UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
^to St^ Research Triangle Park, NC 27711
OFFICE OF
AIR AND RADIATION
MEMORANDUM
DATE: July 31, 2019
SUBJECT: Interlaboratory Results of the OAQPS 2018 Mega PE Speciation Event
FROM: Jenia McBrian (OAQPS)
TO: Tracy Dombek (RTI)
Raul Dominguez (SCAQMD)
Chris Moore (ODEQ)
Xiaoliang Wang (DRI)
Dominique Young (UCD)
CC: Mike Crowe (R4)
Melinda Beaver (OAQPS)
Richard Guillot (R4)
Chris Hall (RIO)
Joann Rice (OAQPS)
Ryan Stokes (OAQPS)
Dena Vallano (R9)
Nealson Watkins (OAQPS)
Introduction
As many of you are aware, the PM2.5 Mega performance evaluation (PE) program was suspended for
approximately 3 years while the program was transitioned from NAREL to OAQPS. OAQPS resumed
the program in late 2017 and completed the first study in early 2018. Draft results were circulated on
October 17, 2018; however, one laboratory had yet to analyze their PE samples because their
instrumentation was in the process of being updated.
All data have now been received and the results presented here are considered final.
This study was conducted as part of EPA's quality assurance oversight of the PM2.5 chemical speciation
air monitoring network (CSN) and the Interagency Monitoring of Protected Visual Environments
(IMPROVE) Program. The purpose of the study was to evaluate specific laboratory performance at those
laboratories that routinely analyze PM2.5 speciation samples. Five laboratories participated in this study:
California Air Resources Board (CARB), Desert Research Institute (DRI), the Oregon Department of
Environmental Quality (ODEQ), Research Triangle Institute (RTI), and the South Coast Air Quality
Management District (SCAQMD).
As in previous studies, each participating laboratory analyzed a set of blind PE filter samples. The PE
samples were prepared by the Office of Air Quality Planning and Standards (OAQPS) at the Research
Triangle Park (RTP), NC facility. For each analysis type, three sets of collocated ambient air filter
samples were collected over varying time periods to ensure sufficient particulate were collected to span
Page 1 of 8
-------
the PM2 5 Network average concentrations. The collocated sampling system was designed and fabricated
at OAQPS in RTP, NC and is used for both the Mega PE and Gravimetric Round Robin PE events. The
sampler can collect up to 32 collocated samples simultaneously and achieves 5% precision between
samples (verified though gravimetric QC studies conducted prior to each PE event and flow checks at
each cyclone prior to every sampling event). Photos of one of the four sampling manifolds containing
eight cyclones, and the collocated sampling system and are shown in Figures 1 and 2, below.
Figure 1. One of four sampling manifolds on the 32-cyclone collocated PE sampler at OAQPS in RTP,
NC
Page 2 of 8
-------
Figure 2. PE Sampling system consisting of 32 PM2 5 cyclones on four sampling manifolds and one
dedicated pump (in pump-box in the foreground)
Each laboratory received the following set of PE speciation samples:
• Anion and Cation Analysis by Ion Chromatography (IC)
o Five Nylon® filter samples (all labs)
o Six Teflon® filter samples (one lab)
• Carbon by Thermal Optical Analysis (TOA)
o Five quartz filter samples
o Four quartz filter samples (one lab)
• Elemental analysis by X-Ray Fluorescence (XRF)
o Five 47 mm Teflon® filter samples
Data Analysis
OAQPS does not have its own laboratories and was unable to successfully qualify external referee labs,
therefore it was not possible to obtain reference values for the PE samples. Since the lab results could not
be evaluated against an assigned value (referee lab result), OAQPS evaluated each result against the
results of the other laboratories participating in the study (interlaboratory comparison).
When the draft data were reported, results submitted as either "ND" (not detected) or "
-------
+2
Figure 3. Relationship between z-score and standard deviation in a normal distribution
Z-scores range from -3 o (falling to the far left of the normal distribution curve) to +3 o (falling to the far
right of the normal distribution curve).
x — [I
a
Where:
• z is the z-score;
• x is the value of the individual analytical result;
• (j, is the population mean across all laboratories for that analyte; and
• o is the standard deviation of the population mean of that analyte.
