|
| 
Cape Verde Observatory
-VOCs-
| Observation |
| Category |
: |
Air sampling observation |
| Situation |
: |
ongoing |
| Time zone |
: |
UTC |
| Sampling |
| Sampling height |
: |
10 |
| Description |
: |
continuous |
| Sampling and analysis frequency |
: |
~ 1 hour time resolution for C2-C8 NMHC, DMS and o-voc including acetone, methanol, acetaldehyde. |
| Sampling environment |
: |
NMHC -mainly from anthropogenic emissions from fossil fuel combustion and natural gas production.
Acetone, acetaldehyde, methanol (oVOC) terrestrial anthropogenic sources from solvent use, industrial processing, and biomass burning, terrestrial biogenic from vegetation emissions.
The oVOC are also produced secondarily from NMHC.
ethene, propene, isoprene, acetaldehyde (and perhaps acetone) from the photochemical degradation of organic matter in the oceans.
DMS from phytoplankton in the oceans.
NMVOCs are removed from the atmosphere primarily by reaction with the hydroxyl radical OH and subsequent photo-oxidation of products ultimately to CO2 and H2O. A by-product of this oxidation (when NOx is present) is the formation of tropospheric ozone (O3) and other secondary pollutants including organic aerosols. Tropospheric O3 has been reported to be increasing both regionally and globally a result has been substantial endeavour to understand the role of longer range transport (e.g. transatlantic transport) of precursors such as CO; and NMVOC in determining O3 trends in remote environments. The strong chemical coupling between O3 and OH provides further impetus to understand what controls CO, NMVOC, and O3 in remote locations. Of particular significance are the remote tropical regions where the majority of CO and CH4 are chemically destroyed by OH. It is here where the atmospheric chemistry of air pollutants couples with climate drivers through the OH radical.
|
| Description for sampling analysis |
: |
Sampling
The air sample is pulled through the main sampling manifold (6 m of 1” Teflon followed by 6 m of 1” stainless steel, residence time 4.5 s) using the main sample pump and a smaller pump pulls the sample through 1/8” stainless steel for the last 1 m to the instrument.
Instrument used from Oct 2006 to April 2013
Agilent 6890 in combination with home built preconcentrator system
Water is removed from the sample using a condensation trap held at -27° C. Ozone is not removed from the air sample. The trap used in this system is packed with the graphitic carbon adsorbents Carbopack B and Carboxen 1000 in series, a combination not thought to be affected by O3 contamination [Koppmann, 2007]. In addition the ambient levels of O3 observed at CVAO are extremely stable on an hour to hour basis (although do vary seasonally) with an annual maximum of ~50 ppbV [Lee et al., 2009] and it is with levels above 90 ppbV that O3 scrubbers are recommended [Koppmann, 2007].
From August 2010 the stainless steel inlet was replaced with a 40mm diameter heated glass inlet followed by 6 m of 1” stainless steel, residence time 4.5 s) using the main sample pump and a smaller pump pulls the sample through 1/8” stainless steel for the last 1 m to the instrument.
Instrument used from May 2013 till ongoing
MARKES preconcentrator Thermal desorption unit and CIA8 Autosampler in combination with Agilent 7890 coupled with FIDs
Since June 1, 2013 a new commercial preconcentrator (MARKES UNITY) and CIA8 Autosampler in combination with Agilent 7890 coupled with Flame Ionisation Detection (FID) system has been used. UNITY contains a 2-stage Peltier cell, which uniformly cools the entire 60-mm sorbent bed to a minimum of -30°C.Dry air flows into the cold trap box creating a slight positive pressure and minimising ingress of water from the laboratory atmosphere.
The cold trap is packed with an appropriate series of sorbent (packing material in not revealed by company) that allows quantitative retention of compounds as volatile as ethene. Before sample is passed through the cold trap, water is removed by passing air through condensation trap immersed in a mixture of antifreeze ethylene glycol and water maintained at -30 deg Celsius and then through another new condensing chiller that is cooled by Sterling coolers and maintained at -40 deg Celsius.
Lines are purged for 5 minutes at the rate of 50 ml per minute before the start of preconcentration step. About 700 ml sample volume is acquired in 20 minutes sampling time with flow rate of 36 ml per minute. During preconcentration step, various flow path stages are displayed on screen.
