Robert Crampton Ph.D Brent Olive PH.D, Don Gamelis PH.D. Argos Scien7fic

Transcription

1 Using Spectral Averaging and Signal Processing to Leverage Exis7ng Short- term Fence Line UV Spectroscopic Data to Retrieve Accurate, Long- term Gas Concentra7ons to Meet New Monitoring Goals Robert Crampton Ph.D Brent Olive PH.D, Don Gamelis PH.D. Argos Scien7fic

2 What is UV DOAS Spectrographic technique for measuring atmospheric components in situ.

3 Ac7ve vs. Passive

4 Principles of DOAS The Beer-lambert law OD i = σ i c i l OD i = optical density, σ i = absorption cross-section, c i = concentration, l = pathlength (for species i)

5 Advantages of DOAS Overlapping absorp7on structures due to different species can be separated Species not an7cipated can be measured later. Spectral fingerprint Presence of Gases verified Warning against unexpected absorbers (residual) Mostly Immune against con7nuous (broad band) ex7nc7on due to e.g. aerosol or molecules High sensi7vity, since many trace gas lines (bands) are used. No analy7c cost associated with the sample (no lab) And Low maintenance costs

6 Disadvantages Atmospheric turbulence induce intensity varia7ons in the spectra. Upfront costs considerable k Only a limited number of molecules have suitable absorp7ons in UV and visible region. Rain, snow, fog and clouds make measurements impossible due to the strong ayenua7on in the UV- Visible region. Path Average concentra7on vs. Peak concentra7on

7

8 Some Gases/features

9 Can detect a wide range of UV absorbing chemical compounds Acrolein Acetaldehyde Ammonia Benzene 1,3-Butadiene Carbon disulfide Formaldehyde Hydrogen sulfide Nitric oxide Nitrogen dioxide Ozone Phenol m-, o-, p-xylene Styrene Sulfur dioxide Toluene..and more

10 Some Example Spectra

11 Hardware

12 Data Trail Raw Spectra - Background - Sample Absorb7on Concentra7on vs Time

13 Deuterium Source Raw Spectrum 4000 Raw Data - Deuterium Source 3500 O3 Benzene Toluene Xylenes NO NO2 SO2 Formaldehyde CS2 CLO2 Counts Acrolein 1-3 But NH3 SO2 single beam H2S Nanometers

14 Xenon Source Raw Spectrum Raw Data - Xenon Source Formaldehyde CS2 Chlorine Dioxide SO Toluene Xylenes 800 Counts 600 Benzene Single Beam 400 O nanometers

15 Background Spectrum Collect a data spectrum in the atmosphere when the target gas is not present. Define this as the background spectrum. Single Beam Raw Spectrum - Background Signal Strength Wavelength (nm)

16 Sample Spectrum Collect a data spectrum in the atmosphere when the target gas is present. Define this as the Sample spectrum. Single Beam Raw Spectrum Sample Spectrum Signal Strength Wavelength (nm)

17 Absorbance Spectrum Subtract the logarithms of the two spectra. This resulting spectrum is defined as an absorbance spectrum. Absorbance Spectrum Absorption Wavelength (nm)

18 Concentra7on 7me series Chalmette Air Monitoring (SO2) PPB :31 12:15 3:05 6:09 8:50 11:32 Time 14:54 6:45 9:58 12:41 3:22 Ambient SO2 10 Min Limit 24 Hour limit Annual Limit

19 Limits of Detec7on How do you know you have a detec7on? Spectral Feature of Gas larger than the noise 2x to 6x depending on false posi7ve vs. false nega7ves - level of cau7on Absorbance spectrum correla7on to ref spec.

20 Current Use UV DOAS Fence- line 5 min 7me resolu7on 5 bbp Benzene LOD Detect Peaks and Leaks Use with Wind data to pinpoint sources

21 Absorbance Spectra Various concentra3ons Benzene, 5 min Absorbance min 0 ppb 5 min.6ppb 5 min 1ppb 5 min 1.8 ppb 5 min 6ppb Nanometers

22 New Monitoring Goals Longer term exposures Lower Levels of detec7on Required Fine Time resolu7on not needed EPA Refinery Rule - 1 ppb Benzene - 2 week sample 7me

23 Absorbance Spectra Various concentra3ons Benzene, 40 min Absorbanc e min 0 ppb 40 min.6 ppb 40 min 1ppb 40 min 1.8 ppb 40 min 6ppb Nanometers

24 Requirements UV system has to save spectra for each small 7me frame Signal Strength changes need to be accounted for Changes in Pressure cause differen7al Oxygen absorp7on which need to be accounted for

25 Theory vs. Reality Limit of detec7on improvement 1/sqrt N 100 7mes sample 7me implies 1/10 L.O.D

26 Theory vs Data RMS Noise Abs Noise Data Theory Sample Time Min

27 Why not as good as Theory Signal strength fluctua7on O2 Absorp7on Atmos Pressure change Missing /bad data Advanced algorithms can get closer to Theory

28 Next Genera7on Argos UV - More thermally stable spectrometer/beyer detector array - 5 min Detec7on limits for benzene in the parts- per- trillion (Based on EPA's ETV methodology for determining detec7on limits) - Detec7on limits generated using field data - Final independent valida7on study completed in Summer 2015

29 New Spectrometer New Spectrometer ABS ppb Benzene 5 min Noise Nanometers

30 Conclusions Exis7ng UV fence- line monitors can meet or exceed new requirements BeYer Time resolu7on 1 ppb LOD Benzene much less than 14 day 7me resolu7on. Several Compounds can be measured Can average spectra from a subset of wind direc7ons much beyer source appor7onment. New Spectrometer promises to deliver 1 ppb L.O.D with 5 min scans