For this study, when:
• z < 2 the analytical result is satisfactory
o 95% of the z-scores are expected to fall in this range for normally distributed data
• 2 < z < 3 the analytical result is considered questionable
o should be investigated by the laboratory
• z > 3 the analytical result is unsatisfactory
Happily, all z-scores for this study were less than 2. See the tables below for a summary of all results.
Results
For the elemental analysis by XRF, results from the top ten CSN wide average elemental concentrations
from June 2016 through May 2018 were included in the interlaboratory comparison. In descending order
by average concentration, these elements shown in Figure 2, below, and are sulfur (S), silicon (Si), iron
(Fe), potassium (K), sodium (Ns), calcium (Ca), aluminum (Al), chlorine (CI), magnesium (Mg), and zinc
(Z).
Page 4 of 8
-------
Network-wide average elemental concentrations
June 2016 - May 2018
• 0.3647
Si
Fe-
K-
Na-
Ca-
Al-
Cl-
Mg"
Zn-
0.1038
0.0912
~ 0.0697
0.0670
0.0623
—~ 0.0398
— 0.0307
0.0163
0.0136
Figure 4. Top-ten CSN-wide average elemental concentrations
As shown in Table 1, below, all z-scores for the selected elements by XRF were below 2.
Table 1. Selected Elements by XRF: Interlaboratory z-score Results
Lab
Event
S
Si
Fe
K
Na
Ca
Al
CI
Mg
Zn
DRI
1
0.82
1.14
0.85
0.50
1.48
0.68
0.70
1.19
0.77
1.26
ODEQ
1.43
0.52
1.44
1.14
0.70
1.44
1.46
0.19
0.77
1.19
SCAQMD
0.10
1.02
0.37
1.06
0.48
0.67
0.59
1.23
1.33
0.04
UCD
0.51
0.63
0.22
0.58
0.30
0.08
0.16
0.15
0.22
0.04
DRI
2
0.71
1.28
0.39
0.17
1.50
0.30
0.81
1.12
0.85
0.62
ODEQ
1.47
0.31
1.47
1.44
0.59
1.49
1.23
0.97
0.85
1.49
SCAQMD
0.22
0.91
0.77
0.85
0.46
0.63
0.81
0.70
1.11
0.43
UCD
0.53
0.67
0.31
0.41
0.45
0.55
0.39
0.55
0.58
0.45
DRI
3
0.90
0.78
0.45
0.58
1.50
0.39
1.50
0.61
0.87
0.77
ODEQ
1.40
0.85
1.49
1.49
0.53
1.49
0.55
1.50
0.87
1.47
SCAQMD
0.01
1.23
0.65
0.32
0.47
0.61
0.52
0.44
0.87
0.35
UCD
0.49
0.40
0.40
0.59
0.50
0.49
0.43
0.45
0.86
0.34
DRI
B1
0.00
1.22
0.50
0.45
1.50
1.12
0.06
0.58
0.02
0.87
ODEQ
0.00
0.40
0.50
0.52
0.52
0.84
0.69
0.58
0.81
0.87
SCAQMD
0.00
0.90
0.50
0.52
0.52
0.84
0.69
1.49
1.40
0.88
UCD
0.00
0.73
1.50
1.50
0.46
0.57
1.43
0.33
0.61
0.85
DRI
B2
0.00
1.12
0.43
0.39
0.50
1.43
0.50
0.62
0.82
1.37
ODEQ
0.00
0.57
0.54
0.55
0.50
0.88
0.50
0.62
0.82
0.74
SCAQMD
0.00
0.83
0.54
0.55
0.50
0.42
0.50
1.48
1.21
0.74
UCD
0.00
0.86
1.50
1.50
1.50
0.13
1.50
0.24
0.43
0.12
Page 5 of 8
-------
For cation and anion analysis by IC, filters were extracted and analyzed for the cations sodium (Na),
ammonium (NH4+), potassium (K), and anions chloride (CI ), nitrate (NO3), and sulfate (SO42). Note that
CI" was added to the analyte list several years ago because there was an interest in quantifying the impact
of sea spray on PM2 5. Desert Research Institute (DRI) has additional samples because they perform these
analyses on both nylon and Teflon filters, so both filter types were included with their PE samples.
