evaluation of the thermo scientific model 2025 sequential … · evaluation of the thermo...

36
Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (<2.5 μm) and Coarse (2.5 – 10 μm) Particulate Matter in the Alberta Oil Sands Region (AOSR) for Trace Element Determination Final Pilot Feasibility Report Prepared For Wood Buffalo Environmental Association #100 – 330 Thickwood Blvd. Fort McMurray, Alberta T9K 1Y1 April 11, 2012 Matthew S. Landis, Integrated Atmospheric Solutions, LLC, Raleigh, NC, USA Eric Edgerton, Atmospheric Research & Analysis Inc., Cary, NC, USA Joseph Graney, SUNY Binghamton, Binghamton, NY, USA

Upload: others

Post on 11-Mar-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

Evaluation of the Thermo Scientific Model 2025 Sequential

Dichotomous Sampler for the Collection of Fine (<2.5 µµµµm) and Coarse

(2.5 – 10 µµµµm) Particulate Matter in the Alberta Oil Sands Region

(AOSR) for Trace Element Determination

Final Pilot Feasibility Report

Prepared For

Wood Buffalo Environmental Association

#100 – 330 Thickwood Blvd.

Fort McMurray, Alberta T9K 1Y1

April 11, 2012

Matthew S. Landis, Integrated Atmospheric Solutions, LLC, Raleigh, NC, USA

Eric Edgerton, Atmospheric Research & Analysis Inc., Cary, NC, USA

Joseph Graney, SUNY Binghamton, Binghamton, NY, USA

Page 2: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

2

Table of Contents

1. INTRODUCTION .................................................................................................................. 3

2. METHODS ............................................................................................................................. 4

2.1. Sequential Dichotomous Samper Filter Collection ............................................................................... 4

2.2. X-ray Fluorescence Analysis................................................................................................................... 8

2.3. Filter Sample Extraction for Multi-Elements Quantification.............................................................. 8

2.4. Multi-Element ICPMS Analysis ............................................................................................................. 8

2.5. Stable Lead Isotope Analysis .................................................................................................................. 9

2.6. Statistical Analysis ................................................................................................................................... 9

3. RESULTS AND DISCUSSION ............................................................................................. 9

3.1. 2010 PM Mass Concentrations ............................................................................................................... 9

3.2. 2011 PM Mass Concentrations ..............................................................................................................14

3.3. ED-XRF Analysis ...................................................................................................................................18

3.4. DRC-ICPMS Analysis ............................................................................................................................18

3.5. Stable Lead Isotope Analysis .................................................................................................................22

4. CONCLUSIONS ................................................................................................................... 26

5. RECOMMENDATIONS ..................................................................................................... 26

5.1. Dichotomous Sampler Field Operation ................................................................................................26

5.2. Dichotomous Sampler Sample Analysis ...............................................................................................26

5.3. Addition of PM Carbon Measurements ...............................................................................................27

5.4. Expansion of the Dichotomous Sampler Network ...............................................................................27

6. RESPONSE TO INITIAL REVIEW COMMENTS ......................................................... 28

6.1. Housing the Sampler Inside the Monitoring Station ...........................................................................28

6.2. Acceptance of Dichotomous Sampler....................................................................................................28

6.3. Movement of Dichotomous Sampler .....................................................................................................29

6.4. Sampling by Difference versus Dichotomous Sampler........................................................................30

7. PATH FORWARD ............................................................................................................... 30

7.1. Evaluation of Data Completeness for the Dichotomous Sampler versus Existing FRMs ................30

7.2. Comparison of Dichotomous Sampler and Existing FRM Mass Data ...............................................31

7.3. Comparison of Dichotomous Sampler and Existing FRM Metals Data ............................................34

8. ACKNOWLEDGEMENTS ................................................................................................. 35

9. REFERENCES ..................................................................................................................... 35

Page 3: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

3

1. Introduction

The Wood Buffalo Environmental Association (WBEA) is a multi-stakeholder, community-based,

not-for-profit association located in Fort McMurray, Alberta, Canada. The WBEA airshed covers

70,000 sq km in northeastern Alberta and includes the Athabasca Oil Sands Region (AOSR).

WBEA measures air quality within the airshed at 15 monitoring sites, from Fort Chipewyan in the

north to Anzac in the south. There are both natural (forest fires) and anthropogenic pollution

emission sources in the AOSR. The anthropogenic air pollution sources are mainly composed of oil

sand mining, oil extraction facilities, oil refining, tailings ponds, and transportation sources. Air

quality management decisions require information on the sources contributing to air pollution to

develop effective air pollution control strategies. Assessing the local and regional scale

contributions of air pollution sources on particulate matter (PM) air quality in the AOSR is currently

limited due to an insufficient amount of PM speciation data required for a comprehensive source

apportionment analysis.

The regulation and monitoring of PM in ambient air focuses on aerosols that can be inhaled into the

respiratory system (e.g., aerodynamic diameter <10 µm (PM10)). Researchers generally recognize

that these aerosols may cause adverse health effects. Atmospheric aerosols commonly occur in two

distinct size modes as shown in Figure 1: the fine (<2.5 µm) mode and the coarse (2.5-10.0 µm)

mode (Seinfeld and Pandis, 2006). The fine or accumulation mode (also termed respirable

particulate matter) is typically attributed to direct emissions from anthropogenic sources or growth

of particles from the gas phase and subsequent agglomeration, while the coarse mode is primarily

made of mechanically abraded or ground particles. Particles that have grown from the gas phase

(either due to condensation, transformation, or combustion) occur initially as very fine nuclei (~

0.05 µm). These particles tend to grow rapidly to accumulation mode particles around 0.5 µm,

which are relatively stable in the air. Because of their initially gaseous origin, particle sizes in this

range include inorganic ions such as sulfate, nitrate, ammonia, combustion-form carbon, organic

aerosols, metals, and other combustion products. Coarse PM, on the other hand, is produced mainly

by mechanical forces such as crushing and abrasion. Therefore, coarse particles typically consist of

finely divided minerals such as oxides of aluminum, silicon, iron, calcium, and potassium. Coarse

particles of soil or dust mostly result from entrainment by wind or from other mechanical action.

Since the size of these particles is normally greater than 2.5 µm, their retention time in the

atmosphere and transport scales are generally substantially shorter than PM2.5.

Page 4: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

4

Figure 1. Particulate Matter Size Mode Distribution.

This report describes a pilot study aimed at evaluating the operation of an automated dichotomous

sampler for the measurement of fine and coarse mode PM and the suitability of the collected filter

samples for trace element quantification for source apportionment modeling. Source apportionment

is the estimation of the contributions to the airborne pollutant concentrations that arise from the

emissions of natural and anthropogenic sources (Hopke, 2009). Statistical data analysis tools called

“receptor models” are applied to evaluate the variance structure of a monitoring data set at a

specific location (receptor site) to reduce its dimensionality and elicit information on the sources of

air pollution. The models utilize unique tracer species combinations from different source types to

quantify the relative importance of sources on observed pollutant concentrations.

The main objectives of this pilot feasibility study are to:

a. Evaluate the operation of the ThermoScientific Model 2025 Sequential Dichotomous PM

Sampler in the AOSR.

b. Investigate appropriate extraction and inorganic analysis methods for dichotomous sampler

Teflon filters.

c. Evaluate if 24-hour sample collected fine and coarse mode mass in the AOSR is sufficient

for quantification of a sufficient number of relevant tracer species for application in routine

PM air quality monitoring and for use in source apportionment analysis.

