Chemical Speciation Network (CSN)
Annual Quality Report
Samples Collected January 1, 2018 through December 31, 2018
Prepared for:
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Research Triangle Park, NC 27711
EPA Contract No. EP-D-15-020
Prepared by:
Air Quality Research Center
University of California, Davis
One Shields Avenue
Davis, CA 95616
January 27, 2020
UCDAVIS
AIR QUALITY RESEARCH CENTER
-------
Table of Contents
1. Executive Summary 4
1.1 Introduction 4
1.2 Data Quality Overview and Issues 4
2. Summary of Laboratory Operation Issues 5
2.1 DRI Ion Analysis Laboratory 5
2.1.1 Analysis Delays 5
2.2 RTI Ion Analysis Laboratory 5
2.2.1 Analysis Delays 5
2.2.2 Laboratory Transition 5
2.3 UC Davis X-Ray Fluorescence Laboratory 7
2.3.1 Application Change 7
2.3.2 Zinc 7
2.3.3 Calcium 7
2.4 DRI Thermal Optical Analysis Laboratory 7
2.4.1 Analysis Delays 7
2.4.2 QC Criteria Failures 7
2.5 UC Davis Thermal Optical Analysis Laboratory 8
2.5.1 Analysis Delays 8
2.5.2 Laboratory Transition 8
3. Quality Issues and Corrective Actions 10
3.1 Data Quality 10
3.1.1 Completeness 10
3.1.2 Comparability and Analytical Precision 12
3.1.3 Blanks 14
3.2 Corrective Actions 24
3.2.1 Elemental Analysis 24
3.2.2 Ion Analysis 25
3.2.3 Carbon Analysis 25
3.2.4 Data Processing 25
4. Laboratory Quality Control Summaries 26
4.1 .A DRI Ion Chromatography Laboratory 26
4.1 .A. 1 Summary of QC Checks and Statistics 26
4.1. A.2 Summary of QC Results 27
4.1 .A.3 Determination of Uncertainties and Method Detection Limits 33
4.1.A.4 Audits, Performance Evaluations, Training, and Accreditations 33
4.1.A.5 Summary of Filter Field Blanks 34
4.1 .B RTI Ion Chromatography Laboratory 34
4.1 .B. 1 Summary of QC Checks and Statistics 35
4.1.B.2 Summary of QC Results 35
4.1 .B.3 Determination of Uncertainties and Method Detection Limits 58
4.1.B.4 Audits, Performance Evaluations, Training, and Accreditations 59
4.1.B.5 Summary of Filter Field Blanks 59
4.2 UC Davis X-Ray Fluorescence Laboratory 60
Page 2 of 126
-------
4.2.1 Summary of QC Checks and Statistics 60
4.2.2 Summary of QC Results 61
4.2.3 Determination of Uncertainties and Method Detection Limits 77
4.2.4 Audits, Performance Evaluations, Training, and Accreditations 78
4.2.5 Summary of Filter Field Blanks 78
4.3.A DRI Thermal Optical Analysis Laboratory 80
4.3.A.1 Summary of QC Checks and Statistics 80
4.3.A.2 Summary of QC Results 81
4.3.A.3 Determination of Uncertainties and Method Detection Limits 94
4.3.A.4 Audits, Performance Evaluations, Training, and Accreditations 94
4.3.A.5 Summary of Filter Blanks 95
4.3.B UC Davis Thermal Optical Analysis Laboratory 95
4.3.B.1 Summary of QC Checks and Statistics 96
4.3.B.2 Summary of QC Results 97
4.3.B.3 Determination of Uncertainties and Method Detection Limits 107
4.3.B.4 Audits, Performance Evaluations, Training, and Accreditations 107
4.3.B.5 Summary of Filter Blanks 108
5. Data Management and Reporting 108
5.1 Number of Events Posted to AQS 108
6. Quality Assurance and Data Validation 109
6.1 QAPP Revisions 109
6.2 SOP Revisions 109
6.3 Summary of Internal QA Activities 109
6.4 Data Validation and Review 110
6.4.1 Summary of Monthly Data Validation Review Results 110
6.5 Uncertainty Estimates and Collocated Precision Summary Statistics 119
7. References 125
Page 3 of 126
-------
1. Executive Summary
1.1 Introduction
The University of CaliforniaDavis (UC Davis) Air Quality Research Center summarizes
quality assurance (QA) annually in this report as a contract deliverable for the Chemical
Speciation Network (CSN) program (contract #EP-D-15-020). The primary objectives of this
report are:
1. Provide the U.S. Environmental Protection Agency (EPA) and other potential users with
graphical and tabular illustrations of quality control (QC) for species measured within the
network.
2. Identify and highlight observations of interest that may have short- or long-term impact
on data quality across the network or at particular sites.
3. Serve as a record and tool for ongoing UC Davis QA efforts.
Each network site includes two samplers: (1) URG 3000N carbon sampler (URG Corporation;
Chapel Hill, NC) for collection of particulate matter on quartz filters; and (2) Met One SASS or
SuperSASS (Met One Instruments, Inc; Grants Pass, OR) for collection of particulate matter on
polytetrafluoroethylene (PTFE) filters and nylon filters. The following analyses are performed:
PTFE filters: filters are analyzed at UC Davis using energy dispersive X-ray fluorescence
(EDXRF) for a suite of 33 elements.
Nylon filters: for samples collected through September 30, 2018, filters were analyzed at
the Desert Research Institute (DRI) using ion chromatography (IC) for a suite of six ions.
For samples collected beginning October 1, 2018, filters are analyzed at Research
Triangle Institute International (RTI) using IC for a suite of six ions.
Quartz filters: for samples collected through September 30, 2018, filters were analyzed at
the Desert Research Institute (DRI) for organic and elemental carbon including carbon
fractions using Thermal Optical Analysis (TOA). For samples collected beginning
October 1, 2018, filters are analyzed at UC Davis for organic and elemental carbon
including carbon fractions using TOA.
Unless otherwise noted, data and discussions included in this report cover samples collected
during the time period January 1, 2018 through December 30, 2018.
1.2 Data Quality Overview and Issues
Section 4 of this report provides laboratory performance details for each of the analytical
measurement techniques. The laboratory performance is detailed in Section 4.1. A (DRI Ion
Chromatography Laboratory, covering analysis of samples collected January 1, 2018 through
September 30, 2018), Section 4.1.B (RTI Ion Chromatography Laboratory, covering analysis of
samples collected October 1, 2018 through December 31, 2018), Section 4.2 (UC Davis X-Ray
Fluorescence Laboratory), Section 4.3.A (DRI Thermal Optical Analysis Laboratory, covering
analysis of samples collected January 1, 2018 through September 30, 2018), and Section 4.3.B
(UC Davis Thermal Optical Analysis Laboratory, covering analysis of samples collected October
1, 2018 through December 31, 2018).
Page 4 of 126
-------
Across the network, completeness determined by the total number of valid samples relative to
the total number of scheduled samples was 96.3% for PTFE filters, 96.2% for nylon filters,
and 93.7%) for quartz filters. As detailed in Section 3.1.1, there were seven sites with less than
75%o completeness for at least one filter type.
No Technical Systems Audit (TSA) of UC Davis was performed by the EPA in 2018.
2. Summary of Laboratory Operation Issues
2.1 DRI Ion Analysis Laboratory
2.1.1 Analysis Delays
Some deliveries of analysis data from DRI to UC Davis were delayed, contributing to
noncompliance with the 120-day requirement for delivery of data to AQS following receipt of
filters by analytical laboratories. See Section 5.1.
2.2 RTI Ion Analysis Laboratory
2.2.1 Analysis Delays
UC Davis issued a subcontract to RTI for ions analysis of filters beginning with samples
collected October 1, 2018. The subcontract analysis laboratory transition from DRI to RTI
resulted in some initial analysis delays. Deliveries for samples collected in October (132 days)
and December 2018 (125 days) were noncompliant with the 120-day requirement for delivery of
data to AQS following receipt of filters by analytical laboratories. See Section 5.1.
2.2.2 Laboratory Transition
Beginning with samples collected October 1, 2018, nylon filters are analyzed for ions using Ion
Chromatography (IC) at RTI. They were previously analyzed for ions using IC at DRI. At the
network level there is no evidence of a step change in the ion concentrations associated with the
laboratory transition (see Figure 2.2-1).
Different filter extraction methods were used by RTI and DRI. RTI performed filter extraction
with one hour of sonication followed by eight hours on a shaker table in a cold room (RTI SOP
Ionsl); DRI performed filter extraction with one hour of sonication followed by one hour on a
shaker table (DRI SOP #2-109r7). UC Davis will continue to closely monitor and evaluate data
to identify changes that may be associated with the laboratory transition.
Page 5 of 126
-------
Figure 2.2-1: Monthly network wide results for ions; data from samples collected January 1, 2018 through
December 31, 2018. Samples collected January 1, 2018 through September 30, 2018 were analyzed by DRI (red
boxes) and samples collected October 1, 2018 through December 30, 2018 were analyzed by RTI (blue boxes). The
thick horizontal lines indicate median and the upper and lower limits of the boxes represent 75th and 25th percentile,
respectively. The whiskers extend to 1.5 xIQR (where IQR is the interquartile range, or the distance between the 25th
and the 75th percentiles).
E
o) i.o
i r
1049 1065 1071 1080 1060 1072 1193 1063 1069 1063 1042 1104
Number indicates count of samples per month
E
o> 0.15
'lo ฐ '1ฐ
8 0.05
c
o
O
Ejzl DRI
RTI
20181 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
$ I
$ I
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 201810 2018 11 2018 12
1
1 = ~ n ~ :
1 t
[ L-r-l 11 t_j_j 1=^ 1T' I"
1
Ejzl DRI
RTI
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
Potassium Ion
$ I
$ I
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
Ejzl DRI
RTI
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
$ I
$ I
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
Page 6 of 126
-------
2.3 UC Davis X-Ray Fluorescence Laboratory
2.3.1 Application Change
The XRF analysis conditions, including the secondary targets and integration times (collectively
referred to as the application), were changed in December 2018 during the XRF instrument
calibrations. The changes were made to lower the detection limits and the variability in some
elements as well as to reduce the overall bias between instruments.
For further details, see Section 4.2.2.5.
2.3.2 Zinc
For analyses performed during this reporting period, periodic zinc contamination was observed
on the daily QC laboratory blank and daily QC multi-elemental reference sample on the EDXRF
instruments, XRF-1 and XRF-4. The cause of these random contamination events was
determined to be related to the instrument design, specifically operation of the sample changer.
Samples analyzed during this period were checked for unusually high zinc mass loadings
compared to site specific and network wide historical values. Nine samples in 2018 with unusual
Zn mass loadings were investigated, with seven of those resulting in reanalysis. Reanalysis
results for one of the cases indicated contamination during the original analysis; the reanalysis
results for this case were reported.
For further detail see Sections 3.2.1.1 and Section 4.2.2.1.
2.3.3 Calcium
During this reporting period, XRF-1, XRF-4, and XRF-5 showed gradual increase in calcium
mass loadings of their QC samples. The calcium buildup was likely caused by atmospheric
deposition or instrument wear on these filters, which are analyzed daily and remain in the
instruments' sample changers indefinitely. This gradual buildup of calcium is not expected on
actual samples which are loaded and analyzed once. However, samples are monitored for
unusually high calcium values and reanalyzed as necessary. During this reporting period there
was one case of reanalysis request because of suspected calcium contamination. The reanalysis
confirmed that contamination was not present and the original results were reported.
