using denuders

19. 19 Mercury Iodide – HgI 2  Originally tested as a substitute for HgCl 2  Hoped that it would be easier to work with  Turned out to be exactly as much of a pain  Behaved the same as HgCl 2 with a capture efficiency: >98%

20. 20 Monomethyl mercury chloride – CH 3 HgCI  Tested by Jonas Sommar (Sweden)  Tests pre-dated thermal method  Used tubular denuder with wet extraction & digestion  Reported a capture efficiency: >94%  (In comparison to 98% for HgCl 2 )  Don’t have a publication reference

21. 21 Further Testing Needed  EPA NERL (Matt Landis, Bob Stevens) are planning on testing a wide variety of mercury compounds for capture efficiency

22. 22 What does a Model 2537A actually measure ? Is there a simple answer ?

23. 23 What We Know  The Model 2537A will respond to HgCl 2 that is presented to the cartridges  Ontario Hydro, 1995 (?)  Had to bypass all front end components to get HgCl 2 into the cartridges  We can’t claim that the 2537A is an elemental Hg analyzer

24. 24 Transport Issues  HgCl 2 does not transport well through sample lines or filters  Will stick onto the materials  May come off later depending on factors such as:  Temperature  Humidity  Composition of sample  ERG & EPA (1997-98)

25. 25 Under Arctic Conditions  Model 2537A functions basically as an elemental Hg analyzer  Evidence: MDE’s were originally discovered by Env. Canada  2537A recorded very low values  We now know that lots of RGM was present during many of those events

26. 26 Precision & Accuracy Caveats for duplicate instrument runs

27. 27 Tests with Prototype 1130’s  Our outdoor air was brought in through a 4” plastic pipe using a 700 l/m blower  Both units ran from same pipe  We do not claim that sample contained true outdoor RGM levels  Got good agreement between units

28. 28 Side by side Tests  Indoor air was simply taken in by the two 1130’s mounted side by side  Not as good agreement for RGM. Why? 1.Denuders more precise at lower concentrations 2.Outdoor air works better than indoor air 3.Sampling wasn’t from a common manifold

29. 29 Early Results - Toronto

30. 30 Early Results - Toronto

31. 31 Duplicate Instruments  Running two instruments side by side is not trivial  Must be sampling exactly the same air  Even minor differences in location will have a large impact  Have seen this with 2537A for years  Much more of an effect with RGM/HgP

32. 32 Caveats  Sampling manifold issues  First instrument in chain will contaminate sample manifold when blowing back zero air during desorption  Precise syncing will help, but not eliminate this problem  Transport of RGM & HgP along manifold  Effect of intrusion by sampling inlets  Scavenging/contamination  Isokinetic sampling if monitoring particulates

33. 33 Caveats  Instrument flow rates are critical!  Must calibrate all 2537A and 1130 MFM’s before running any tests

34. 34 Effects of Sodalime Trap

35. 35 Effects of Sodalime Trap  Works well in most cases  Bad sodalime can either scavange or augment mercury from sample  Good sodalime can go bad simply by being stored after opening  Must be kept above dew point of sample air

36. 36 Calibration for RGM & HgP ?

37. 37 Calibration - RGM  Very difficult to get long term stability in lab  Even harder in the field  Likely to be used as a check, rather than as an actual calibration source

38. 38 Calibration - HgP  Even more difficult !  Some issues:  Picogram amounts required  Controlling size distribution  Sample introduction

39. 39 Denuder Coating Techniques ?

40. 40 Two techniques  Original super-saturated method  EPA method

41. 41 RPF Refill Techniques ?

42. 42 RPF Standard Technique  There is none !

43. 43 End of Session 2