2. Methods

2.1. Sequential Dichotomous Sampler Filter Collection

A ThermoScientific (Franklin, MA) model 2025 sequential dichotomous particulate matter (PM)

sampler (a U.S. EPA designated Federal Equivalent Method for PM2.5) was installed in the AMS-1

site shelter at Fort McKay in late 2009 (Figure 2). After several test runs the sampler began

continuous operation collecting daily samples on February 22, 2010. The sampler uses a PM10

impactor inlet (Figure 3) to remove particles with an aerodynamic diameter greater than 10 µm, and

an internal virtual impactor

Page 5: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

5

Figure 2. AMS-1 Fort McKay Installation of Sequential Dichotomous Sampler.

Figure 3. AMS-1 Fort McKay Sequential Dichotomous Sampler Inlet.

Page 6: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

6

to separate the coarse (2.5 - 10.0 µm) and fine aerosols (<2.5 µm) onto separate filters (Loo and

Cork, 1998; Figure 4). The virtual impactor accelerates incoming PM10 aerosols using a jet to

impart sufficient momentum that they resist the lateral sheer of the Q2 major flow and traverse into

the receiving jet and are captured onto the PMcoarse filter.

The virtual impactor results in the collection of all coarse mode particles from the Q0 total flow and

the fine mode particles in the Q1 minor flow (Figure 4) on the coarse filter. As a result, the fine

mode and corrected coarse mode concentrations (mass and trace elements) are adjusted for this

artifact using equation 1 and 2, respectively.

= (1)

= ∗ !"# $ (2)

Where: CFine= Concentration PM2.5 (µg m-3

)

CCoarse = Concentration PMCoarse (µg m-3

)

VFine= Volume of Air Through PM2.5 Filter (m-3

)

VCoarse = Volume of Air Through PMCoarse Filter (m-3

)

VTotal = Volume of Air Through Sampler (m-3

)

MFine= Mass on Fine Filter (µg)

MCoarse = Mass on Coarse Filter (µg)

Measurement Technologies Laboratories (MTL; Minneapolis, MN) 47 mm Teflon membrane filters

with Teflon support rings were procured, pre-weighed, installed into filter cassettes/magazines,

shipped to WBEA, received from WBEA, and post-weighed by Atmospheric Research & Analysis,

Inc (ARA) at its laboratory in Morrisville, NC. Each filter magazine was loaded with 15 filter

cassettes, enough for two weeks of unattended sampling and a field blank. The standard

ThermoScientific stainless steel filter support screens were replaced with custom cross-linked

Teflon-coated support screens in the filter cassettes to avoid potential trace level metals

contamination of the filters. Filters were pre- and post-weighed by ARA in a Class 1000 clean

environment using a Mettler Toledo (Columbus, OH) Model UMX2 micro-balance fitted with an

MTL Model AH225-6 robotic auto-handler (Figure 5). The ARA clean environment maintains

temperature (± 0.1 ºC) and relative humidity (± 2 %) within strict tolerances to ensure consistent

results. The MTL AH225 system performs five replicate weighings of each filter and automatically

minimizes electrostatic effects by utilizing a static discharge Po α-particle emission and Faraday

pan. The balance is zeroed before and after each filter weighing and 1-NIST-traceable Class A

weight and 2 unexposed reference filters are weighed every 6 hours. 3-sigma uncertainties for the

NIST-traceable weight and reference filters are typically 1.0 microgram and 1.6 microgram,

respectively, or <0.1 µg m-3

for a 24-hour sample. ARA performs zero and buoyancy corrections

and reports the mean ± standard deviation weight of each filter.

WBEA contracted site operation personnel received the filter magazines shipped by ARA,

exchanged filters, downloaded sampler data files, maintained the sampler, and reshipped sampled

filters back to ARA. The dichotomous sampler files containing the filter, interval and input data

Page 7: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

7

Figure 4. Schematic of ThermoScientific Dichotomous Sampler Virtual Impactor.

Where: Qo = 16.7 liters min-1

(Total Flow); Q1 = 1.7 liters min-1

(Minor Flow); Q2 = 15.0 liters min-1

(Major Flow)

Figure 5. Picture of ARA’s MTL Model AH225-6 Robotic Weighing System.

Page 8: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

8

were also e-mailed to ARA. These data files contain information for each sample filter such as date

sampled, filter ID, cassette ID, sample volume, flows, temperatures, and quality assurance flags.

2.2. X-ray Fluorescence Analysis

Energy-dispersive X-ray fluorescence analysis (EDXRF) is commonly used as a nondestructive

analytical method for quantifying elemental content in ambient PM samples (Solomon et al., 2001).

EDXRF involves excitation of the constituent atoms in a sample with a nearly bi-chromatic X-ray

beam from a secondary target. As the atoms relax back to the ground state, they emit X-rays whose

energies are characteristic of the element. The fluorescent X-rays impinge on a detector and the

resulting spectrum has an energy profile that is directly related to the elements and their

concentrations in the sample. The spectra are subsequently processed. Multiple linear regression

analysis is used to de-convolute the pulse height spectrum into its background and constituent

element peaks by least-squares fitting of stored elemental thin-film library spectra and lot specific

MTL filter backgrounds. Subsequent processing performs attenuation and interference corrections

and converts raw data to reportable information. Calibrations are performed empirically with thin-

film standards, and verified routinely via analysis of NIST SRM 2783 (Air Particulate Matter on

Filter Media). WBEA dichotomous filters were analyzed using ARA’s PANalytical Epsilon 5

EDXRF spectrometer.

2.3. Filter Sample Extraction for Multi-Element Quantification

Dichotomous filters were digested by ARA personnel with a CEM Corporation (Matthews, NC)

Mars Express microwave digestion system in a cocktail of ultra-pure H2O2, HF and HNO3 with

heating to180ºC for 40 minutes. After cooling, ASTM Type II ultrapure (Resistivity ≥ 18.2

MΩ⋅cm) water was added to each vessel to bring the extraction sample up to a final volume of 25

ml. A standard reference material (SRM) was also digested in triplicate: U.S. National Institute of

Standards and Technology (NIST) 1648a (urban particulate matter).

2.4. Multi-Element ICPMS Analysis

Inductively Coupled Plasma Mass Spectroscopy (ICPMS) has broad acceptance as a method for

determining a wide range of elements in atmospheric PM due to technological advances that

continue to improve sensitivity and reduce interferences (Grohse, 1999). In ICPMS, the filter

sample extract is aspirated through a nebulizer and injected as a droplet aerosol into an argon radio

frequency plasma. In the plasma, the bombardment of the droplets by free electrons causes removal

of the solvent, breakdown of molecules to atoms, and ionization of the atoms to give them a charge

so they can be identified by mass spectroscopy. The ions are extracted from the plasma through a

differentially pumped vacuum interface using a series of electrostatic ion lenses that repel negative

ions and direct positive ions into a quadrupole mass spectrometer. The ions are then sorted

according to their mass-to-charge ratio and individual ions are detected by a counting electron

multiplier. ICPMS provides detection limits on the order of 1-1000 parts per trillion for

approximately 65 elements with a linear dynamic range in excess of eight orders of magnitude

(Solomon et al., 2001).

Sample extracts were analyzed using ARA’s Perkin Elmer (Waltham, MA) Sciex DRCII dynamic

reaction cell ICPMS (DRC-ICPMS) and quantified for 42 elements. The DRC method was

optimized with O2 reagent gas for AsO and NH3 reagent gas for Se quantification, respectively.

Two independent reference solutions and one SRM (NIST 1643e, trace elements in water) were

analyzed to confirm the DRC-ICPMS calibration and six digestion spikes (0.1-1.0 ppb) were

analyzed to assess matrix interferences and estimate method detection limits.

Page 9: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

9

2.5. Stable Lead Isotope Analysis

A portion of each extract from the twelve samples that were run on the DRC-ICPMS were

subsequently analyzed for lead (Pb) isotope ratios using a High Resolution Magnetic Sector Field

ICPMS (HR-ICPMS) at U.S. EPA in Research Triangle Park, NC. The samples represented three

high mass and three low mass sample collection periods for the coarse and fine material. Four of

the twelve samples were coarse and fine samples collected on the same dates (two sets of sample

pairs).