For further detail see Section 3.2.1.2 and Section 4.2.2.1.
2.4 DRI Thermal Optical Analysis Laboratory
2.4.1 Analysis Delays
Some deliveries of analysis data from DRI to UC Davis were delayed, contributing to
noncompliance with the 120-day requirement for delivery of data to AQS following receipt of
filters by analytical laboratories. See Section 5.1.
2.4.2 QC Criteria Failures
In some cases, DRI analyzed samples while instruments were operating outside of the defined
QC criteria. There are instances of impacted data for samples collected during 2018.
Per direction from the EPA, these data were redelivered to AQS with the QX (Does Not Meet
QC Criteria) qualifier flag applied.
Page 7 of 126
-------
For further detail see Section 3.2.3.1 and Section 4.3.A.2.
2.5 UC Davis Thermal Optical Analysis Laboratory
2.5.1 Analysis Delays
UC Davis began TOA analysis of filters beginning with samples collected October 1, 2018. The
analysis laboratory transition from subcontractor DRI to UC Davis resulted in some initial
analysis delays. Deliveries for samples collected in October (132 days) and December 2018 (125
days) were noncompliant with the 120-day requirement for delivery of data to AQS following
receipt of filters by analytical laboratories. See Section 5.1.
2.5.2 Laboratory Transition
Beginning with samples collected October 1, 2018, quartz filters are analyzed for carbon using
Thermal Optical Analysis (TOA) at UC Davis. They were previously analyzed for carbon using
TOA at DRI. At the network level, there is no evidence of a step change in the organic carbon
(OC) or elemental carbon (EC) concentrations associated with the laboratory transition, but the
EC to OC ratios are slightly elevated after the transition, especially at higher percentiles (Figure
2.5-1).
Page 8 of 126
-------
Figure 2.5-1: Monthly network wide results for organic carbon by reflectance (OCR), elemental carbon by
reflectance (ECR) and ratio of ECR to OCR (ECR/OCR); data from samples collected January 1, 2018 through
December 31, 2018. Samples collected January 1, 2018 through September 30, 2018 were analyzed by DRI (red
boxes) and samples collected October 1, 2018 through December 30, 2018 were analyzed by UC Davis (blue
boxes). The thick horizontal lines indicate median, and the upper and lower limits of the boxes represent 75th and
25th percentile, respectively. The whiskers extend to 1.5 xIQR (where IQR is the interquartile range, or the distance
between the 25th and the 75th percentiles).
OCR
DRI
Ejzl UCD
2018 1
2018 2
2018 3 2018 4 2018 5 2018 6 2018 7 2018
2018 9 2018 10 2018 11 2018 12
ECR
E
D)
a.
c
o
TO
CD
" 0.
O
O
DRI
Ejzl UCD
2018 1 2018 2 2018 3 2018 4 2018 5 2018 6 2018 7 2018 8 2018 9 2018 10 2018 11 2018 12
ECR/OCR
DRI
Ejzl UCD
2018 1 20182 20183 20184 20185 20186 2018 7 20188 20189 2018 10 2018 11 2018 12
Individual sites within the network are also evaluated in an effort to identify changes in data
trends that may be related to the laboratory transition. A distinct change in the ECR
concentration corresponding to the laboratory transition is observed at the Charleston NCore site
(AQS ID #54-039-0020; Figure 2.5-2). The change is also pronounced in the highest temperature
elemental carbon fraction (e.g. EC3) and is likely related to small differences in the operating
temperatures of the instruments.
Page 9 of 126
-------
Figure 2.5-2: Time series of elemental carbon by reflectance (ECR) and EC3 concentrations at the Charleston
NCore site (AQS ID #54-039-0020); data from samples collected January 1, 2017 through December 31, 2018.
ECR
ฆ
ฆ
ฆ
. #
ฆ
m
ฆ
%
ฆ
*
.
#
*
ฆ
. v
ฆ
ฆ
. '
%
b
m
m
.*
m
%
ฆ
. ,
Jan 2017 Mar2017 May 2017 Jul2017 Sep2017 Nov2017 Jan2018 Mar2018 May2018 Jul 2018 Sep2018 Nov2018 Jan2019
EC3
ฆ
ai
m
ฆ
s
i
v.-.
t .
'
' .*
.
Vซซ
p
ฆ
ft
a.
^TT
'v
;. .
Jan 2017 Mar2017 May 2017 Jul 2017 Sep2017 Nov2017 Jan2018 Mar2018 May 2018 Jul 2018 Sep2018 Nov2018 Jan2019
3. Quality Issues and Corrective Actions
3.1 Data Quality
3.1.1 Completeness
Completeness is evaluated network wide by filter type, and determined by the total number of
valid samples relative to the total number of collected and scheduled samples (Table 3.1-1). The
completeness is comparable for PTFE and nylon filters which are both collected by the Met One
SASS / Super SASS sampler; however, the number of invalid samples is higher for quartz filters,
which are collected by the URG sampler. Quartz filters flagged with the QX qualifier, as detailed
in Section 2.4.2, were not invalidated and are included in the count of valid samples.
Table 3.1-1: Network sample completeness by filter type, January 1, 2018 through December 31, 2018. The total
number of scheduled samples is calculated from the sampling schedule (does not include field blanks). The total
number of collected samples is the actual number of samples collected in the field.
Filter
Type
Total Number
of Scheduled
Samples
Total Number
of Collected
Samples
Number
of Valid
Samples
Number
of Invalid
Samples
% Valid
(relative to #
collected samples)
% Valid
(relative to # of
scheduled samples)
PTFE
13,410
13,400
12,918
482
96.4
96.3
Nylon
13,410
13,400
12,894
506
96.2
96.2
Quartz
13,410
13,397
12,567
830
93.8
93.7
Page 10 of 126
-------
Across the network there were seven sites with sample completeness less than 75% for at least
one filter type (Table 3.1-2). Five of the seven cases had low completeness resulting from invalid
quartz filters.
Table 3.1-2: Network sites with less than 75% sample completeness (relative to the number of collected samples,
and determined for null codes applied at the filter level) for at least one filter type, January 1, 2018 through
December 30, 2018. For each filter type, the percentage of different null codes is listed relative to the total number
of null codes per site. For null code definitions, see Table 3.1-3.
AQS ID#
Location
Completeness (%)
Null Codes
PTFE
Nvlon
Quartz
PTFE
Nylon
Quartz
12-011-0034-5
(Region 4)
Broward County, FL
(NCore/STN)
74.6%
74.6%
86.1%
AN (29%)
BA (29%)
Other (42%)
AN (29%)
BA (29%)
Other (42%)
BA (53%)
Other (47%)
15-003-0010-5
(Region 9)
Kapolei, HI
78.5%
73.6%
82.0%
AO (27%)
AB (19%)
Other (54%)
AO (22%)
AH (19%)
Other (59%)
AH (36%)
AN (23%)
Other (41%)
17-031-4201-5
(Region 5)
Northbrook, IL
93.4%
94.3%
71.3%
AO (38%)
AV (38%)
Other (25%)
AO (43%)
AV (43%)
SV (14%)
AH (94%)
AO (6%)
32-003-0540-5
(Region 9)
Jerome Mack Middle
School, NV
99.2%
99.2%
68.9%
AH (100%)
AH (100%)
AH (90%)
AK (5%)
AN (5%)
42-045-0109-5
(Region 3)
Marcus Hook, PA
98.4%
98.4%
62.3%
AG (100%)
AG (100%)
AH (96%)
AN (4%)
49-049-4001-5
(Region 8)
Lindon, UT
100.0%
100.0%
42.6%
...
...
AH (100%)
72-021-0010-5
(Region 2)
Bayamon, Puerto Rico
(NCore/STN)
95.0%
95.0%
24.8%
AV (50%)
AF (33%)
BJ (17%)
AV (50%)
AF (33%)
BJ (17%)
AH (90%)
AV (6%)
Other (4%)
Samples can be invalidated for a variety of reasons, as detailed in the UCD CSN TI801C and the
Data Validation for the Chemical Speciation Network guide. Null codes indicate the reasons for
invalidation (Table 3.1-3).
Page 11 of 126
-------
Table 3.1-3: Number and type of null codes applied at the filter level to SASS and URG samples from January 1,
2018 through December 30, 2018. Codes are ordered by frequency of occurrence.
Null
(ode
SASS
ptii:
SASS
\\lon
I R(.
Qusirl/
Null ( ode Description
AU
0
0
1
Monitoring Waived
AS
0
0
2
Poor Quality Assurance Results
AW
l
1
0
Wildlife Damage
BI
0
0
2
Lost or damaged in transit
AK
1
0
3
Filter Leak
AZ
2
2
0
Q C Audit
AM
0
3
2
Miscellaneous Void
AC
2
2
2
Construction/Repairs in Area
BE
2
2
2
Building/Site Repair
BB
3
3
2
Unable to Reach Site
SA
4
4
4
Storm Approaching
AL
4
4
6
Voided by Operator
AI
6
6
3
Insufficient Data (cannot calculate)
SV
2
2
12
Sample Volume Out of Limits
AQ
3
7
7
Collection Error
AR
4
6
14
Lab Error
AB
19
19
14
Technician Unavailable
AG
21
21
18
Sample Time out of Limits
AJ
26
31
4
Filter Damage
AO
24
24
15
Bad Weather
BA
15
15
42
Maintenance/Routine Repairs
BJ
52
59
28
Operator Error
AF*
49
49
71
Scheduled but not Collected
AV
78
78
66
Power Failure
AN
104
105
166
Machine Malfunction
AH
67
70
355
Sample Flow Rate or CV out of Limits
* Records that receive this flag can be associated with events that were not sampled.
3.1.2 Comparability and Analytical Precision
Analytical precision is evaluated by comparing data from repeat analyses, where two analyses
are performed on the same sample using either the same instrument (duplicate) or different
instruments (replicate). Reliable laboratory measurements should be repeatable with good
precision. Analytical precision includes only the uncertainties associated with the laboratory
handling and analysis, whereas collocated precision (Section 6.5) also includes the uncertainties
associated with sample preparation, field handling, and sample collection. Analytical precision is
used internally as a QC tool.
Comparisons of ion mass loadings from repeat analyses (replicates and/or duplicates) on nylon
filters analyzed by IC show agreement (Figure 3.1-1).
Page 12 of 126
-------
Figure 3.1-1: Ion repeat analysis (replicates and/or duplicates) results; data from samples collected January 1, 2018
through December 31, 2018. Samples collected January 1, 2018 through September 30, 2018 were analyzed by DRI
(red points) and samples collected October 1 2018 through December 31, 2018 were analyzed by RTI (blue points).
Ammonium
Chloride
Nitrate
w
w 60
_>*
TO
TO 40
*->
CO
15 on
o_ 20
0)
CD
0-
y = -0.1 +
r2 = 0.99
1.01/
if
$
m
> /
/
/
/
/
d)
CL
D) 40-
c SO-
TS
CD
O 20"
CO
CO 10-
0-
30-
20-
10-
0-
0 + 1x
&
/
/
/
/
&
/
Jr
/
/
/
/
0 10 20 30
Sodium Ion
15-
10-
5-
0-
y = - 0 + 1x /
r2 = 1 /
0 20 40 60
Potassium Ion
y = 0.01 + 0.99x /
r2 = 0.99
/
/
/
/
/
/
/
J*
/
"I 1 1 1 T1 '"I 1 1 V
0 10 20 30 40 0 5 10 15
150-
100 -
50-
0-
y = 0.01 + 1x
r2 = 1 '
0 50 100 150
Sulfate
40-
20-
0-
>^=-0.01+ 1x
/
/
/
r2 = 1
/
~
Laboratory
DRI
RTI
Mass Loading (|ag per filter): routine filter
Comparison of carbon mass loadings from repeat analyses (replicates and/or duplicates) on
quartz filters analyzed by TOA generally show agreement (Figure 3.1-2).