Once all the target analytes have been collected and focused in the cold trap, the trap oven heats rapidly reaching rates in excess of 60°C/sec for the first critical stages of trap desorption and all analytes are desorbed in 50:50 split onto two PLOT alumina deactivated by sodium sulphate columns (one 50 m for NMHCs and other 10 m CP LOWOX for oxy VOCs ) kept in temperature programmed oven (Agilent 7890). Analytes from each column are fed into respective FIDs and output from detector is processed in Agilent chemstation software.
|
| Instrument and Analysis |
| Measurement method |
: |
Gas Chromatography (FID) |
| Current status and history of instrument |
: |
Since October 2006 and ongoing.
The dual channel GC-FID and optic injector heating is from Agilent, UK. The rest of the instrumental set-up is from a combination of commercial and home-built elements including chillers, pelter controls, valves etc.
Instrument used from May 2013 till ongoing
MARKES preconcentrator Thermal desorption unit and CIA8 Autosampler in combination with Agilent 7890 coupled with FIDs
Since June 1, 2013 a new commercial preconcentrator (MARKES UNITY) and CIA8 Autosampler in combination with Agilent 7890 coupled with Flame Ionisation Detection (FID) system has been used. UNITY contains a 2-stage Peltier cell, which uniformly cools the entire 60-mm sorbent bed to a minimum of -30°C.Dry air flows into the cold trap box creating a slight positive pressure and minimising ingress of water from the laboratory atmosphere.
The cold trap is packed with an appropriate series of sorbent (packing material in not revealed by company) that allows quantitative retention of compounds as volatile as ethene. Before sample is passed through the cold trap, water is removed by passing air through condensation trap immersed in a mixture of antifreeze ethylene glycol and water maintained at -30 deg Celsius and then through another new condensing chiller that is cooled by Sterling coolers and maintained at -40 deg Celsius.
Lines are purged for 5 minutes at the rate of 50 ml per minute before the start of preconcentration step. About 700 ml sample volume is acquired in 20 minutes sampling time with flow rate of 36 ml per minute. During preconcentration step, various flow path stages are displayed on screen.
Once all the target analytes have been collected and focused in the cold trap, the trap oven heats rapidly reaching rates in excess of 60°C/sec for the first critical stages of trap desorption and all analytes are desorbed in 50:50 split onto two PLOT alumina deactivated by sodium sulphate columns (one 50 m for NMHCs and other 10 m CP LOWOX for oxy VOCs ) kept in temperature programmed oven (Agilent 7890). Analytes from each column are fed into respective FIDs and output from detector is processed in Agilent chemstation software. Quantification is achieved by using multi-component standard gas mixture. Having seen the stability of FIDs, calibrations are performed once in a month since August 2012.
|
| Description of instrument |
: |
|
| Calibration |
| Current scale employed in the measurement |
: |
APEL REIMER 54 CYLINDER NUMBER 236351, NPL 30 COMPONENT MIXTURE 2008, CYLINDER NUMBER =D838940, NPL 30 COMPONENT MIXTURE 2012, CYLINDER NUMBER =D860619, NPL 30 COMPONENT MIXTURE 2014, CYLINDER NUMBER D910414
|
| Measurement calibration |
: |
Calibrations are performed once a week in triplicate using a standard cylinder (low ppbV level, Apel-Riemer Environmental Inc, Denver, US). The triplicate samples are then averaged and applied to the subsequent week’s data assuming a linear regression in detector response through zero. The calibrations are bracketed by nitrogen blank samples to check for contaminations in the system. The FIDs show very consistent calibration responses that vary by no more than 5% for any species. Measurement uncertainty for NMHC arises largely from the uncertainties associated with the calibrant gas (quoted as 5%), and also in sample volume calculation and peak integration inconsistency. Volumes of sample and calibrant are identical and so a cancellation of errors may be considered here.
Calibration for acetone, methanol and acetaldehyde are performed using permeation sources. Permeation tubes are weighed every three months and the linearity of responses with differing sample volumes was confirmed (R2 = 0.82 to 0.99). The permeation tube emission flow is diluted by nitrogen to generate user controlled concentrations of OVOCs in the range 4 - 20 ppbV. The response of individual OVOCs is consistent as a function of both sample volume collected and concentration generated over the period of observations to within ± 15%). Toluene was generated within the permeation source as a cross check to the cylinder standards for NMHCs. The average response to toluene (on the NMHC channel) from the permeation tube is then compared to the NMHC calibration of toluene using the calibration cylinder (low ppbV level, Apel-Riemer Environmental Inc, Denver, US) and all the OVOC responses were adjusted to the more frequent weekly NMHC calibration. This method allows for correction of any mass flow controller error in the dilution of the permeation tube flow within the dilutant nitrogen gas flow. Response factors (peak areas/ppbv) were applied to the data; these are calculated for periods between the 3-monthly weighings and calibrations by assuming a linear change in response with time.