Lead has four major isotopes, 204, 206, 207, and 208. 208

Pb is formed from the radioactive decay of 232

Th, 207

Pb from 235

U, and 206

Pb from 238

U. 204

Pb is referred to as common lead; it has no

radioactive parent, and is much less abundant than the other isotopes. The uranium and thorium

parents have differing decay rates resulting in predictable changes in lead isotope ratios. Plotting

results as 208

Pb/206

Pb isotope ratios (y-axis) versus 207

Pb/206

Pb isotope ratios (x-axis) often yields

a linear array (or a triangular field). Individual data points on such plots typically reflect the age at

which lead was incorporated into the host rock (ore, coal, sediments, oil, or oil sands for example).

The Pb ratios of the parent material are preserved during the subsequent process(es) that emitted the

lead into the environment. On 208

Pb/206

Pb versus 207

Pb/206

Pb plots, older lead will be found in the

upper right quadrant of the diagram, and younger lead in the lower left quadrant. The decrease in 207

Pb/206

Pb ratios from old to young lead reflects the difference in decay rate of the parent 235

U and 238

U isotopes. The difference in 208

Pb/206

Pb can reflect differences in the amount of parent 232

Th

and 238

U. More thorogenic source materials generate higher 208

Pb/206

Pb ratios. Plotting elemental

concentrations on the y-axis versus either 207

Pb/206

Pb or 208

Pb/206

Pb isotope ratios on the x-axis

can result in clusters of data points that are related to the source of the lead in the sample of interest.

2.6. Statistical Analysis

All data management and statistical analysis was conducted using SAS (Cary, NC) v9.2.

3. Results and Discussion

This final report covers the results of the sequential dichotomous sampler operation at the WBEA

AMS-1 Fort McKay site from February 22, 2010 to July 25, 2011, during which ARA processed

399 dichotomous filter sample pairs. This amount represents all the data that ARA has finalized

and provided to date for this period. To evaluate the potential for trace element quantification of the

dichotomous study filters, a subset of samples were selected for EDXRF and DRC-ICPMS analysis.

One hundred seven (107) fine/coarse sample pairs from March 2010 through July 2011 were

selected for EDXRF analysis. Thirty five (35) fine/coarse sample pairs from the EDXRF subset

were then selected for DRC-ICPMS analysis representing a distribution of high and low mass

sample periods. The results are presented and discussed below.

3.1. 2010 PM Mass Concentrations

From February 22, 2010 to December 24, 2010 ARA received 237 daily dichotomous filter sample

pairs from the 306 possible sampling days during the period, resulting in a data completeness of

77%. Reasons for the instrument not sampling include: (i) filter exchange error putting instrument

in “wait” mode until the site operator clears error and restarts; (ii) site operator not loading new

filter magazines before instrument runs out of filters in the supply magazine; and (iii) operator error.

The dichotomous sampler data files indicate that there were typically several days between filter

changes where the sampler ran out of filters and entered “stop” mode before the new filter

magazines were loaded and the sampler restarted. Following QA/QC data screening, there were

Page 10: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

10

208 valid daily dichotomous filter sampling pairs. All but one of the 29 dichotomous filter sample

pairs that were screened out were invalidated due to total sample runs times that were less than 23

hours (resulting from power or sampler problems).

The 2010 PM2.5 and PMcoarse results are summarized below in Table 1. The average 24-hour PM2.5

mass concentration at AMS-1 Fort McKay was 5.7 ± 5.0 µg m-3 (mean ± standard deviation) and the

average 24-hour PMcoarse mass concentration was 7.3 ± 6.0 µg m-3

in 2010. Relative frequency

histograms of the fine and coarse PM data are presented in Figure 6a-b. Box plots depicting the

measures of central tendency and variance of the fine and coarse PM distributions are presented in

Figure 7. All 416 valid dichotomous filters received were above the ARA weighing detection limit

(3σ) of 0.12 µg m-3

.

Table 1. Summary Statistics of Valid AMS-1 PM (µµµµg m-3

) from Feb –Dec 2010 (n=208).

Mean Median Std. Dev. Min 25% Q1 75% Q3 Max

PM2.5 5.7 4.8 5.0 0.4 3.0 7.3 52.3

PMCoarse 7.3 5.6 6.0 0.3 2.8 9.8 30.2

Figure 6a. Relative Frequency Histogram of 2010 AMS-1 PM2.5 Concentrations.

Fine PM (µg m-3

)

0 10 20 30 40 50

Count

0

20

40

60

80

100

120

Page 11: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

11

Page 12: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

12

Figure 6b. Relative Frequency Histogram of 2010 AMS-1 PMcoarse Concentrations.

Coarse PM (µg m-3)

0 5 10 15 20 25 30

Count

0

10

20

30

40

50

Page 13: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

13

Figure 7. Box Plots of 2010 AMS-1 PM Mass Concentrations.

Fine PM Coarse PM

PM

g m

-3)

0

10

20

30

40

50

60

90th Percentile

75th Percentile

25th Percentile

10th Percentile

Median

PM data from July 2010, presented in Figure 8, show a four-day PM2.5 mass excursion resulting in

the year’s highest concentration value of 52.3 µg m-3

, while the PMcoarse mass does not reflect a

similar dramatic increase. These data could represent a local impact from forest fires that could be

confirmed by investigating black carbon and/or potassium concentrations on the collected

dichotomous sample filters.

Page 14: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

14

Figure 8. July 2010 PM2.5 and PMcoarse Mass Concentrations at AMS-1.

3.2. 2011 PM Mass Concentrations

From January 9, 2011 to July 25, 2011 ARA received 162 daily dichotomous filter sample pairs

from the 193 possible sampling days during the period, resulting in a data completeness of 84%.

Improved data completeness in 2011 was achieved by the operators as they addressed most of the

operation and troubleshooting issues that were identified. The only remaining significant source of

missed potential samples during this period was the site operator not loading new filter magazines

before instrument runs out of filters in the supply magazine. The dichotomous sampler data files

indicate that there were typically 1-5 days (3 ± 1) between filter changes where the sampler ran out

of filters and entered “stop” mode before the new filter magazines were loaded and the sampler

restarted. Following QA/QC data screening, there were 151 valid daily dichotomous filter sampling

pairs. All 11 of the dichotomous filter sample pairs that were screened out were invalidated due to

total sample run times that were less than 23 hours.

Seven of the eleven samples that ran less than 23 hours were associated with significant forest fire

impacted samples collected between May 18, 2011 and June 15, 2011. When the loading on the

filter reaches a point where the flow is reduced to below 90% of the programmed set point of either

channel, the sampler will shut down to (i) prevent damage to the system and (ii) to prevent the PM10

and PM2.5 cut points from significantly changing. When the sampler reaches the start time for the

next sample the filters are exchanged and the sampler starts again until those filters are also

overloaded. While samples that ran less than 23 hours are considered not valid indicators of the

daily concentrations, the concentrations are real and representative of that period of time when they

ran. The run time associated with these samples ranged from 4.3 – 15.8 hours (8.5 ± 5). The PM2.5

0

20

40

60

80

100

0

10

20

30

40

50

60

7/2/2010 7/5/2010 7/8/2010 7/11/2010 7/14/2010 7/17/2010 7/20/2010 7/23/2010 7/26/2010 7/29/2010

PM2.5 PMcoarse %Fine

Page 15: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

15

and PMcoarse results associated with the forest fire impacted samples that ran less than 23 hours are

summarized below in Table 3.

Table 3. Summary Statistics of Invalid AMS-1 PM (µµµµg m-3

) from May 18 – June 15, 2011

During Forest Fire Impact (n=7).