Additionally, the DRI B2 Teflon PE sample was found to have two filters adhered together. DRI
extracted these samples separately and reported both results, which were included in the data analysis. As
shown in Table 2, below, all z-scores were below 2.
Results reported by laboratories as "ND" or "
-------
Table 2. Cations and Anions by IC: Interlaboratory z-scorc Results
Lab
Event
Na
NH4+
K
CI
I\I03-
S042"
CARB
1
NS
0.90
0.03
NR
0.85
0.07
DRI
0.86
1.27
0.96
0.70
0.08
0.41
DRI
1.16
0.05
1.36
0.95
1.96
1.65
ODEQ
0.63
1.54
0.33
NR
0.44
1.10
RTI
0.17
0.21
0.67
0.48
0.50
0.10
SCAQMD
1.22
0.38
1.30
1.17
0.08
0.94
CARB
2
1.47
0.28
0.49
NR
0.96
0.13
DRI
0.10
0.55
1.05
0.37
0.03
1.02
DRI
1.55
1.92
1.38
1.49
1.89
1.52
ODEQ
0.12
0.49
0.39
NR
0.59
0.23
RTI
0.26
0.56
0.32
0.62
0.36
0.35
SCAQMD
0.57
0.61
1.22
0.50
0.01
1.21
CARB
3
0.09
0.25
0.28
NR
0.77
0.33
DRI
0.39
0.37
0.74
0.03
0.12
0.08
DRI
1.95
1.52
1.38
1.34
1.94
1.49
ODEQ
0.03
1.50
0.26
NR
0.58
0.88
RTI
0.61
0.17
0.79
1.06
0.50
0.44
SCAQMD
0.83
0.47
1.34
0.25
0.21
1.30
CARB
B1
NS
NS
NS
NR
NS
NS
DRI
1.03
0.58
0.48
0.71
0.59
0.52
DRI
0.66
0.58
0.67
0.98
0.56
1.15
ODEQ
NS
NS
NS
NR
NS
NS
RTI
0.64
NS
NS
0.59
NS
NS
SCAQMD
1.05
1.15
1.15
1.09
1.15
0.64
CARB
B2
NS
NS
NS
NR
NS
NS
DRI
0.43
0.00
0.02
0.37
0.38
0.50
DRI
0.44
0.00
1.11
0.56
0.87
0.50
DRI
1.14
0.00
0.23
0.56
0.76
0.50
ODEQ
NS
NS
NS
NR
NS
NS
RTI
NS
NS
NS
NS
1.50
NS
SCAQMD
0.00
0.00
0.00
0.00
0.00
0.00
NS: Not scored because reported result was either ND or
-------
For organic carbon analysis, results for organic carbon (OC), elemental carbon (EC), and total carbon
(TC) were analyzed and compared across laboratories. Note that one lab was not provided a filter for
Event 1, so was not included in that comparison. As shown in Table 3, below, all z-scores were below 2.
Table 3. Carbon by TOA: Interlaboratory z-score Results
Lab
Event
OC
EC
TC
CARB
1
0.77
1.40
0.22
DRI
1.23
0.11
1.18
SCAQMD
0.86
0.96
1.22
UCD
0.40
0.33
0.26
CARB
2
0.52
0.87
0.32
DRI
1.72
1.48
1.59
ORD
0.53
0.45
0.80
SCAQMD
0.70
0.16
0.80
UCD
0.01
0.90
0.32
CARB
3
0.32
1.41
1.13
DRI
1.35
1.41
0.60
ORD
1.42
0.09
1.52
SCAQMD
0.04
0.11
0.02
UCD
0.21
0.02
0.23
CARB
B1
0.05
0.59
0.06
DRI
0.92
1.72
0.91
ORD
0.92
0.59
0.93
SCAQMD
0.44
0.59
0.44
UCD
1.45
0.05
1.45
CARB
B2
0.62
0.45
0.62
DRI
0.96
0.45
0.96
ORD
0.44
0.45
0.44
SCAQMD
1.44
0.45
1.44
UCD
0.66
1.79
0.65
Page 8 of 8
------- |