Page 13 of 126
-------
Figure 3.1-2: Carbon repeat analysis (replicates and/or duplicates) results; data from samples collected during
January 1.2018 through December 31, 2018. Samples collected January 1, 2018 through September 30, 2018 were
analyzed by DR1 (red points) and samples collected October 1, 2018 through December 31, 2018 were analyzed by
UC Davis (blue points). Elemental carbon (EC) fractions are indicated as (1) through (3), organic carbon (OC)
fractions are indicated as (1) through (4). Organic pvrolyzed (OP), elemental carbon (EC), and organic carbon (OC)
are shown by reflectance (R) and transmittance (T).
EC1
EC2
EC3
ECR
400
300
200
100
0
to
w
^ 150
CD
ro 100
0)
Q.
a) 50
y = -0.29 + 1.02x
r2 - 0.97 '
P
-------
handling and analysis, and adsorption of gases during sampling and handling. Additionally, field
blanks are used to calculate method detection limits (MDLs; see Section 3.1.3.2).
There is some variability in field blank mass loadings by species and month, as shown in Figure
3.1-3 through 3.1-8 for ions measured from nylon filters, and Figure 3.1-9 and 3.1-10 for organic
carbon and elemental carbon, respectively, measured from quartz filters. The 10th percentile of
network sample concentrations is indicated in Figure 3.1-3 through Figure 3.1-10 to facilitate
understanding of field blank concentrations in context of network sample concentrations; 90% of
network sample concentrations fall above the indicated 10th percentile. As part of the validation
process (see Section 6), field blank outliers are investigated but are only invalidated if there is
cause to do so. Artifact correction (Section 3.1.3.1) and MDL (Section 3.1.3.2) calculation
methods are robust to accommodate occasional outliers.
For most species there does not appear to be a step change in the field blank time series
corresponding with the October 2018 laboratory transitions (see Section 2.2.2 and Section 2.5.2),
though ammonium (Figure 3.1-3) and potassium ion (Figure 3.1-6) median field blank mass
loadings appear slightly elevated, which may be related to the laboratory transition. Nitrate field
blank mass loadings were elevated in February 2018 (Figure 3.1-5) corresponding with
temporary use of Pall Ultipore N66 nylon filters, which were used when the supply of MTL
nylon filters was depleted.
Figure 3.1-3: Time series of ammonium measured on nylon filter field blanks (FB), for field blanks collected
January 1, 2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were
collected. The colored (red, 2016; green. 2017; blue, 2018) horizontal lines indicate median, and the upper and
lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme
data point that is no more than 1.5 xIQR (where IQR is the interquartile range, or the distance between the 25th and
the 75th percentiles). The dots are all of the points that lay outside the whiskers. Black vertical dotted line indicates
laboratory transition from DRI to RTI. The black horizontal dashes indicate the 10th percentile of network samples.
4
3
10th %-ile of
network samples
U)
3
Field blanks
(year)
$ 2016
$ 2017
$ 2018
1
0 = -J-i 4^
2016 01 2016 07 2017 01 2017 07 2018 01 2018 07
Time (Year Month)
Page 15 of 126
-------
Figure 3.1-4: Time series of chloride measured on nylon filter field blanks (FB), for field blanks collected January
1. 2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were
collected. The colored (red, 2016; green 2017; blue, 2018) horizontal lines indicate median, and the upper and
lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme
data point that is no more than 1.5 xIQR (where IQR is the interquartile range, or the distance between the 25th and
the 75th percentiles). The dots are all of the points that lay outside the whiskers. Black vertical dotted line indicates
laboratory transition from DRI to RTI. The black horizontal dashes indicate the 10th percentile of network samples.
T
T
5E
2016 01 2016 07 2017 01 2017 07 2018 01 2018 07
Time (Year Month)
10th %-ile of
network samples
Field blanks
(year)
$ 2016
$ 2017
$ 2018
Page 16 of 126
-------
Figure 3.1-5: Time series of nitrate measured on nylon filter field blanks (FB), for field blanks collected January 1,
2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were collected.
The colored (red, 2016; green. 2017; blue, 2018) horizontal lines indicate median, and the upper and lower limits of
the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme data point that is
no more than 1.5>
-------
Figure 3.1-6: Time series of potassium ion measured on nylon filter field blanks (FB), for field blanks collected
January 1, 2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were
collected. The colored (red, 2016; green 2017; blue, 2018) horizontal lines indicate median, and the upper and
lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme
data point that is no more than 1.5 xIQR (where IQR is the interquartile range, or the distance between the 25th and
the 75th percentiles). The dots are all of the points that lay outside the whiskers. Black vertical dotted line indicates
laboratory transition from DRI to RTI. The black horizontal dashes indicate the 10th percentile of network samples.
xi-iili:
is
2016 01 2016 07
2017 01 2017 07 2018 01
Time (Year Month)
2018 07
10th %-ile of
network samples
Field blanks
(year)
$ 2016
$ 2017
$ 2018
Page 18 of 126
-------
Figure 3.1-7: Time series of sodium ion measured on nylon filter field blanks (FB), for field blanks collected
January 1, 2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were
collected. The colored (red, 2016; green 2017; blue, 2018) horizontal lines indicate median, and the upper and
lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme
data point that is no more than 1.5 xIQR (where IQR is the interquartile range, or the distance between the 25th and
the 75th percentiles). The dots are all of the points that lay outside the whiskers. Black vertical dotted line indicates
laboratory transition from DRI to RTI. The black horizontal dashes indicate the 10th percentile of network samples.
10th %-ile of
network samples
Field blanks
(year)
$ 2016
$ 2017
$ 2018
2016 01 2016 07 2017 01 2017 07 2018 01
Time (Year Month)
2018 07
Page 19 of 126
-------
Figure 3.1-8: Time series of sulfate measured on nylon filter field blanks (FB), for field blanks collected January 1,
2016 through December 31, 2018. Gaps in time series are present when no nylon filter field blanks were collected.
The colored (red, 2016; green 2017; blue, 2018) horizontal lines indicate median and the upper and lower limits of
the boxes represent 75th and 25th percentile, respectively. The whiskers extend to the most extreme data point that is
no more than 1.5>
-------
Figure 3.1-9: Time series of organic carbon by reflectance (OCR) measured on quartz filter field blanks (FB), for
field blanks collected January 1, 2016 through December 31, 2018. Gaps in time series are present when no quartz
filter field blanks were collected. The colored (red, 2016; green, 2017; blue, 2018) horizontal lines indicate median,
and the upper and lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers extend to
the most extreme data point that is no more than 1.5>
| 30
Ui
O)
c
T3
TO
O
_l
O)
-------
Figure 3.1-10: Time series of elemental carbon by reflectance (ECR) measured on quartz filter field blanks (FB),
for field blanks collected November 20, 2015 through December 31, 2018. Gaps in time series are present when no
quartz filter field blanks were collected. The colored (red, 2016; green, 2017; blue, 2018) horizontal lines indicate
median, and the upper and lower limits of the boxes represent 75th and 25th percentile, respectively. The whiskers
extend to the most extreme data point that is no more than 1.5>
-------
both field and laboratory processes, thus it is expected that field blank mass loadings are
generally higher than lab blanks which have only been handled in a laboratory environment and
have less opportunity for mishandling and contamination. When the MDL determined from field
blanks is lower than the analytical MDL (calculated by the laboratories using laboratory blanks),
the analytical MDL is assigned as a floor value.
The average MDLs calculated for this reporting period (samples collected January 1, 2018
through December 31, 2018) are compared to those calculated using the same method from the
previous reporting period (samples collected February 1, 2017 through December 31, 2017)
(Table 3.1-4). MDLs calculated during this reporting period compare well with those from the
previous reporting period for many species. However, there are some cases where 2018 MDLs
are lower (improved) or higher (degraded): (1) elemental species Ca and K 2018 MDLs are both
lower relative to 2017 MDLs; (2) ion species sodium ion and sulfate 2018 MDLs are higher and
lower, respectively, relative to 2017 MDLs; (3) most carbon species and fractions have higher
2018 MDLs relative to 2017 MDLs. MDL differences may be related to changes in filter media
cleanliness, EDXRF application changes (see Section 2.3.1 and Section 4.2.2.5), and laboratory
transitions (see Section 2.2.2 and Section 2.5.2).
Table 3.1-4: Average method detection limits (MDLs) and percentage of reported data above the MDLs for all
species, calculated for data from samples collected February 1, 2017 through December 31, 2017 (previous
reporting period) and January 1, 2018 through December 31, 2018 (current reporting period). Elemental carbon (EC)
fractions are indicated as (1) through (3), organic carbon (OC) fractions are indicated as (1) through (4). Organic
pyrolyzed (OP), elemental carbon (EC), and organic carbon (OC) are shown by reflectance (R) and transmittance
(T). Species shown in bold have differences >50% between those reported for the previous reporting period (2017)
and the current reporting period (2018).
Species
2017 (previous reporting period)
2018 (current reporting period)
Average MDL, jig/mJ
% Above MDL
Average M DL, fig/m3
% Above MDL
Ag
0.017
2.7
0.016
2.6
A1
0.038
30.1
0.032
37.0
As
0.003
6.2
0.002
4.3
Ba
0.081
1.5
0.080
1.3
Br
0.005
15.8
0.005
12.0
Ca
0.034
56.8
0.018
74.1
Cd
0.016
2.9
0.016
3.2
Ce
0.096
1.3
0.095
1.3
CI
0.007
31.9
0.005
38.2
Co
0.003
0.9
0.003
0.9
Cr
0.004
20.1
0.003
25.4
Cs
0.056
2.5
0.054
2.2
Cu
0.011
10.9
0.011
9.8
Fe
0.027
79.8
0.018
88.7
In
0.037
0.0
0.038
0.0
K
0.012
98.4
0.005
99.2
Mg
0.042
12.5
0.043
14.4
Mn
0.006
8.3
0.006
8.2
Na
0.088
20.9
0.089
23.5
Ni
0.002
14.7
0.002
16.2
Page 23 of 126
-------
Species
20I"7 (pre\ious reporting period)
20IS (ciirrcnl reporting period)
\\erage M 1)1-jig/m'
"i. \bo\e MM.
\\erage M 1)1.. ug/nr1
"i. \bo\e MM.