Mean Median Std. Dev. Min 25% Q1 75% Q3 Max

PM2.5 351.7 423.8 157.3 152.0 175.9 479.5 522.9

PMcoarse 44.3 40.4 11.0 36.9 38.5 44.5 68.5

The 2011 valid PM2.5 and PMcoarse results are summarized below in Table 4. The average 24-hour

PM2.5 mass concentration at AMS-1 Fort McKay was 8.9 ± 19.7 µg m-3 (mean ± standard deviation)

and the average 24-hour PMcoarse mass concentration was 6.6 ± 5.6 µg m-3

in 2011. Relative

frequency histograms of the fine and coarse PM data are presented in Figure 9a-b. Box plots

depicting the measures of central tendency and variance of the fine and coarse PM distributions are

presented in Figure 10. Of the 302 valid dichotomous filters received by ARA for weighing all but

one coarse filter (0.11 µg m-3

) were above the weighing detection limit (3σ) of 0.12 µg m-3

.

Table 4. Summary Statistics of Valid AMS-1 PM (µµµµg m-3

) from Jan – Jul 2011 (n=151).

Mean Median Std. Dev. Min 25% Q1 75% Q3 Max

PM2.5 8.9 4.8 19.7 0.8 3.2 7.4 170.5

PMcoarse 6.6 4.5 5.6 0.1 2.4 9.4 25.1

The 2011 PM2.5 results were skewed higher (8.9 ± 19.7) versus the 2010 results (5.7 ± 5.0 µg m-3)

due to forest fire impacted samples collected at the site. Three high concentration forest fire

samples in particular collected at the site (May 29, May 30, and June 8) were significant outliers

143.4, 170.5, and 98.0, respectively (Figure 9a and Figure 10). Eleven other samples around this

same time period were invalidated as discussed above and summarized in Table 4 due to sample

filter overloading. If all eleven (11) forest fire impacted samples that ran less than 23 hours were

included in the analysis, they would have had an enormous impact on the 2011 summary statistics

(Table 5).

Table 5. Summary Statistics of All AMS-1 PM (µµµµg m-3

) from Jan – Jul 2011 (n=162).

Mean Median Std. Dev. Min 25% Q1 75% Q3 Max

PM2.5 23.7 5.0 78.6 0.8 3.3 8.0 522.9

PMcoarse 8.4 5.1 9.7 0.1 2.7 10.4 68.5

Page 16: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

16

Figure 9a. Relative Frequency Histogram of 2011 AMS-1 PM2.5 Concentrations.

Fine PM (µg m-3)

0 20 40 60 80 100 120 140 160 180

Count

0

20

40

60

80

100

120

140

Page 17: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

17

Figure 9b. Relative Frequency Histogram of 2011 AMS-1 PMcoarse Concentrations.

Coarse PM (µg m-3)

0 5 10 15 20 25 30

Co

un

t

0

10

20

30

40

50

Page 18: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

18

Figure 10. Box Plots of 2011 AMS-1 PM Mass Concentrations (NOTE: Axis Break).

Fine PM Coarse PM

PM

g m

-3)

0

10

20

30

40

50

100

150

200

90th

Percentile

75th

Percentile

25th

Percentile10

th Percentile

Median

3.3. ED-XRF Analysis

Method detection limits (MDLs) were calculated by ARA for the major, minor, and trace elements

that their PANalytical Epsilon 5 instrument was calibrated to quantify, and they are presented in

Table 6. In general, it was found that the EDXRF instrument (i) was suitable for Al, Si, S, K, Ca,

Fe, and Mn, (ii) was potentially suitable for Na, Mg, Ti, and Zn, and (iii) was unsuitable for Cu, As,

Se, Pb, Ni, V, and Cd in the WBEA dichotomous filter samples from AMS-1. While the EDXRF

instrument did a nice job on sulfur and crustal elements, its inability to adequately detect the

anthropogenic tracers for the source types in the AOSR (e.g., V, Ni, Se, and Pb) would preclude it

from providing adequate data useful for routine PM air quality monitoring as well as for source

apportionment analysis.

3.4. DRC-ICPMS Analysis

Lower detection limits were calculated by ARA for the major, minor, and trace elements that their

Perkin Elmer Sciex DRCII instrument was calibrated to quantify, and are presented in Table 6. In

general, the DRC-ICPMS instrument was capable of providing excellent detection for most

elements of interest, particularly the crustal (Al, Fe, Ca, Si, La, Ce, Sm) and anthropogenic (V, Ni,

Se, and Pb) tracers for the sources in the AOSR. Results for the platinum group elements (Pt and

Pd) were generally at or below instrument detection limits. Larger sample volumes or sample pre-

concentration would be necessary to consistently quantify these elements. DRC-ICPMS analysis

results for the thirty five (35) filter pairs are presented in Table 7 (PM2.5) and Table 8 (PMcoarse).

Page 19: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

19

Results show that virtually all elements of interest are detectable, even at the lowest ambient

concentrations.

Table 6. ARA DRC-ICPMS and XRF Method Detection Limits (ng m-3

).

Element DRC-ICPMS Isotope DRC-ICPMS MDL XRF MDL

Li 7 0.0366

Be 9 0.0025

Na 23 2.4210 5.62

Mg 26 1.4244 -

Al 27 5.2355 5.66

Si 28 51.6464 3.96

K 39 1.8733 0.93

Ca 44 23.8882 2.51

Ti 49 0.2247 1.62

V 51 0.0094 -

Cr 53 0.1853 -

Mn 55 0.1520 3.11

Fe 56 29.4447 3.45

Ni 62 0.1434 -

Cu 65 0.7259 1.02

Zn 68 0.4420 1.32

Se 78 0.0160 3.91

Rb 85 0.0036 -

Sr 88 0.0233 -

AsO 91 0.0067 1.89

Nb 93 0.0102 -

Mo 98 0.0193 -

Pd 108 0.0052 -

Cd 114 0.0020 -

Sn 118 0.0234 -

Sb 123 0.0178 -

Cs 133 0.0009 -

Ba 137 0.1403 18.06

La 139 0.0009 -

Ce 140 0.0017 -

Nd 143 0.0009 -

Ta 181 0.0001 -

W 182 0.0058 -

Pt 195 0.0003 -

Pb 208 0.1742 3.61

Th 232 0.0003 -

U 238 0.0007 -

Page 20: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

20

Table 7. DRC-ICPMS Analysis Results of 35 PM2.5 samples at AMS-1 (ng m-3

).

Element Isotope n (>MDL) Mean Stdev Min Max

Li 7 29 0.0765 0.0577 0.0023 0.2165

Be 9 22 0.0020 0.0019 0.0001 0.0061

Na 23 32 16.90 39.27 0.7433 217.36

Mg 26 34 19.36 20.28 0.6049 104.74

Al 27 33 54.24 48.88 6.89 178.9

Si 28 30 128.35 112.80 6.999 392.8

K 39 34 171.02 315.37 2.697 1064.6

Ca 44 29 63.07 56.51 6.15 291.9

Ti 49 33 1.71 2.24 0.01 9.65

V 51 33 0.208 0.171 0.004 0.796

Cr 53 24 0.596 0.730 0.013 2.504

Mn 55 34 2.494 3.268 0.027 17.252

Fe 56 33 57.898 49.368 0.469 184.732

Ni 62 25 2.168 9.861 0.007 49.496

Cu 65 22 1.717 2.977 0.017 14.083

Zn 68 31 17.015 30.103 0.037 105.793

Se 78 33 0.199 0.480 0.012 2.770

Rb 85 34 0.454 0.759 0.004 3.042

Sr 88 33 0.352 0.291 0.007 1.295

As 91 34 0.271 0.396 0.002 1.486

Nb 93 22 0.007 0.007 0.001 0.027

Mo 98 32 0.546 0.055 0.001 0.210

Pd 108 17 0.023 0.049 0.001 0.196

Cd 114 34 0.552 1.587 0.001 7.824

Sn 118 33 0.116 0.147 0.007 0.734

Sb 123 30 0.050 0.065 0.001 0.290

Cs 133 34 0.012 0.014 0.001 0.053

Ba 137 33 0.910 0.898 0.041 4.557

La 139 34 0.036 0.031 0.002 0.117

Ce 140 34 0.069 0.060 0.005 0.230

Nd 143 34 0.028 0.025 0.001 0.094

Ta 181 30 0.0005 0.0005 0.0003 0.0021

W 182 28 0.0093 0.0088 0.0004 0.0365

Pt 195 17 0.0006 0.0009 0.0001 0.0038

Pb 208 32 0.762 0.826 0.020 3.389

Th 232 34 0.010 0.010 0.0001 0.0365

U 238 32 0.0031 0.0035 0.0001 0.0141

Page 21: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

21

Table 8. DRC-ICPMS Analysis Results of 35 PMcoarse samples at AMS-1 (ng m-3

).