The precision for determination of NMVOC concentrations that are close to the detection limit (often the case for some NMHCs at Cape Verde) is determined mainly from random error in the peak integration. Experimental assessment of the detector noise and response shows a power relationship (y = axb) relating concentration, x to the RMS standard deviation, y. Therefore when peak size becomes comparable to the magnitude of detector noise the relative error approaches 100%, but for larger peaks the relative error is closer to 1%. A different power relationship exists for each compound because it depends on the area/ppbV. These relative errors relate to very small absolute errors, which are calculated for each peak individually. For typical clean MBL mixing ratios of 50 pptV, the error for C2-C4 is 7%, >C4 = 20%, acetone = 20%, acetaldehyde = 35%, and methanol = 35%. For 5 pptV, the error for C2-C4 is 20%, > C4 = 20%, acetone = 30%, acetaldehyde = 40%, and methanol = 40%. The detection limits are 2.5 pptV for C2-C4 NMHC, 1 pptV for >C4 NMHC, 3 pptV for acetone, 18 pptV for acetaldehyde and 7 pptV for methanol.
|
| Scale and calibration(treasability) |
: |
MEASUREMENT SCALE: APEL REIMER 54 CYLINDER NUMBER 236351,
NPL 30 COMPONENT MIXTURE 2008, CYLINDER NUMBER =D838940
AND NPL 30 COMPONENT MIXTURE 2012, CYLINDER NUMBER =D860619
From Oct 2006 to 19 Feb 2009, AR 54 component cylinder was used then from 20 Feb till 30 March 2012, NPL 2008 was used and
from April 2012 to Dec 2013, NPL 2012 was used.
This revised version (submitted in July 14) is based on compiling all peak areas from Oct 2006 till Dec 2013 and all calibration factors during this period and applying monthly calibration factors on respective month's data. |
| Data Processing |
| Measurement unit |
: |
ppt |
| Data processing |
: |
Chromatograms are integrated manually by the operator and the closest in time reponse factor (areas/ppbv) calculated from the calibration data is applied to give the species' concentration.
Nov 2011 onwards Data processing:
Data from 19 Nov 2011 are automatic integrated using GCWerks software. Chromatograms from Agilent Chemstation software are imported into GCWerks software. Integration parameters (peak width and peak threshold) are set in the software and chromatograms are reviewed one by one. If peaks are not integrated properly then appropriate integration parameters are applied to particular cases. Files are generated in dat format and then exported to XL files.
|
| Processing for averaging |
: |
Processing for Hourly Data:
Output direct from instrument, a little over a 1 hour resolution.
Processing for Daily Data:
The data is averaged between 01/xxx 00:00 and 02/xxx 00:00 to give the daily data (approx~ 20 points).
Processing for Monthly Data:
The data from the instrument is averaged between 01/xxx and 01/yyy to give the monthly data. |
| Data flag |
: |
Error Flag = 0 Good data
Error Flag = 1 Reduced quality data
Error Flag = 2 Below detection limit
Error Flag = 3 Invalid or missing data |
| Data remarks |
: |
|
| Other Information |
| Scientific aim |
: |
NMVOC are useful proxys for the interpretation of measurements of species such as CO (and also CH4), sharing some common production and loss terms. Long-term measurements can be used as pointers towards seasonal changes of OH and for general characterisation of the air in this region. Measurements such as these particularly within the tropical regions are integral in the validation of the outputs of global atmospheric models. |
| Reference |
: |
Read, K. A., A. S. Mahajan, L. J. Carpenter, M. J. Evans, B. V. E. Faria, D. E. Heard, J. R. Hopkins, J. D. Lee, S. J. Moller, A. C. Lewis, L. Mendes, J. B. McQuaid, H. Oetjen, A. Saiz-Lopez, M. J. Pilling and J. M. C. Plane (2008), Extensive halogen-mediated ozone destruction over the tropical Atlantic Ocean, Nature, 453(7199), 1232-1235.
Read, K.A. et al, Intraannual cycles of NMVOCs in the tropical troposphere and their use for interpreting seasonal variability in CO, submitted to JGR, 2009
|
|
| submitted by University of York |
|
|