P
0.002
8.0
0.002
7.3
Pb
0.012
7.4
0.012
6.9
Rb
0.009
0.3
0.009
0.2
S
0.005
99.5
0.004
99.5
Sb
0.040
2.3
0.039
1.8
Se
0.005
1.6
0.005
1.4
Si
0.020
83.7
0.016
82.3
Sn
0.050
0.5
0.049
0.4
Sr
0.007
2.2
0.007
1.7
Ti
0.003
41.4
0.003
44.1
V
0.002
7.9
0.001
6.7
Zn
0.003
78.9
0.003
79.6
Zr
0.036
1.0
0.036
0.7
Ammonium
0.006
81.6
0.005
95.2
Chloride
0.047
60.1
0.036
71.1
Nitrate
0.036
98.5
0.035
99.0
Potassium Ion
0.047
29.5
0.061
10.4
Sodium Ion
0.016
66.6
0.026
73.5
Sulfate
0.047
99.4
0.025
99.6
Elemental Carbon (EC1)
0.007
99.5
0.015
99.9
Elemental Carbon (EC2)
0.009
95.5
0.017
97.7
Elemental Carbon (EC3)
0.002
3.6
0.003
22.2
Elemental Carbon (ECR)
0.013
99.4
0.018
99.8
Elemental Carbon (ECT)
0.012
98.9
0.016
99.7
Organic Carbon (OC1)
0.019
76.8
0.015
84.9
Organic Carbon (OC2)
0.036
99.5
0.035
99.8
Organic Carbon (OC3)
0.053
98.7
0.077
96.1
Organic Carbon (OC4)
0.012
99.7
0.034
96.1
Organic Carbon (OCR)
0.081
99.6
0.134
99.5
Organic Carbon (OCT)
0.083
99.6
0.138
99.6
Organic Pyrolyzed (OPR)
0.008
72.4
0.022
78.8
Organic Pyrolyzed (OPT)
0.010
93.9
0.028
94.2
3.2 Corrective Actions
To ensure ongoing quality work, UC Davis reacts as quickly and decisively as possible to
unacceptable changes in data quality. These reactions are usually in the form of investigations,
and, if necessary, corrective actions. The following subsections describe significant corrective
actions undertaken for data from samples collected during 2018.
3.2.1 Elemental Analysis
3.2.1.1 Zinc
Page 24 of 126
-------
As discussed in Section 2.3.2 and Section 4.2.2.1, the design of the sample changer arm on the
EDXRF instruments results in sporadic cases of zinc contamination. During this reporting
period, seven filters identified as having potential contamination were reanalyzed.
3.2.1.2 Calcium
As discussed in Section 2.3.3 and Section 4.2.2.1 laboratory QC filters that are exposed to the
environment for prolonged periods for repeat analysis show a general increase in calcium mass
loadings. These increases are not observed if the QC filter is cleaned with air or replaced with a
new filter. The contamination appears to occur mostly on filters that are analyzed multiple times
and therefore should not impact routine samples or field blanks. Even so, CSN sample and field
blank filters were monitored during QA checks for calcium contamination. During this reporting
period, one filter identified as having potential contamination was reanalyzed.
3.2.1.3 Cr and Ni Contamination
UC Davis identified a potential Cr and Ni contamination issue that impacts data from prior to the
contract transition (November 20, 2015) through this reporting period. The sampler modules may
be the source of contamination and are being investigated. Wipes from inside the sampling
modules were collected and analyzed by ICP-MS at RTI. Additionally, screens and screws from
the sampling modules were analyzed by EDXRF at UC Davis. Results are forthcoming and the
investigation is ongoing.
3.2.2 Ion Analysis
No corrective actions during this reporting period.
3.2.3 Carbon Analysis
3.2.3.1 QC Criteria Failures
As discussed in Section 2.4.2, in some instances, DRI analyzed samples while instruments were
operating outside of the defined QC criteria.
Per direction from the EPA, these data will be redelivered to AQS with QX (Does Not Meet QC
Criteria) qualifier flag applied.
For further detail see Section 2.4.2 and Section 4.3.A.2.
3.2.3.2 Data Flagging (LJ Flag)
UC Davis identified a potential source of uncertainty in the OC and EC split upon thermogram
review. Specifically, for heavily loaded quartz samples or samples that contain light-absorbing
materials that volatilize at a lower temperature, pyrolysis of OC does not lead to any more light
absorbed by the sample deposit (i.e. no further decrease in the laser signals). As a result, the laser
signal does not return to its initial value, thus no OC/EC split can be automatically determined.
In these cases, the split is forced to correspond with the system switch to the oxidizing
environment, and any pyrolyzed OC (i.e. OP) is quantified as EC. The reported OC is the lower
limit and EC is the upper limit. Beginning with samples collected November 2018, UC Davis has
started applying the 'LJ' AQS flag (LJ: Identification of analyte is acceptable; Reported value is
an estimate) for these cases. 5.1% (64 out of 1249) and 5.7% (77 out of 1353) of the samples
collected in November and December 2018 have the LJ flag applied, respectively.
3.2.4 Data Processing
Page 25 of 126
-------
3.2.4.1 Data Flagging Modifications
Data are flagged as part of the CSN data validation process as detailed in the UCD CSN TI801C
and the Data Validation for the Chemical Speciation Network guide. Flags are applied
throughout the sampling, filter handling, analysis, and validation processes, using automated
checks and on a case-by-case basis. The use and application of flags evolves as problems are
identified and remedied, and also in response to process improvements that are implemented to
improve the quality and consistency of data for the end user.
4. Laboratory Quality Control Summaries
4.1.A DRI Ion Chromatography Laboratory
The DRI Ion Chromatography Laboratory, as a subcontractor to UC Davis, received and
analyzed nylon filters from batches 39 through 47 covering the sampling period January 1, 2018
through September 30, 2018. Analysis of these samples was performed June 14, 2018 through
December 14, 2018. Using ion chromatography, DRI analyzed for both anions (i.e., chloride [CI"
], nitrate [NO3"], and sulfate [SO42"]) and cations (i.e., sodium [Na ], ammonium [NH4+], and
potassium[K+]) using three DIONEX ICS-5000+ systems (Chow and Watson, 2017) and two
DIONEX ICS-6000 systems and reported the results of those analyses to UC Davis.
4.1.A.1 Summary of QC Checks and Statistics
Samples were received by the DRI Ion Chromatography Laboratory following the chain-of-
custody procedures specified in DRI SOP #2-117. Samples were analyzed using DIONEX ICS-
5000+ or ICS-6000 Systems following DRI SOP #2-228 for anions and DRI SOP #2-229 for
cations. QC measures for the DRI ion analysis are summarized in Table 4.1. A -1. The table
indicates the frequency and standards required for the specified checks, along with the
acceptance criteria and corrective actions.
During daily startup, an eight-point calibration was performed over the range from 0.02 to 3.0
|ig/mL (i.e., 0.02, 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 3.0 |ig/mL) before beginning analysis. Then
two deionized-distilled water (DDW) samples and a method blank were analyzed, followed by
two types of QC control standards: (1) 1.0-2.5 |ig/mL QC standards diluted from NIST certified
Dionex standard solutions; and (2) DRI-made control standards (i.e., 1.00 |ig/mL CI", 1.00
|ig/mL NO3", 1.00 |ig/mL SO42" for anions and 0.39 |ag/mL NH4+ and 1.03 |ag/mL Na+ for
cations). During routine analysis, after every 10 samples, one duplicate, one DDW, and a
selected QC standard (same as calibration solution concentrations; diluted from certified
Environmental Research Associates (ERA) stock solutions) at various concentrations (0.1-3.0
|ig/mL) were analyzed.
Page 26 of 126
-------
Table 4.1.A-1: DRI quality control measures for ion (anion and cation) analysis by ion chromatography.
Requirement
Multipoint
Calibration
F requeney
Daily or every batch of-100,
whichever comes first
Cal i 1)ration Stand a rd
NIST certified ERA
Aceeptance Criteria
ฑ 10% of certified
value
Corrcetive Aetion
Identify and correct problem before
analyzing samples; and recalibrate
Minimum
Detection Limit
(MDL)a
Initially, then annually or after
major instrument maintenance
Nylon filter lab blanks
(7 or more)
Within ฑ 10% of
previous instrument
limit
Troubleshoot instrument and check
filter lots
DDW
Four initially to establish
background, followed by one
every 10 samples
DDW with resistance >
18 MQ
Within 3 standard
deviations of MDLs3
Verify instrument response to DDW
without extraction
Method blank b
One for every 40 samples
DDW with resistance >
18 MQ
Within 3 standard
deviations of MDLs3
Check instrument response for DDW
with extraction
QC Control
Standards
Daily or every run
DRI-made or Dionex
NIST-certified multi-
component standard
solution
ฑ 10% of listed value
Rerun the QC standard and reanalyze
samples between this standard and
previous QC standard
QC Check
Standards
Every 10 samples
NIST-certified multi-
component standard
solution from ERA
ฑ 10% of listed value
Reanalyze samples between this
standard and previous check standard
Duplicates0
10% of samples
N/A
ฑ 10% when value >
10 x MDLa
Reanalysis of duplicate sample
a MDL indicated here is an internal laboratory QA indicator, distinct from the MDL reported to AQS.
b 15 mL DDW solution that follows the same extraction procedure as the sample extraction.
0 Duplicate indicates analysis results obtained from two different aliquots of the same filter sample extract analyzed
on the same instrument.
4.1.A.2 Summary of QC Results
Table 4.1-1 outlines corrective actions for failed QC checks. For failed method blanks,
instrument malfunction was ruled out first. Next, the blank was reanalyzed to rule out
contamination during the extraction process and within the IC system. For the cases of failed
method blanks in Table 4.1.A-2, reanalysis of the blanks resulted in concentrations below QC
threshold and sample data are not affected. In the cases where the Dionex and DRI-made QC
control standards (Tables 4.1.A-3 and 4.1.A-4; run after the multipoint calibration and before
sample analysis) failed to pass the acceptance criteria, the multipoint calibration, the QC control
standard, and any samples that were analyzed were rerun to ensure that the QC standards passed
acceptance criteria. For cases where the ERA QC check standards failed (Table 4.1.A-5;
analyzed every 10th sample), all samples between the failed standard and the nearest previous
passing QC standard were reanalyzed. Reported sample data all passed acceptance criteria for the
QC standards. Duplicate analyses (Table 4.1.A-6) that exceeded acceptance criteria were
reanalyzed and compared to the original analysis. If the second duplicate met acceptable
tolerance, the first duplicate data point was considered spurious and was replaced. If the second
duplicate analysis did meet tolerance standards, all ten samples in the set were reanalyzed.
Sample data are not affected by reanalyzing duplicates.
4. l.A. 2.1 Method Blanks
Table 4.1. A-2 lists the number of method blanks analyzed during this reporting period and their
concentration statistics. Both median and average concentrations are near or below the MDLs
Page 27 of 126
-------
(MDL indicated here is an internal laboratory QA indicator, distinct from the MDL reported to
AQS).
Table 4.1.A-2: Method blank counts and concentrations for all reported ions for the analysis period 6/14/2018
through 12/14/2018 (samples collected 1/1/2018 through 9/30/2018).
Inns
CI
NOr
so;
Na*
NH4
k:
( CHI III
286
286
286
286
286
286
Median (,u^/ml.)
0.001
0.003
0.000
0.000
0.000
0.000
A\erase (fi$*/ml.)
0.001
0.004
0.001
0.001
0.000
0.000
Si. I)e\. (uii/iiil.)
0.002
0.005
0.006
0.001
0.000
0.001
Mill (,us/nil.)
0.000
0.000
0.000
0.000
0.000
0.000
Max Uiii/ml.)
0.021
0.039
0.088
0.010
0.002
0.011
# l.\eee(l3xM 1)1.'
0
4
1
0
0
0
a MDL indicated here is an internal laboratory QA indicator, distinct from the MDL reported to AQS.