Element Isotope n (>MDL) Mean Stdev Min Max

Li 7 35 0.351 0.339 0.001 1.325

Be 9 34 0.0122 0.0114 0.0002 0.0436

Na 23 32 25.52 24.52 1.03 104.73

Mg 26 35 86.96 80.61 2.25 324.09

Al 27 35 375.40 366.45 14.72 1445.69

Si 28 35 761.12 735.65 2.04 2854.59

K 39 35 112.10 101.25 13.51 438.92

Ca 44 35 450.86 450.77 5.96 2054.31

Ti 49 35 8.89 8.30 0.28 34.27

V 51 35 0.878 0.760 0.003 2.557

Cr 53 27 0.919 2.026 0.014 10.706

Mn 55 34 9.23 13.04 0.05 71.51

Fe 56 35 329.93 316.80 1.77 1393.09

Ni 62 29 0.896 2.306 0.034 12.700

Cu 65 21 1.377 1.677 0.008 5.444

Zn 68 33 2.765 3.192 0.068 13.284

Se 78 35 0.123 0.162 0.002 0.899

Rb 85 35 0.575 0.536 0.031 2.226

Sr 88 35 1.544 1.425 0.006 6.221

As 91 35 0.084 0.078 0.002 0.316

Nb 93 30 0.038 0.036 0.004 0.146

Mo 98 28 0.080 0.091 0.002 0.433

Pd 108 20 0.0342 0.1064 0.0003 0.4841

Cd 114 34 0.0084 0.0176 0.0001 0.1031

Sn 118 29 0.0721 0.1565 0.0006 0.8501

Sb 123 28 0.0396 0.0583 0.0001 0.2165

Cs 133 35 0.0280 0.0271 0.0004 0.0944

Ba 137 35 3.775 3.280 0.024 16.358

La 139 36 0.2189 0.2065 0.0014 0.8569

Ce 140 36 0.4114 0.3852 0.0008 1.5645

Nd 143 35 0.1810 0.1686 0.0009 0.6661

Ta 181 35 0.0046 0.0099 0.0001 0.0554

W 182 32 0.0499 0.0518 0.0008 0.1906

Pt 195 21 0.0013 0.0027 0.0001 0.0122

Pb 208 28 0.2237 0.2781 0.0011 1.4190

Th 232 34 0.0595 0.0547 0.0031 0.2219

U 238 34 0.0149 0.0138 0.0007 0.0572

Page 22: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

22

3.5. Stable Lead Isotope Analysis

Forty-two elements were quantified using the DRC-ICPMS. For purposes of this pilot feasibility

study, the results for elements that are representative tracer species for sources in the AOSR will be

included on the data plots. The elements chosen were Aluminum (Al), Calcium (Ca), Uranium (U),

Vanadium (V), Lead (Pb) and Zinc (Zn).

On a 208

Pb/206

Pb versus 207

Pb/206

Pb plot from the twelve (12) dichotomous samples analyzed in

this pilot feasibility study, a good range in isotopic ratios was found, indicating the potential for Pb

isotopes to assist in source type identification in the AOSR. This potential is exemplified when

either mass or element concentrations are combined on the same plot with Pb isotope ratios. On

such plots, several fields of data points (clusters) were found. The clusters likely correspond to

differences in source materials. A total of three or four fields of data clusters are found on Figures

11 and 12.

Figure 11. Relationship between PM Mass and 207

Pb/206

Pb.

0.0

5.0

10.0

15.0

20.0

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Ma

ss (

ug

/m3

)

Coarse Mass

Fine Mass

Page 23: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

23

Figure 12. Relationship between 208

Pb/206

Pb and 207

Pb/206

Pb.

High mass concentration in the coarse fraction samples corresponds best to samples with elevated

Al, Ca, U and V concentrations (Figure 13). The lowest 207

Pb/206

Pb ratios are found in these high

mass concentration samples, clustering near a 207

Pb/206

Pb ratio of 0.835. High mass concentrations

found in the fine fraction samples corresponded best to elevated Zn and Pb concentrations. The 207

Pb/206

Pb ratios in these samples clustered near values of 0.862. The two other sets of data

clusters include samples with low concentrations for most elements, but differences in Pb isotope

ratios (0.860 versus 0.870) and Zn concentrations are still apparent.

2.0600

2.0700

2.0800

2.0900

2.1000

2.1100

2.1200

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

20

8 P

b /

20

6 P

b

Coarse

Fine

Page 24: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

24

Figure 13(a-f). Relationship between Trace Elements and 207

Pb/206

Pb.

Some indication of the source of the PM collected by the dichotomous sampler can be elucidated by

coupling the Pb isotope ratios to AOSR source type samples that had previously been analyzed for

Pb isotope ratios (Joe, need reference here). Three groups of source type samples were analyzed that

tend to cluster into distinct fields in isotope space. The lowest 207

Pb/206

Pb and 208

Pb/206

Pb ratios are

found in processed oil sand samples (e.g., fly ash and coke samples from the oil upgrading

facilities), the mid range 207

Pb/206

Pb and 208

Pb/206

Pb ratios are found in the raw oil sands, and the

highest 207

Pb/206

Pb and 208

Pb/206

Pb ratios were found in the tailings sand (raw oil sand from which

the oil had been extracted).

When comparing the dichotomous samples to the source samples, it appears that high mass, coarse

fraction samples are most similar to the raw oil sands signature in Pb isotope space (Figure 14).

This suggests the dichotomous samples captured fugitive dust emissions from mining operations on

the high mass concentration days. The signature of the Pb isotopes from the other dichotomous

samples (both coarse and fine fractions) clusters into fields that are closest to the signature of the

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Ura

niu

m (

ng/m

3)

U Coarse

U Fine

0.000

0.500

1.000

1.500

2.000

2.500

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Lead

(n

g/m

3)

Pb Coarse

Pb Fine

0

200

400

600

800

1000

1200

1400

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Alu

min

um

(n

g/m

3)

Al Coarse

Al Fine

0.000

2.000

4.000

6.000

8.000

10.000

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Zin

c (n

g/m

3)

Zn Coarse

Zn Fine

0

200

400

600

800

1000

1200

1400

1600

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Calc

ium

(n

g/m

3)

Ca Coarse

Ca Fine

0.000

0.500

1.000

1.500

2.000

2.500

0.8300 0.8400 0.8500 0.8600 0.8700 0.8800

207 Pb / 206 Pb

Van

ad

ium

(n

g/m

3)

V Coarse

V Fine

(a) (b)

(c) (d)

(e) (f)

Page 25: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

25

tailings sand. The tailings sand, however, might also be expected to have a similar signature to that

of background soils away from the mining sites. At this time, the source of materials with high

concentrations of Zn and Pb in the fine mass fraction has not been identified. In addition, none of

the dichotomous samples captured a significant amount of material that corresponds to the

processed oil sand Pb isotope signatures.

Figure 14. Relationship between Dichotomous Sample and Source Sample Pb Isotopes.

1.960

1.980

2.000

2.020

2.040

2.060

2.080

2.100

2.120

2.140

0.7800 0.8000 0.8200 0.8400 0.8600 0.8800 0.9000207 Pb / 206 Pb

20

8 P

b /

20

6 P

b

Processed

Oil Sand

Tailings Sand

Dichot Coarse

Dichot Fine

Page 26: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

26

4. Conclusions

• The ThermoScientific Model 2025 Sequential Dichotomous PM Sampler operated well in

the AOSR. Data completeness (77%) suffered mainly due to site operation procedures

rather than mechanical problems with the sampler. Recommendations for increased sample

collection completeness are presented in Section 5.1 of this report.

• The PM mass concentration data provided by ARA’s robotic weighing system was robust,

with all but one (99.8%) of the filters being above their weighing detection limit (3σ) of

0.12 µg m-3

.

• EDXRF analysis of the dichotomous sampler filters provided reliable trace element data for

sulfur and some other mainly crustal elements (Al, Si, K, Ca, Fe, Mn). These data are

amenable for use in evaluating regional sulfate, windblown dust, and extraction efficiency of

subsequent DRC-ICPMS analysis, but not for providing data for adequate source type

identification from routine PM air quality monitoring or adequate data for a robust statistical

receptor modeling effort.

• DRC-ICPMS analysis of the dichotomous sampler filters provided reliable data for a

comprehensive list of trace elements. The resulting data (even from low mass samples) are

amenable for routine PM air quality monitoring and a robust statistical receptor modeling

effort.

• HR-ICPMS stable Pb isotope analysis appears to have great potential for use as a source

apportionment tool in conjunction with a statistical receptor modeling effort.

5. Recommendations

5.1. Dichotomous Sampler Field Operation

Review of the Dichotomous Sampler Data files revealed that during the 2010 and 2011 sampling

years, the site technicians responsible for the operation of the sampler allowed it to run out of filters

at the end of each 2 week filter change cycle. This means the sampler ran out of filters and

automatically shut down on a routine basis, resulting in overall data completeness of only about

80%. Typically, 2-4 days later new filters were installed and the sampler was restarted. It is

recommended that a new sampling SOP be implemented that directs the site operators to load new

filter magazines while the last filter in the previous magazine is running. This procedural change

will minimize missed sample days, avoid confusion at the ARA lab concerning the status of the

instrument, and result in regular filter shipment and instrument maintenance scheduling. Field

operations and data management could also be tightened by routine (daily) automated polling of

data from the dichotomous sampler. This would allow near real-time detection of sample exchange

failures and equipment malfunctions by WBEA personnel and would allow better filter tracking by

ARA personnel.

5.2. Dichotomous Sampler Sample Analysis

It is recommended that an analysis plan be developed for archived dichotomous filter samples

collected to date using DRC-ICPMS (multi-element) and HR-ICPMS (Pb stable isotopes), with a

sufficient number of samples analyzed to satisfy the requirements of contemporary statistical

receptor models such as Positive Matrix Factorization (PMF) and Unmix that require a relatively

large data set (n>100) to provide feasible analytical solutions. A one in three day analysis plan

Page 27: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

27

would provide analytical results for approximately 122 filter pairs. Annual receptor modeling

results will provide (i) a temporal record of significant sources impacting AMS-1 allowing WBEA

to track the relative strengths of source types in the AOSR as oil production increases, and (ii) the

data to evaluate the efficacy of adopted abatement measures.

5.3. Addition of PM Carbon Measurements

It is recommended that instrumentation to measure organic carbon (OC) and black carbon (BC) be

deployed to supplement the dichotomous sampler measurements. Particulate carbon is a major

component of ambient PM and can account for 25-50% of PM mass (Solomon et al., 2001; Landis

et al., 2001). Black carbon is emitted primarily from anthropogenic sources, such as diesel vehicle

exhaust, due to incomplete combustion. Organic carbon is emitted from anthropogenic sources,

biogenic sources and wildfires; and is formed in the atmosphere from gaseous precursors

(secondary aerosols). The addition of OC and BC measurements at AMS-1, in conjunction with the

dichotomous sampler and the URG AIM 9000 sampler, will provide data for (i) a comprehensive

PM mass reconstruction to fully elucidate the chemistry of aerosols in the AOSR, and (ii) an

integrated source apportionment analysis into the source types contributing to all the major PM

components in the OASR.

OC/BC measurements can be made using automated continuous instruments such as the Sunset

Instruments (Portland, OR) Model 4 analyzer or from integrated filter-based collection. FRM/FEM

samplers (16.7 LPM) configured with quartz filters can provide sufficient mass to make reliable

OC/BC measurements. The use of high volume PM2.5 samplers such as the Tisch (Village of

Cleves, OH) model TE-1202 (113 LPM) or model TE-1000 (226 LPM) are typically used for the

collection of samples for subsequent speciation of primary and secondary organic species,

respectively. The speciation of OC would provide important information on (i) the sources of

primary organic aerosol emissions, (ii) the importance of atmospheric chemistry leading to the

formation of secondary aerosols, and (iii) the overall impact of OC on PM concentrations in the

AOSR. The nature of the raw oil sands and the resulting on-site upgrading activities may result in

unique emission profiles that may prove extremely beneficial in elucidating the relationship

between oil production activities and PM concentrations in the AOSR. Unique tracers for the

AOSR activities might include napthenic acids and unique tracers for forest fires might include

levoglucosan.

5.4. Expansion of the Dichotomous Sampler Network

It is recommended that a plan be developed for the expansion of the dichotomous monitoring

network to include (i) a regional background site, (ii) a significant source impacted site, and (iii) a

downwind site. The addition of a background site will allow quantification of the AOSR local

source enhancement to ambient PM. The addition of a source impacted site would assist in the

elucidation of the geographic location of significant PM air pollution sources in the AOSR by

enabling the use of hybrid receptor models like quantitative transport bias analysis (QTBA). The

QTBA model can incorporate meteorological and multiple site monitoring data to generate spatial

probability fields of source areas (Keeler and Samson, 1989). The downwind site will provide

information on transport scales for fine and coarse PM and will be instrumental in constraining

atmospheric dispersion and transport models.

As part of the network expansion, it is also recommended that an ongoing one in three day analysis

plan be developed for the inorganic speciation of PM2.5 and PMcoarse. Over time, this recommended

level of effort will generate a unique time series of PM2.5 and PMcoarse mass and trace element data,

as well as generate an invaluable inventory of PM2.5 and PMcoarse samples for use in special studies

Page 28: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

28

(e.g., Pb isotope analysis, investigation of forest fire impacts, extreme event analysis). Collection of

daily samples will provide the ability to investigate transient events that could otherwise be missed

on a one in three day sampling plan and provide additional filters for analysis to make up for

maintenance or repair related sampler down time.

6. Response to Initial Review Comments

6.1. Housing the Sampler Inside the Monitoring Station

Question: The current sampler is housed inside the station, while typical applications are for

outside installations. Are there any concerns around certain compounds being lost due to being in a

cold ambient environment and then sitting on a filter for several days at room temperature? How

much does the internal/external sampling conditions factor into the final result?

Answer: In the winter, there may be a potential for the loss of some semi-volatile compounds.

However, all reference methods for the weighing of filters call for the equilibration of the filters for

a minimum of 24 hours at 20-23ºC and 30-40% relative humidity. Therefore, we do not think there

would be an additional significant loss of sample mass over and above that expected to be lost

during the filter equilibration and weighing process.