4.1.A. 2.2 QC Control Standards and Check Standards
Table 4.1. A-3 and Table 4.1. A-4 list the analysis statistics for Dionex and DRI-made ion QC
control standards, respectively. The control charts of these analyses are shown in Figure 4.1.A-1.
The average difference between the measured and nominal concentrations are within the ฑ10%
limit (Table 4.1.A-1), although a few individual checks failed the 10% acceptance criteria.
Corrective actions for failed analyses are shown in Table 4.1.A-1. Table 4.1.A-5 summarizes
analysis statistics for the ERA QC check standards at different concentration levels. Some
individual standards failed QC criteria, but were reanalyzed following the procedure outlined in
Table 4.1.A-1. All reported CSN sample ion concentrations passed the QC control and check
standard verification.
Table 4.1.A-3: Statistics for Dionex ion QC control standards for the analysis period 6/14/2018 through 12/14/2018
(samples collected 1/1/2018 through 9/30/2018).
Ions
Nominal
Uiii/ml.)
( Oil III
Median
(,iis/ml.)
A\erase
(iili/inl.)
Min
(,us/iul->
Max
Ulli/llll.)
# Kail
A\e " i.
Ueeo\ er\
" i. Si. I)e\.
CI
1.000
121
0.971
0.972
0.900
1.027
0
97.2%
2.1%
NO.
1.000
121
0.931
0.933
0.901
1.085
0
93.3%
2.6%
SO,-
1.000
121
0.966
0.972
0.927
1.092
0
97.2%
2.6%
Na
1.000
122
0.975
0.975
0.931
1.031
0
97.5%
1.5%
Mli
1.250
122
1.308
1.302
1.157
1.348
0
104.2%
2.8%
k
2.500
122
2.523
2.525
2.294
2.737
0
101.0%
6.0%
Page 28 of 126
-------
Table 4.1.A-4: Statistics for DRI-made ion QC control standards for the analysis 6/14/2018 through 12/14/2018
(samples collected 1/1/2018 through 9/30/2018).
Ions
Nominal
(HK/mL)
Count
Median
(Hซ/mL)
Average
(Hซ/mL)
Mill
(jig/mL)
Max
(fi${/niL)
#Fail
Ave %
Reeoverv
% St. Dev.
CI
1.000
110
0.997
1.000
0.928
1.122
1
100.0%
2.5%
NO.
1.000
110
0.967
0.970
0.903
1.090
0
97.0%
3.5%
SOr
1.000
110
0.954
0.957
0.901
1.088
0
95.7%
2.6%
\a
1.030
105
1.006
1.009
0.960
1.076
0
100.9%
1.9%
Mli
0.390
105
0.380
0.379
0.338
0.405
1
97.1%
1.0%
k
0.000
105
0.000
0.001
0.000
0.006
0
NA
0.1%
aNA=Not applicable
Table 4.1.A-5: Statistics for ERA QC control standards for the analysis period 6/14/2018 through 12/14/2018
(samples collected 1/1/2018 through 9/30/2018).
Ion
Nominal
(,uu/ml.)
Coiinl
Median
(,iiu/ml.)
A\era lie
(,iiu/ml.)
Min
(,iiu/ml.)
Max
Ulli/llll.)
A\e"'i.
Ueeo\ er\
" i. Si. I)e\.
CI
0.1
18
0.086
0.089
0.082
0.099
88.6%
0.6%
0.2
84
0.180
0.183
0.161
0.318
91.3%
1.7%
0.5
293
0.467
0.469
0.278
0.548
93.7%
2.1%
1
277
0.967
0.969
0.869
1.107
96.9%
3.5%
2
218
1.992
2.002
1.849
2.329
100.1%
6.5%
3
231
3.031
3.046
2.806
3.488
101.5%
9.2%
NO.
0.1
18
0.083
0.086
0.076
0.099
85.6%
0.7%
0.2
84
0.175
0.176
0.138
0.216
87.8%
1.4%
0.5
293
0.454
0.453
0.255
0.544
90.5%
2.5%
1
277
0.937
0.941
0.822
1.098
94.1%
3.4%
2
218
1.982
1.990
1.744
2.345
99.5%
7.5%
3
231
3.039
3.056
2.740
3.438
101.9%
10.9%
SOr
0.1
18
0.088
0.092
0.082
0.106
91.7%
0.8%
0.2
84
0.185
0.184
0.152
0.213
91.9%
1.4%
0.5
293
0.470
0.471
0.324
0.584
94.2%
2.4%
1
277
0.957
0.962
0.841
1.149
96.2%
3.8%
2
218
1.987
1.991
1.768
2.455
99.6%
6.8%
3
231
3.033
3.043
2.757
3.690
101.4%
10.4%
\a
0.1
18
0.073
0.074
0.058
0.092
73.9%
1.0%
0.2
79
0.175
0.175
0.143
0.202
87.6%
1.2%
0.5
291
0.462
0.463
0.404
0.512
92.5%
1.1%
1
278
0.973
0.975
0.861
1.088
97.5%
2.4%
2
222
1.999
2.005
1.812
2.219
100.3%
3.8%
3
234
3.025
3.036
2.703
3.352
101.2%
6.1%
Mli
0.1
18
0.083
0.082
0.073
0.094
82.1%
0.6%
Page 29 of 126
-------
Ion
Noiniiiiil
(uii/inl.)
Coiinl
M i'(l i;in
Ulli/llll.)
A\CTilliC
(11 Vl/in I )
Min
(fiU/ml.)
M;i\
Uili/inl.)
Ucc(i\ er\
" i. Si. I)e\.
0.2
79
0.186
0.186
0.167
0.211
93.1%
0.8%
0.5
291
0.500
0.498
0.449
0.537
99.6%
1.2%
1
278
1.008
1.005
0.824
1.065
100.5%
2.3%
2
222
1.992
1.992
1.772
2.159
99.6%
4.1%
3
234
3.004
3.018
2.812
3.528
100.6%
7.7%
k
0.1
18
0.076
0.080
0.069
0.099
79.7%
1.1%
0.2
79
0.186
0.186
0.142
0.252
92.8%
1.9%
0.5
291
0.486
0.483
0.390
0.571
96.6%
2.8%
1
278
1.011
1.002
0.811
1.114
100.2%
3.8%
2
222
1.996
1.994
1.459
2.169
99.7%
7.3%
3
234
3.001
3.026
2.791
3.606
100.9%
11.1%
Page 30 of 126
-------
Figure 4.1.A-la: Control charts for Dionex ion QC control standards for the analysis period 6/14/2018 through
12/14/2018 (samples collected 1/1/2018 through 9/30/2018). The limits are ฑ10% of the nominal concentrations (red
dashed lines).
1.15
g 1.10
O)
3 1.05
c
o
1.00
o 0.90
O 0.85
0.80
CI" (Nominal 1.000 |jg/mL)
.1-
V
1.15
1.10
1.05
1.00
0.95
0.90
0.85
0.80
N03- (Nominal 1.000 |jg/mL)
W
'-Srvg
.v
ป
1.15
_l
1.10
ฃ
O)
3.
1.05
C
O
1.00
TO
C
0.95
a)
u
c
0.90
o
o
c
0.85
O
0.80
Analysis Date
S042' (Nominal 1.000 jjg/mL)
a-
*
1.15
_i
1.10
ฃ
O)
3.
1.05
C
O
5
1.00
C
a)
0.95
o
c
o
0.90
o
c
o
0.85
0.80
Analysis Date
Na+ (Nominal 1.000 jjg/mL)
tJC
1.40
1.35
1.30
1.25
1.20
1.15
1.10
Analysis Date
NH4+ (Nominal 1.25 jjg/mL)
aA,
"
sAi
O
o
2.8
2.7
2.6
2.5
2.4
2.3
2.2
Analysis Date
K+ (Nominal 2.5 jjg/mL)
y^,.V= ~
i .. " *ซ
*
:?
COCOCOCOCOCOCOCO
OOOCOCOI^I^i^
CD h- CO O) ฆ* O
Analysis Date
Analysis Date
Page 31 of 126
-------
Figure 4.1.A-lb: Control charts for DRI-made ion QC control standards for the analysis period 6/14/2018 through
12/14/2018 (samples collected 1/1/2018 through 9/30/2018). The limits are ฑ10% of the nominal concentrations (red
dashed lines), except for K+ which is 3 xMDLa (red dashed lines).
1.15
1.10
1.05
1.00
0.95
0.90
0.85
0.80
CI- (Nominal 1.000 ng/mL)
-J"
s
1.15
1.10
1.05
1.00
0.95
0.90
0.85
0.80
N03- (Nominal 1.000 |jg/mL)
v
1.15
1.10
_l
ฃ
O)
1.05
3.
C
1.00
O
TO
0.95
C
a)
u
0.90
c
o
o
0.85
c
o
0.80
Analysis Date
S042' (Nominal 1.000 jjg/mL)
i
..1
o
o
1.15
1.10
1.05
1.00
0.95
0.90
0.85
0.80
Analysis Date
Na+ (Nominal 1.000 jjg/mL)
0.44
_i
ฃ
0.42
O)
3.
0.40
C
O
0.38
TO
C
0.36
0)
U
c
o
0.34
o
c
O
0.32
0.30
Analysis Date
NH4+ (Nominal 0.390 (jg/mL)
:<
0.020
0.015
0.010
0.005
0.000
-0.005
-0.010
-0.015
-0.020
Analysis Date
K+ (Nominal 0.000 jjg/mL)
Analysis Date Analysis Date
1MDL indicated here is an internal laboratory QA indicator, distinct from the MDL reported to AQS.
Page 32 of 126
-------
4. l.A. 2.3 Duplicate Analyses
Table 4.1.A-6 gives the criteria and summary statistics for duplicate analysis results. Duplicate
analysis results are obtained from two different aliquots of the same filter sample extract run on
the same instrument. The criteria used for each ion were that 1) if the average concentration was
less than 10 times the lower quantifiable limit (LQL), the absolute value of the average
difference should be less than ten times the LQL, and 2) if the average concentration was greater
than or equal to ten times the LQL, then the relative percent different difference (RPD) should be
less than 10%. LQLs are given in Tables 4.1.A-7a and 4.l.A-7b. The LQLs are used as internal
QA indicators, distinct from the MDLs reported to AQS. A total of 1,010 duplicate analyses
were run for samples taken during the reporting period, excluding samples with field or analysis
flags.
Table 4.1.A-6: Ion duplicate analysis criteria and statistics for the analysis period 6/14/2018 through 12/14/2018
(samples collected 1/1/2018 through 9/30/2018).
Range
Criteria
Statistic
Na
nh4
K
CI
NOj
SO/
U nits
All
Count
1010
1010
1010
1010
1010
1010
Count
995
305
961
994
777
364
No. Fail
0
0
0
0
0
0
%Fail
0
0
0
0
0
0
%
Ion< 10 x
10 x
RPDa
Mean
1.0%
2.5%
2.8%
0.7%
0.8%
1.2%
RPD
LQL
<10%
St. Dev.
1.7%
2.7%
3.5%
0.7%
1.0%
1.5%
RPD
Max
7.0%
12.0%
13.5%
2.9%
6.8%
9.3%
RPD
Min
0.0%
0.0%
0.0%
0.1%
0.0%
0.0%
RPD
Median
0.6%
1.4%
1.5%
0.6%
0.4%
0.7%
RPD
aRPD= 100 x absolute value [original sample - duplicate sample] / [(original sample + duplicate sample) / 2]
4.1.A.3 Determination of Uncertainties and Method Detection Limits
For discussion of Method Detection Limits (MDLs) see Section 3.1.3.2.