The inorganic species targeted by DRC-ICPMS analysis for the proposed source apportionment

analysis are non-volatile and would not be significantly impacted by sampling inside the shelter. If

the reports recommendation to initiate carbon PM measurements is implemented, a total PM mass

reconstruction could be performed and any loss of PM mass could be quantified. If additional

dichotomous samplers are procured for deployment, we suggest a short collocated sampling

experiment during the winter with one sampler inside and one sampler outside. This test would

provide a sound basis for future sampler placement. Because of the negative impact of cold

temperatures on the sampler seals, pumps, and filter exchange mechanisms it is preferable that the

samplers are located inside the shelter.

6.2. Acceptance of Dichotomous Sampler

Question: Would the wider audience of monitoring stakeholders accept the use of these samplers

based upon this report and in comparison with other sampling methods? Based on the sample

location recommendation, are 3 stations enough to represent the airshed for PM data?

Answer: The ThermoScientific model 2025 sequential dichotomous PM sampler is a U.S. EPA

designated Federal Equivalent Method for PM2.5 that has been evaluated against both manual and

sequential FRM samplers and found to be equivalent (Poor et al., 2002; Chen et al., 2011). In

addition, dichotomous samplers have seen broad application in both the international PM

monitoring and research communities (Dzubay and Stevens, 1975; McFarland et al., 1978; Stevens,

et al., 1980; O'Conner and Jaklevic, 1981; Loo and Cork, 1988; Chan et al., 1997; Wei et. al, 1999;

Cabada et al., 2004; Kim et al., 2005), including the Canadian National Air Pollution Surveillance

(NAPS) Network (Brook et al., 1996) and the seminal work linking PM and health effects such as

the Harvard Six Cities Study (Dockery et al., 1993). The science is clear and we anticipate no

problems with the wider audience of monitoring stakeholders accepting the use of dichotomous

samplers.

Page 29: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

29

Recommendation 5.4 suggests the addition of three (3) dichotomous sampling sites to the existing

site AMS #1 at Fort McKay making a total of four (4) sites. PM2.5 concentrations should be well

characterized with a long term four sampler network in the AOSR, given the dispersion and mixing

characteristics of fine mode aerosols. PMcoarse concentration can vary significantly over short

spatial and temporal scales. This proposed network design (collection and analysis) will provide

very important data, and should be adequate to address the goals identified above including (i) to

quantify the local source enhancement, (ii) to elucidate transport scales of PM, (iii) to constrain

atmospheric dispersion and transport models, and (iv) to identify the geographic location of

significant PM air pollution source types in the AOSR.

6.3. Movement of Dichotomous Sampler

Question: Why wouldn’t the current sampler be moved to another location, rather than install

additional samplers? This would be consistent with WBEA’s typical approach for non-compliance

sampling.

Answer: If the only goal of the dichotomous sampling strategy is to quantify PM mass for

regulatory compliance, moving a single sampler around the network may be satisfactory. If the goal

is to understand the dynamics of PM emission, transport, and atmospheric deposition; then

simultaneous sampling at multiple locations is required. Spatial dynamics, statistical association,

and variance/covariance structure between network sites is critical to addressing the research goals

identified above.

Page 30: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

30

6.4. Sampling by Difference versus Dichotomous Sampler

Question: While this report seems to show that the dichotomous sampler works, is it really any

better than sampling for PM2.5 fine separate from PMcoarse? Various industry approvals require

sampling of PM2.5 and PM10 and this is completed through the separate filter measurements.

Answer: There are three main reasons why we feel the dichotomous sampler is a superior approach

to sampling for PM2.5 and PM10 separately, and calculating PMcoarse by difference:

(1) The chemistry of fine mode (acidic) and coarse mode (basic/alkaline) aerosols are very

different. Collecting them together on a common filter allows for surface chemistry to occur

and can lead to increased gas phase artifact formation and aerosol decomposition. By design,

the dichotomous sampler collects all the PMcoarse aerosols from the total sampler flow of 16.7

LPM on a filter with an actual 1.67 LPM flow rate (pre-concentration). This means there is an

order of magnitude less volume of air flowing through the filter leading to (i) much less chance

of gas phase artifact formation and (ii) lower face velocity/pressure drop which leads to less

aerosol decomposition.

(2) There is always an associated uncertainty with analytical chemistry results. In many cases the

signal to noise ratio is a major determiner of lower limits of detection and overall uncertainty.

For anthropogenic species that are present predominantly in the fine mode aerosols, the PM2.5

correction may be larger than the total PMcoarse concentration, leading to higher overall

uncertainties in the analytical results and hindering the subsequent hypothesis testing and

statistical model applications for the data.

(3) Logistically, running one sampler per site instead of two reduces the time and effort required

for the operation, maintenance, data collection, and QA/QC procedures.

7. Path Forward

7.1. Evaluation of Data Completeness for the Dichotomous Sampler versus Existing FRMs

The dichotomous study team did not have knowledge of the existence of the FRM samplers at

AMS-1 prior to receipt of the review comments on the initial report, and therefore did not have a

comparison between the samplers in mind during the filter selection for chemical analysis. WBEA

subsequently provided the study team a 2010 and a 2011 data file containing a total of 44 PM2.5 and

107 PM10 FRM sample results from Fort McKay. We did not receive a schedule of planned FRM

sampling days to compare to the actual filters collected and successfully analyzed, so overall

completeness could not be assessed. However, we ran the samples in the data set through the

QA/QC screening program that was developed for the dichotomous data set and was able to

calculate the percentage of reported filter collections that passed QA/QC and compare these results

to the dichotomous sampler in Table 9. The QA/QC screening parameters are (i) a valid daily

sample must have a run time of 24 ± 1 hour, and (ii) the PM mass must be greater than 0 µg m-3

.

The reasons for the dichotomous samples not passing QA/QC are discussed in sections 3.1 and 3.2,

including significant forest fire impacted samples causing excessive loading on the filters and

reaching a point where the sampler automatically shuts down to protect hardware components.

Page 31: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

31

Table 9. Collected Filters that Pass QA/QC from FRM and Dichotomous Samplers.

Sampler Total Samples Total Passed QA/QC Percentage Passed

FRM PM2.5 44 20 45%

FRM PM10 107 46 43%

Dichot PM2.5 399 360 90%

Dichot PM10 399 360 90%

7.2. Comparison of Dichotomous Sampler and Existing FRM Mass Data

There is very limited overlap in the FRM and dichotomous sampler data sets. The PM2.5 FRM

sampler began operation on February 2, 2011 and the dichotomous sampler ended operation on July

25, 2011. During the overlap time frame, nineteen (18) PM2.5 FRM samples were contained in the

data set provided by WBEA, of those samples eight (7) passed QA/QC criteria (February 2, 8, 20,

26; March 4, 10, 16). The dichotomous sampler was not run on February 8 or March 10, 2011,

leaving five (5) valid observations for comparison. Figure 15 depicts the relationship between the

PM2.5 FRM and the dichotomous PM2.5. The dichotomous sampler error bars are the weighing

uncertainty provided by ARA. The slope of the linear regression line is 0.74 and the coefficient of

determination is 0.99. On average, the two samplers provided highly correlated results with the

FRM sampler measuring 26% higher PM2.5 mass.

Page 32: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

32

Figure 15. PM2.5 Relationship between FRM and Dichotomous Samplers at AMS #1.

WBEA FRM (µg m-3)

0 2 4 6 8 10 12 14 16

WB

EA

Dic

ho

t F

ine

(µg m

-3)

0

2

4

6

8

10

12

14

16

Dichot = 0.7406 * FRM + 0.8856

r2 = 0.9942

Data from the WBEA continuous TEOM PM2.5 monitor at Fort McKay were also provided to the

study team. The relationship between the dichotomous sampler and the TEOM for the five

common sampling days between the FRM and dichotomous sampler are presented in Figure 16.

The slope of the linear regression line is 1.39 and the coefficient of determination is 0.997. On

average, the two samplers provided highly correlated results with the dichotomous sampler

measuring 39% higher PM2.5 mass. During this small sample comparison window in the winter of

2011, all three sampling methods were highly correlated with mass concentrations reported by the

samplers FRM>dichot>TEOM. All the valid 2011 TEOM and dichotomous sampler PM2.5 mass

data was also compared and is presented in Figure 17. When evaluating the longer data period

(January – July, 2011) including winter and summer seasons, the instruments compared very well

with a slope of 1 and a coefficient of determination of 0.95.

Page 33: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

33

Figure 16. PM2.5 Relationship between TEOM and Dichotomous Sampler at AMS #1.

WBEA TEOM (µg m-3)

0 2 4 6 8 10 12

WB

EA

Dic

ho

t ( µ

g m

-3)

0

2

4

6

8

10

12

Dichot = 1.393 * TEOM - 0.009

r2 = 0.997

Page 34: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

34

Figure 17. PM2.5 Relationship between TEOM and Dichotomous Sampler at AMS #1

(January 2011 – July 2011; n=149).

WBEA TEOM (µg m-3

)

0 25 50 75 100 125 150 175

WB

EA

Dic

hot

(µg

m-3

)

0

25

50

75

100

125

150

175

Dichot = 1.00 * TEOM + 0.63

r2 = 0.952

In all of 2010 & 2011 there are eight (8) days with valid PM2.5 and PM10 FRM filters (July 26, 2011;

August 7, 2011; August 13, 2011; August 19, 2011; September 30, 2011; October 6, 2011; October

12, 2011; October 30, 2011). Since the dichotomous sampler study ended on July 25, 2011, there

are no days in common for which to base a comparison.

7.3. Comparison of Dichotomous Sampler and Existing FRM Metals Data

The dichotomous sampler filters for the five (5) days in common with the PM2.5 FRM are currently

being extracted in preparation for analysis by DRC-ICPMS by ARA. An addendum to this report

detailing this comparison will be submitted shortly.

Page 35: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

35

8. Acknowledgements

The pilot sequential dichotomous sampler PM research study described in this report was funded by

WBEA. We thank Kevin Percy (WBEA), Allan Legge (Biosphere Solutions), and Robert K.

Stevens for their support and insight. The content and opinions expressed by the authors in this

report are their own and do not necessarily reflect the views of the Wood Buffalo Environmental

Association (WBEA) or of the WBEA membership.

9. References

Brook, J.R.; Dann, T.F.; Burnett, R.T. 1997. The relationship among TSP, PM10, PM2.5, and

inorganic constituents of atmospheric particulate matter at multiple Canadian locations. Journal of

the Air and Waste Management Association, 47, 2-19.

Chan, Y.C.; Simpson, R.W.; McTainsh, G.H.; Vowles, P.D.; Cohen, D.D.; Bailey, G.M. 1997.

Characterization of chemical species in PM2.5 and PM10 aerosols in Brisbane, Australia. Atmos

Environ, 31, 3773-3785.

Chen, F.L.; Vanderpool, R.; Williams, R.; Dimmick, F.; Grover, B.D.; Long, R.; Murdocj, R. 2011.

Field Evaluation of portable and central site PM samplers emphasizing additive and differential

mass concentrations estimates. Atmos Environ, 45, 4522-4527.

Dockery, D.W.; Pope III, A.; Xu, X.; Spengler, J.D.; Ware, J.H.; Fay, M.E.; Ferris Jr., B.G.;

Speizer, F.E. 1993. An association between air pollution and mortality in six U.S. cities. New

England Journal of Medicine, 329, 1753-1759.

Dzubay, T.G.; Stevens R.K. 1975. Ambient Air Analysis with Dichotomous Sampler and X-ray

Fluorescence Spectrometer. Environ. Sci. Technol. 9, 663-668.

Cabada, J.C.; Rees, S.; Takahama, S.; Khlystov, A.; Pandis, S.N.; Davidson, C.I.; Robinson, A.L.

2004. Mass size distributions and size resolved chemical composition of fine particulate matter at

the Pittsburgh supersite. Atmos Environ, 38, 3127-3141.

Grohse, P.M. 1999. Trace element analysis of airborne particles by atomic absorption spectroscopy,

inductively coupled plasma-atomic emission spectroscopy, and inductively coupled plasma-mass

spectrometry. In Environmental analysis of airborne particles (Advances in environmental,

industrial and process control technologies; v.1). Eds. Landsberger, S. and M. Creatchman, the

Netherlands: Gordon and Breach Science Publishers, pp. 1 - 65.

Hopke, P.K. 2009. Theory and application of atmospheric source apportionment. In Air Quality

and Ecological Impacts: Relating Sources to Effects, Ed. A.H. Legge. Elsevier Science, Amsterdam,

Netherlands, pp. 1-33.

Keeler, G.; Samson, P. 1989. Spatial Representativeness of Trace Element Ratios. Environ. Sci.

Technol. 23, 1358-1364.

Kim, E.; Hopke, P.K.; Pinto, J.P.; Wilson, W.E. 2005. Spatial variability of fine particle mass,

components, and source contributions during the regional air pollution study in St. Louis. Environ.

Sci. Technol. 39, 4172-4179.

Landis, M.S.; Norris, G.A.; Williams, R.W.; Weinstein J. Personal exposures to PM2.5 mass and

trace elements in Baltimore, MD, USA. Atmos Environ 2001, 35, 6511-6524.

Lin, J.J. 2002. Characterization of the major chemical species in PM2.5 in the Kaohsiung City,

Taiwan. Atmos Environ 2002, 36, 1911-1920.

Page 36: Evaluation of the Thermo Scientific Model 2025 Sequential … · Evaluation of the Thermo Scientific Model 2025 Sequential Dichotomous Sampler for the Collection of Fine (

36

Loo, B.W; Cork, C.P. Development of High Efficiency Virtual Impactors. Aerosol Sci. Technol.

1988, 9, 167-176.

McFarland, A.R.; Ortiz, C.A.; Bertch Jr, R.W. 1978. Particle collection characteristics of a single

stage dichotomous sampler. Environ. Sci. Technol. 12, 679-682.

O’Conner, B.H.; Jaklevic, J.M. 1981. Characterization of ambient aerosol particulate samples from

the St. Louis area by X-Ray power diffractometry. Atmos Environ, 15, 1681-1690.

Poor, N.; Clark, T.; Nye, T.; Tamanini, T.; Tate, K.; Stevens, R.K.; Atkeson, T. 2002. Field

performance of dichotomous sequential PM air samplers. Atmos Environ, 36, 3289-3298.

Seinfeld, J.H.; Pandas, S.N. 2006. Atmospheric Chemistry and Physics – From Air Pollution to

Climate Change. Second edition, John Wiley and Sons, Inc., New York, NY.

Solomon, P.A.; Norris, G.A.; Landis, M.S.; Tolocka, M. 2001. Chemical analysis methods for

atmospheric aerosol components. In Aerosol Measurement, Baron, P.A. and Willeke, K eds, John

Wiley & Sons, Inc., 261-293.

Stevens, R.K.; Dzubay, T.G.; Shaw Jr., R.W.; McClenny, W.A.; Lewis, C.W.; Wilson, W.E. 1980.

Characterization of the aerosol in the Great Smokey Mountains. Environ. Sci. Technol. 14, 1491-

1498.

Wei, F; Teng, E.; Wu, G.; Hu, W.; Wilson, W.E.; Chapman, R.S.; Pau, J.C.; Zhang, J. 1999.

Ambient concentrations and elemental compositions of PM10 and PM2.5 in four Chinese cities.

Environ. Sci. Technol. 33, 4188-4193.