For discussion of analytical uncertainty and total uncertainty see Section 3.1.2 and Section 6.5,
respectively.
4.1.A.4 Audits, Performance Evaluations, Training, and Accreditations
4. l.A. 4.1 System Audits
UC Davis contracted a third-party auditor (Technical & Business Systems; Placerville, CA) to
perform a Laboratory Systems Audit of the DRI Ion Chromatography Laboratory. The audit was
Page 33 of 126
-------
conducted on September 19, 2018. No issues were identified that affected data quality; auditors
provided minor recommendations for improved documentation and tracking, and assured QA/
QC documentation was in agreement with existing procedures.
4. l.A. 4.2 Performance Evaluations
No performance evaluations were reported during the time period.
4.1.A.4.3 Training
All new laboratory staff receive training for performing tasks described in the SOPs relevant to
their assigned work.
4. l.A. 4.4 A ccreditations
There are no accreditations for analysis of ions on aerosol filters by Ion Chromatography.
4.1.A.5 Summary of Filter Field Blanks
Over the sampling period (January 1, 2018 through September 30, 2018) there were 1,250 valid
nylon filter field blanks. Table 4.1-7a and Table 4.1-7b summarize the field blank statistics.
Table 4.1.A-7a: Nylon filter field blank statistics in ng/mL for the analysis period 6/14/2018 through 12/14/2018
(samples collected 1/1/2018 through 9/30/2018).
Ions
Coiinl
Median
(,iiii/ml.)
A\era tie
Ulli/llll.)
Min
(.11^/1111 - >
Max
Ulli/llll.)
S(. I)e\.
(Uli/llll.)
CI
1250
0.007
0.011
0.000
0.264
0.018
NO3
1250
0.014
0.021
0.000
0.221
0.020
SCV-
1250
0.004
0.007
0.000
0.188
0.012
Na+
1250
0.002
0.004
0.000
0.200
0.013
NH4+
1250
0.000
0.001
0.000
0.009
0.001
K+
1250
0.000
0.001
0.000
0.022
0.002
Table 4.1.A-7b: Nylon filter field blank statistics in |ig/filter (extraction volume 15 mL) for the analysis period
6/14/2018 through 12/14/2018 (samples collected 1/1/2018 through 9/30/2018).
Ions
(oil III
Median
(fiji/l'illen
A\era tie
(fiji/l'illen
Min
(fiii/l'illen
Max
(fiii/l'illen
S(. I)e\.
(fiii/l'illen
CI
1250
0.098
0.165
0.000
3.956
0.274
NO3
1250
0.204
0.315
0.000
3.313
0.307
SCV-
1250
0.056
0.108
0.000
2.819
0.183
Na+
1250
0.025
0.066
0.000
3.003
0.196
NH4+
1250
0.006
0.014
0.000
0.132
0.021
K+
1250
0.000
0.010
0.000
0.326
0.026
4.1.B RTI Ion Chromatography Laboratory
The RTI Ion Chromatography Laboratory, as a subcontractor to UC Davis, received and
analyzed extracts from nylon filters for batches 48 through 50, covering the sampling period
October 1, 2018 through December 30, 2018. Analysis of these samples was performed January
Page 34 of 126
-------
25, 2019 through March 26, 2019. Using ion chromatography, RTI analyzed for both anions (e.g.
chloride [CI"], nitrate [NO3"], and sulfate [SO42"]) and cations (e.g. sodium [Na+], ammonium
[NH4+], and potassium[K+]) using five Thermo Dionex ICS systems (three anion systems, two
cation systems) and reported the results of those analyses to UC Davis.
4.1.B.1 Summary of QC Checks and Statistics
Samples are received by the RTI Ion Chromatography Laboratory following the chain-of-
custody procedures specified in RTI SOP #Ionsl. Samples are analyzed using Thermo Dionex
ICS-2000 and ICS-3000 systems following RTI SOP #Ionsl. Extraction procedures are
documented on worksheets which are maintained with the batch files. The QC measures for the
RTI ion analysis are summarized in Table 4.1.B-1. The table details the frequency and standards
required for the specified checks, along with the acceptance criteria and corrective actions.
Stated QC criteria are verified and documented on review worksheets, and reviewers document
QC criteria not met, corrective actions, samples flagged for reanalysis, and subsequent reanalysis
dates.
Table 4.1.B-1: RTI quality control measures for ion (anion and cation) analysis by ion chromatography.
Ko(|iiiiviiioiil
l"iV(|ueno
Acceplance ( rilcriii
( ปnvcli\e
Action
Calibration regression
Daily
R2 >0.999
Investigate;
Repeat
calibration
Continuing calibration
verification (CCV)
check standard; RTI
dilution of a
commercially
prepared, NIST-
traceable QC sample
Daily, immediately after
calibration and at every 10
samples
Measured concentrations < 0.050 ppm:
within 35% of known values.
Measured concentrations >0.050 ppm:
within 10% of known values.
Investigate;
reanalyze
samples
Duplicate sample
3 per batch of 50 samples
RPD = 10% at lOx MDL
RPD = 200% at MDL
Investigate;
reanalyze
Spiked sample extract
2 per batch of 50 samples
Recoveries within 90 to 110% of target
values
Investigate;
reanalyze
Reagent blanks
One reagent blank per reagent
used (DIH20 and/or eluent)
At least one per day
No limit set. The data is compiled for
comparability studies. < 10 times MDL
Investigate;
reanalyze
Round Robin
(External QA by
USGS)
4/month
Not applicable; data reported and
compared annually
Investigate
Reanalysis
5% per batch reanalyzed on
different day and as requested
MDL to 10 times MDL: RPD up to 200%,
10 to 100 times MDL: RPD < 20%,
>100 times MDL: differences within 10%
Investigate from
batch reanalyze
samples if
needed
4.1.B.2 Summary of QC Results
RTI followed the quality control criteria stated in Table 4.1.B-1. Instruments were recalibrated
when calibration failed to meet the criteria. For cases where CCV failures occurred during
analyses, samples bracketed by the CCV failure were reanalyzed. When duplicate precision or
spiked sample recoveries failed to meet the criteria, the duplicated samples or matrix spike
sample plus additional samples (5% of analytical batch) were reanalyzed. The original data were
only replaced with reanalysis data in cases where precision between the reanalysis and original
Page 35 of 126
-------
result failed to meet the criteria. For cases where check samples failed to meet the reanalysis
criteria, the remaining samples not already reanalyzed from the batch were reanalyzed.
4. l.B. 2.1 Calibration regression
Ion chromatographs are calibrated daily with calibration standards prepared as serial dilutions of
a NIST traceable stock standard. Anion instruments are calibrated from 10 - 2000 parts per
billion (ppb) for chloride and from 50 - 10000 ppb for nitrate and sulfate. A high calibration
standard at 5000 ppb for chloride and 25000 ppb for sulfate and nitrate are used in the calibration
curve only for samples exceeding 10000 ppb. Cation instruments are calibrated from 10 - 1000
ppb for sodium, ammonium and potassium. A high calibration standard at 3000 ppb is used only
for samples whose concentrations exceed 1000 ppb. The correlation coefficients for the daily
calibration must be at least 0.999. If this criterion is not met, the curve is investigated. A
calibration standard or standards that are suspect are removed from the curve and not used for
calculations. If the calibration still fails to meet the stated QC criteria, the situation is further
investigated until it has been confirmed that the instrument is performing correctly.
After calibration, an analytical sequence is assigned to an instrument and includes 50 batch
samples, extraction QC checks, three sets of replicate samples, two matrix spikes, and continuing
calibration verification (CCV) standards.
4.1.B.2.2 Continuing calibration verification (CCV) check standard
Instrument QC samples are used to verify the initial and continuing calibration of the ion
chromatographs. These solutions are prepared at the low, medium, medium-high and high end of
the calibration curve. Table 4.1.B-2 and 4.1.B-3 lists the concentrations.
Table 4.1.B-2: Target concentrations for anion CCV check standards for the analysis period 1/25/2019 through
3/26/2019 (samples collected 10/1/2018 through 12/31/2018).
QC Sample
CI (ppb)
NOa (ppb)
SO/ (ppb)
Insli'iiinenl Low QC
200
600
1200
1 iislru 1110111 Medium QC
500
1500
3000
luslrnmeiil Medium High QC
1000
3000
6000
lusl ru meiil lliiili QC
2000
6000
12000
Table 4.1.B-3: Target concentrations for cation CCV check standards for the analysis period 1/25/2019 through
3/26/2019 (samples collected 10/1/2018 through 12/31/2018).
QC Sample
Na+ (ppb)
NH4+ (ppb)
K+ (ppb)
Inslnimenl Low QC
20
20
20
1 iisl ru men 1 Medium QC
250
250
250
Inslnimenl Medium lli^h QC
750
750
750
Inslriimenl lliiili QC
2000
2000
2000
At least two CCV check standards are analyzed immediately after the calibration standards and a
single CCV check standard is analyzed after every ten samples throughout the batch. When an
instrument CCV check standard falls outside of the control limits, impacted samples are
reanalyzed. If a CCV check standard fails, and there is a second CCV check standard measured
immediately following the failure, samples are not reanalyzed. The failed CCV check standard,
Page 36 of 126
-------
samples flagged for reanalysis, and date of reanalysis are documented on the review worksheet
and maintained with the complete set of batch records for each batch analyzed.
Control charts were prepared for anion (Figure 4.1.B-1) and cation (Figure 4.1.B-2) CCV check
standards. Most CCV check standards were within the stated control limits. There were two
cases where CCV check standards failed the QC criteria: (1) one of these was a medium-high
CCV check standard for sodium, and the impacted samples were reanalyzed; (2) the other was a
low CCV check standard for potassium, the impacted samples were not reanalyzed because there
was a second CCV check standard that was successful.
For the purpose of demonstrating instrument to instrument performance, control charts for the
lowest CCV check standards were generated, where instruments A9 and A10 were compared for
anions (Figure 4.1.B-3) and instruments C3 and C6 were compared for cations (Figure 4.1.B-4).
The control charts illustrate consistent performance between instruments.
Figure 4.1.B-1: Control charts for anion CCV check standards at low, medium, medium-high, and high
concentrations measured in ng/mL (see Table 4.1.B-2) forthe analysis period 1/25/2019 through 3/26/2019
(samples collected 10/1/2018 through 12/31/2018). Red lines show upper and lower control limits set at ฑ10% of the
nominal concentrations for the low, medium, medium-high, and high standards. Blue lines show upper and lower
warning limits.
230 i
220
= 210
1 200
S 190 -
0 180 -
Instrument Low QC - CI
170
a*4*aฑ *11'
iiMi t ป ,
*4 iiiii
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19 3/31/19
Analysis Date
Instrument Low QC - N03
660 -
640 -
620
600
580
560 -
540 -
520 -
A A A 1 a A
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19 3/31/19
Analysis Date
Page 37 of 126
-------
Instrument Low QC - S04
1320 -
F
1270 -
3
C
o
1220
(TJ
1170 -
1475 -
c
a
1425 -
c
o
u
1375 -
1325
kA a*aAa i ฑA
M AiAA*
A A A A
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19 3/31/19
Analysis Date
Instrument Medium QC - S04
3275 -
F
3175 -
3
C
o
3075 -
CD
2975 -
*->
-------
Instrument Medium-High QC - CI
_i 1090 -
E
= 1040 -
c
o
g 990 -
c
1990 -
c
1940 -
1890
u
1840 -
1790 -
1/15/19
Instrument High QC - CI
h " *
A AAAA
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Page 39 of 126
-------
Instrument High QC - N03
6590
J
E
6390
"53
D
6190
s
g
5990
c
d)
5790
<-ป
C
o
5590
o
5390
'iiiuuiiUiiii
"aa aAAAA a
aAA^I
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
13290 -|
ฃ 12790 -
M
3
C
O
flj
ฃ 11790 -
ซ
o 11290
10790
1/15/19
Instrument High QC - S04
A AAAA
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Figure 4.1.B-2: Control charts for cation CCV check standards at low, medium, medium-high, and high
concentrations measured in ng/mL (see Table 4.1.B-3) forthe analysis period 1/25/2019 through 3/26/2019
(samples collected 10/1/2018 through 12/31/2018). Red lines show upper and lower control limits set at ฑ35% of the
nominal concentrations for the low standards and ฑ10% of the nominal concentrations for the medium, medium-
high. and high standards. Blue lines show upper and lower warning limits.
30 n
25
20
a 15 -
10
1/15/19
Instrument Low QC - NH4
iiiirstibipr "
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19
4/5/19
Page 40 of 126
-------
Instrument Low QC - Na
30
E
aB 25
20
a 15 -I
10
1/15/19
iiiit 4 lit** J1 S;
1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
30 -i
E
~SB 25 -
o
'5 20
15 -
10
1/15/19
Instrument Low QC - K
A
1/25/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
280 -I
E 270 -
260
250
ฃ 240
o 230 -
o
220
1/15/19
Instrument Medium QC - NH4
~
^ li<(! i' >'.
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
280 -I
E 270 -
260
250
S 240
o 230 -
u
220
1/15/19
Instrument Medium QC - Na
^ ** j
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Page 41 of 126
-------
Instrument Medium QC - K
280
1
F
270 -
nn
=3
260 -
C
O
250 -
c
0)
240 -
c
o
u
230
220 -
:!ia' % a,, i: ;
~ I
1/15/19 1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
850
, 800
o
ฃ 750 A
700
650
1/15/19
Instrument Medium High - NH4
* A*
J-44** if** ^
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
850
i
Is 800
3
C
O
tt 750
c 700
o
650
1/15/19
Instrument Medium-High QC - Na
5i'i4 4 aii* ^ (i ** 4
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
850
. 800
o
S 750 H
700
650
1/15/19
Instrument Medium High QC - K
iiV ** *aaa a *
A ^A A A * * 1
Ai*
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Page 42 of 126
-------
2250 -i
1
fc
2150 -
w>
3
C
o
2050 -
(0
1950 -
0>
u
O
1850 -
u
1750
1/15/19
Instrument High QC - NH4
A
^ " :a :
A A V a/aa a
i aa i
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
2250 -|
_i
2150 -
S3
| 2050 -
nj
| 1950 -
<0
u
ง 1850 -
o
1750
1/15/19
Instrument High QC - Na
'jli' A *A ii'u "
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
2250 -i
-j
b
2150 -
3
C
o
2050 -
fD
ฃ
1950 -
QJ
1850 -
u
1750
1/15/19
Instrument High QC - K
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
4/5/19
Page 43 of 126
-------
Figure 4.1.B-3: Control charts for anion and cation CCV check standards showing comparability between
instruments (A9 and A10 for anions; C3 and C6 for cations) at low concentrations (see Table 4. l.B-2 and Table
4.1.B-3) for the analysis period 1/25/2019 through 3/26/2019 (samples collected 10/1/2018 through 12/31/2018).
Red lines show upper and lower control limits set at ฑ10% of the nominal concentrations for anions and ฑ35% of the
nominal concentrations for cations. Blue lines show upper and lower warning limits.
Instrument Low QC - CI A9
230 i
ฃ 220
txO
3 210 -
c
0
% 200
to
1 190 A
o 180
u
170
A A A. A A A A A
A AAA A
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19
Analysis Date
Instrument Low QC - CI A10
230 -i
ฃ 220
CU)
3 210 A
c
0
% 200 -
nj
1 190 -
o 180
u
170
a a*aa a a*a aa AaaA
A AAAAAAA i
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19
Analysis Date
Instrument Low QC - N03 A9
670 i
bo
= 620
c 570 -
520
1 1 1 1 1 1 1 1 1 1 1
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19 3/31/19
Analysis Date
Page 44 of 126
-------
Instrument Low QC - N03 A10
670
.E
oa
620
570
520
1 4*aA . i A*A
1 4 ***** *
'Z1
{ **iA*
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19 3/31/19
Analysis Date
Instrument Low QC - S04 A9
1320 -
1
E
CLD
1270 -
3
C
O
1220 -
03
*->
1170 -
a>
c
1120 -
u
1070
A A A A A 1 A a
A A
A A A A A A
A A A '
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19
Analysis Date
Instrument Low QC - S04 A10
1320 -
j
,E
eปo
1270
3
C
o
1220 -
+3
OJ
+->
c
1170 -
u
c
o
1120 -
u
1070
AA aAA4 A a . A* AA*AA
A A A A A t i
2/4/19 2/9/19 2/14/19 2/19/19 2/24/19 3/1/19 3/6/19 3/11/19 3/16/19 3/21/19 3/26/19
Analysis Date
i 30
3
c 25
o
| 20
g 15
u
ง 10
Instrument Low QC - NH4 C3
^ ^ ^ 4t ^ *
u 1/15/19 1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
4/5/19
Page 45 of 126
-------
4 30
3
c 25
o
a 20
g 15
Instrument Low QC - NH4 C6
^ ^ ^ I ^ * i *
5 10
u 1/15/19 1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
4/5/19
E
"SS 30
3
c 25 H
o
20
S 15
u
o 10
Instrument Low QC - Na C3
***** ^4 ***** *
*i
1/15/19 1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
4/5/19
30
25
TS 20
i-
ฆH
C 15
<1> "LJ
10
Instrument Low QC - Na C6
u*t * *#** ti**i i* *
-------
Instrument Low QC - K C6
iiw1 a* u4 ^ ii*1* *i 4
4. 30
tlQ
S*
0
X 20
(TJ
1 15
oj
u
n 10
u 1/15/19 1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
4.1.B.2.3 Duplicate Samples
Duplicate analysis results are obtained from two different aliquots of the same filter sample
extract run on the same instrument sequentially in the analytical batch. Each analytical batch
includes three sets of duplicate samples. The relative percent difference (RPD) for duplicate
samples must be within ฑ 10% when sample concentrations are greater than ten times the
analytical MDL and within ฑ 100% when sample concentrations are at or up to ten times the
analytical MDL. There was a total of 256 duplicate samples analyzed for anions (Figure 4.1.B-
4), with two cases where the RPD did not meet the QC criteria for chloride and one case for
sulfate; all RPD results met the QC criteria for nitrate. There was a total of 251 duplicate samples
analyzed for cations (Figure 4.1.B-4), with two cases where the RPD did not meet the QC
criteria for potassium and three cases for sodium; all RPD results met the QC criteria for
ammonium. In all cases where duplicate precision fails to meet the QC criteria, five samples
(duplicate plus four randomly selected samples) from the analysis set are reanalyzed. If any of
the reanalyzed samples fail to meet the QC criteria, the entire batch is reanalyzed.
Page 47 of 126
-------
Figure 4.1.B-4: Ion duplicate analysis results for the analysis period 1/25/2019 through 3/26/2019 (samples
collected 10/1/2018 through 12/31/2018).
Chloride Duplicate Precision
Nitrate Duplicate Precision
1600
1
ฃ
1400
~5a
3
1200
C
o
1000
c
800
-------
4.1. B. 2.4 Spiked Sample Extracts
Matrix spikes are performed on 4% (two per batch of 50) of the samples analyzed. The matrix is
deionized (DI) water, and spike samples typically meet the QC criteria with failures most likely
from introduced contamination. All spike recoveries met the QC criteria except for one chloride
case (Figure 4.1.B-5); the sample and four other samples from the same analysis set were
reanalyzed.
Figure 4.1.B-5: Time series of recovery percentage for anion and cation of matrix spikes for the analysis period
1/25/2019 through 3/26/2019 (samples collected 10/1/2018 through 12/31/2018).
Chloride Matrix Spike Recovery
115
Nitrate Matrix Spike Recovery
115
no
no
ฃ 105
>
o
o
v
^ 100
c
-------
4.1.B.2.5 Reagent Blanks and Spikes
Analyses began with the analysis of two DI water instrument blanks which clean the sample loop
prior to injection of calibration standards. Method blanks and laboratory control spikes (LCS) are
used to measure the background contamination that could be introduced during the extraction,
sample handling, or analysis processes. At the time of filter extraction, an empty extraction vial
is included as a method blank at a rate of 1 for every 50 samples. Empty extraction vials are also
spiked with exact volumes of concentrated solutions for both anions and cations a rate of 1 for
every 25 samples for LCS analysis. The same volume of water (20.0 mL) is added to the method
blank and LCS vials as is added to the vials with the filter samples to be extracted.
Figure 4.1.B-6: Concentrations of anions and cations in DI water blanks and method blanks for the analysis period
1/25/2019 through 3/26/2019 (samples collected 10/1/2018 through 12/31/2018). Black line indicates the analytical
method detection limit.
Chloride Measured in Instrument DI Blanks and Method Blanks
11
o
u
15
10
5
0
MDL = 5 ppb
mSm
M*
1 3 5 7 9 11131517192123252729313335373941434547495153555759616365676971737577
Sequential Blanks Analyzed 01/25/2019 - 03/26/2019
Chloride Instalment DI Blank Chloride Method Blanks
10
Nitrate Measured in Instrument DI Blanks and Method Blanks
MDL = 8 ppb
I Is
ra =
ฆ*->
c
a>
o
c
o
u
13 5 7 9 11131517192123252729313335373941434547495153555759616365676971737577
Sequential Blanks Analyzed 01/25/2019 - 03/26/2019
Nitrate Instrument DI Blank Nitrate Method Blanks
12
s e 7
ฆB "5S
o
V
-3
Sulfate Measured in Instrument DI Blanks and Method Blanks
MDL = 11 ppb
10 20 30 40 50 60 70
Sequential Blanks Analyzed 01/25/2019 -03/26/2019
Sulfate Instrument DI Blank Sulfate Method Blanks
80
90
Page 50 of 126
-------
Sodium Measured in Instrument Dl Blanks and Method Blanks
60
E -i 40
I 1
ro =s 20
o
u
ZJ
CO
03
(D
ฃ
I i
CD 3
O
U
MDL = 3 ppb
Vw* I .
o 10
20 30 40 50 60 70
Sequential Blanks Analyzed 01/25/2019 - 03/26/2019
80
90
Sodium Instrument Dl Blank Sodium Method Blank
Ammonium Measured in Instrument Dl blanks and Method Blanks
MDL = 3 ppb
WN - ซ. ป ซV
10 20 30 40 50 60 70
Sequential Blanks Analyzed 01/25/2019-03/26/2019
Ammonium Instrument Dl blank Ammonium Method Blank
80
90
10
ra
o>
E
I i
ra 3
o
u
Potassium Measured in Instrument Dl blanks and Method Blanks
MDL = 3 ppb
* _ ซM
_ *
10 20 30 40 50 60
Sequential Blanks Analyzed 01/25/2019 - 03/26/2019
Potassium Instrument Dl Blank Potassium Method Blank
70
80
90
The laboratory does not use the reagent blanks (instrument Dl blanks and method blanks) or the
LCS analyses for control purposes. Because the concentrations in the LCS (Table 4.1.B-4 and
Table 4.1.B-5) are very close to the CCV check standards, it is useful to compare the LCS results
with the CCV check standard criteria for evidence of outlier frequency. The LCS analyses
(Figure 4.1.B-7) have more frequent outliers relative to the CCV check standards (Figure 4.1.B-1
and Figure 4.1.B-2), suggesting that background contamination is introduced during the sample
handling and processing of samples and not typically due to instrumental issues. The method
blanks and LCS analysis results are useful for internal laboratory quality control, as they can
alert the analyst to background issues early during the analysis process. Review of the LCS and
method blank results relative to the CCV check standards is performed for each analysis set.
Table 4.1.B-4: Target concentrations for anion LCS for the analysis period 01/25/2019 through 03/26/2019
(samples collected 10/1/2018 through 12/31/2018).
QC Sample
CI (ppb)
NOs (ppb)
SO/ (ppb)
LCS Low
196
392
1180
LCS Med
476
1430
2860
LCS High
2000
6000
12000
Page 51 of 126
-------
Table 4.1.B-5 Target concentrations for cation LCS for the analysis period 01/25/2019 through 03/26/2019 (samples
collected 10/1/2018 through 12/31/2018).
QC Sample
Na+ (ppb)
NH4+ (ppb)
K+ (ppb)
LCS Low
20
20
20
LCS Med
276
276
276
LCS High
769
769
769
Figure 4.1.B-7: Control charts for anion and cation LCS analyses relative to the CCV check standard QC criteria for
the analysis period 1/25/2019 through 3/26/2019 (samples collected 10/1/2018 through 12/31/2018). Red lines show
upper and lower control limits per the CCV check standard QC criteria Blue lines show upper and lower warning
limits.
230 -|
ฃ 220
210
200
S 190
o 180 -
u
170
1/15/19
Laboratory Control Spike Low QC - CI
a*aaAaaaA
AAฑ aaAaaa AAaA ^ A A * aA a*aaA *a
1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19 4/5/19
1330 -|
E 1280 -
w
3 1230 -
c
0
X 1180 -
-------
J 540 -
"SB 520 -
| 500
ฆ2 480 -
| 460 -
| 440 -
o 420
u
400 4
Laboratory Control Spike MED QC - CI
1
A
*MA1 'Z'1' aAA*aA / A
1/15/19 1/25/19 2/4/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
Laboratory Control Spike Med QC - S04
I
3125 -
00 3025 -
3
C
O
2925 -
2825 -
0)
2725 -
c
o
u
2625 -
2525
1/15/19
A*a4aA 1'il1 a / 1
1/25/19 2/4/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
i65o Laboratory Control Spike Med QC - N03
E 1600
m 1550 -
% 1500
.2 1450 -
n 1400
c 1350 -
S 1300 -
o 1250
u 1200
ft" A ^ A ฑ
1/15/19 1/25/19 2/4/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
Laboratory Control Spike High QC - CI
2190
ฃ 2140
"if 2090 -
= 2040 -
1 1990 -
c 1940
S 1890 -
3 1840
1790
1/15/19
1 a 1"AAa 4A a a
* ii1 ป1* A A A AiA 4 *
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Page 53 of 126
-------
6590
6390
6190
5990
5790
5590
5390
1/
30 i
25 -
20 -
15 -
10 --
1/15,
30 n
25 -
20 -
15 -
10
Laboratory Control Spike High QC - S04
* i"1 i 1 * * 1
A
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Laboratory Control Spike High QC - N03
A *AA *AaA A A A
1 ii*
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Laboratory Control Spike Low QC - NH4
A
A
a A i* A v
aa^aa4 1 4
1 4 A aa
AA A a A A
A
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Laboratory Control Spike Low QC - Na
A A A ^ ^
A * 'i4 iA A A*AA A 4 '4 Ai*AA 'i
1 1 1 1 1 1 1
1/25/19 2/4/19 2/14/19 2/24/19 3/6/19 3/16/19 3/26/19
Analysis Date
Page 54 of 126
-------
30 n
ฃ
"oB 25 -
o
ซ 20 A
u 15 -
c J-J n
o
u
10
1/15/19
Laboratory Control Spike Low QC - K
At i
A* *iAt *
Ia * A A 4
1/25/19 2/4/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
Laboratory Control Spike Medium QC - NH4
310 -|
I
F
300
3
290
C
o
280
+->
270
c
c 240 -
o
u
220
1/15/19
viv4 v a;a^a aa*/ iaa
A A *aAa
1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19
Page 55 of 126
-------
Laboratory Control Spike High QC - NH4
880
.E
^ 830 -
c
0
1 780
c
a>
y 730
680
1/15/19
A^A aaaaa a* *A *aAA ^ aaaa a a A ^ aa* 4a
A
1/25/19 2/4/19
2/14/19 2/24/19
Analysis Date
3/6/19 3/16/19 3/26/19
Laboratory Control Spike High QC - Na
, 830 -
4= 780
S 730 -
680
1/15/19
iAA \4aa aaa Aa *aAa a aAaa aa4
A a A
iA *A
1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19
Laboratory Control Spike High QC - K
880 -I
I
,ฃ
^ 830 -
c
o
'S 780
c
a
y 730 -
680
1/15/19
ll1 i* V'i4 *aA* ill 1 1 * A*
A A A
1/25/19 2/4/19
2/14/19 2/24/19 3/6/19
Analysis Date
3/16/19 3/26/19
4.1.B.2.6 Round robin (USGS)
The RTI Ions Chromatography Laboratory participated in the National Atmospheric Deposition
Program/Mercury Deposition Network Interlaboratory Comparison Program. The program is
administered by the United States Geological Survey (USGS) Branch of Quality Systems. Four
samples per month were sent to participating laboratories for analysis. A website reporting
participant results is currently in development; a report for the 2018 results is available upon
request.
4.1.B.2.7 Reanalysis
Five percent of all samples are reanalyzed using different instruments and different calibration
curves. Samples are flagged for reanalysis during analyst review of analytical results, and
reasons include poorly integrated peaks and cases where one peak is significantly higher than the
other peaks in the chromatograph (particularly for cations peaks, which elute very close
together). In these cases, the sample may be diluted for reanalysis. Samples are also flagged if
Page 56 of 126
-------
the QC criteria for reanalysis samples are not met. When more than one analysis within an
analysis set fails to meet the QC criteria as outlined in Table 4.1.B-1, the whole set of samples is
reanalyzed. The majority of reanalyzed samples are from QC criteria failure for background
contamination from sodium, chloride, and/or potassium detected in either the original or
reanalysis result. In cases where the entire set of samples were reanalyzed, background
contamination did not propagate through the whole set.
During this reporting period there were cases of sodium, potassium, chloride, and sulfate
contamination from the pre-cleaned filter caps used in the analysis vials. The problem was
identified quickly and analysis data for impacted samples were not reported; these samples were
reanalyzed using caps that were verified as clean. The faulty caps were from a new manufacturer
batch, and the laboratory had a supply of clean filter caps from a previous batch that were used
until a suitable replacement was found. The laboratory began purchasing and cleaning caps
without a filter and found no background issues with these caps.
During this reporting period, there were 731 samples reanalyzed for anions and 752 samples
reanalyzed for cations (Figure 4.1.B-8). At most, about 1.5% (10-11 samples) of the samples
required edits (failed criteria for precision between the original and reanalysis result) for both
sodium and chloride. In these cases, the reanalysis result was reported only for the ion with the
poor precision. The failures were likely caused by contamination introduced during the analyses.
The percentage was slightly higher for potassium, however most of the edits required were for
samples flagged for reanalysis to check baseline level samples.
Page 57 of 126
-------
Figure 4.1.B-8: Ion reanalysis results for the analysis period 1/25/2019 through 3/26/2019 (samples collected
10/1/2018 through 12/31/2018).
Chloride Reanalysis Precision
8000
7000
6000
5000
4000
3000
2000
1000
0
y = 1.0218x-1.5529
R2 = 0.9993
0 2000 4000 6000 8000
Conentration measured ppb
(first analysis)
Sulfate Reanalysis Precision
y = 1.0093x-2.2134
R2 = 0.9987
qj ro
E ฃ
0 500 1000 1500 2000
Concentration measured ppb
(first analysis)
Nitrate Reanalysis Precision
| v = 0.9971x-1.1601 0
R2 = 0.9995
10000
Concentration measured ppb
(first analysis)
Ammonium Reanalysis Precision
5000
3 ฃ 3000
O)
E ฃ
c -o
2000
1000
y = 0.9991x + 0.8858
fp = 0.9886
1
Z
2000
4000
6000
Concentration measured ppb
(first analysis)
Sodium Reanalysis Precision
oj 2
ฃ ฃ
1400
1200
1000
800
600
400
200
y = 0.9795x +0.4708
~
R2 = 0.9917
>
Concentration measured ppb
(first analysis)
oj 2
ฃ ฃ
450
400
350
300
250
200
150
100
50
0
Potassium Reanalysis Precision
y = 0J489x+.0.5D9
R2 = 0.9227
VL
0 100 200 300 400
Concentration measured ppb
(first analysis)
4.1.B.3 Determination of Uncertainties and Method Detection Limits
For discussion of Method Detection Limits (MDLs) see Section 3.1.3.2.
For discussion of analytical uncertainty and total uncertainty see Section 3.1.2 and Section 6.5,
respectively.
Page 58 of 126
-------
4.1.B.4 Audits, Performance Evaluations, Training, and Accreditations
4. I.B.4.1 System Audits
The prime contractor (UC Davis) did not conduct any audit of the RTI Ion Chromatography
Laboratory during this reporting period.
4. LB.4.2 Performance Evaluations
RTI performance was satisfactory in the Interlaboratory OAQPS 2018 Mega PE Speciation
Event.
4.1.4.3 Training
All new laboratory staff receive training for performing tasks described in the SOPs relevant to
their assigned work.
4.1.4.4 A ccreditations
There are no accreditations for analysis of ions on aerosol filters by Ion Chromatography.
4.1.B.5 Summary of Filter Field Blanks
Over the sampling period (October 1, 2018 through December 31, 2018) there were 420 valid
nylon filter field blanks. Table 4. l-6a and Table 4. l-6b summarize the field blank statistics.
Table 4.1-6a: Nylon filter field blank statistics in ng/mL for the analysis period 1/25/2019 through 3/26/2019
(samples collected 10/1/2018 through 12/31/2018).
Ions
Count
Median
(Hซ/mL)
Average
(jig/mL)
Min
(jig/mL)
Max
(Hg/mL)
St. Dev.
(Hซ/mL)
CI
420
0.003
0.004
0.000
0.332
0.016
no3
420
0.004
0.017
0.000
3.661
0.179
scv-
420
0.001
0.006
0.000
0.953
0.047
Na+
420
0.002
0.003
0.000
0.041
0.003
nh4+
420
0.003
0.013
0.000
2.335
0.114
K+
420
0.001
0.004
0.000
0.369
0.019
Table 4.1-6b: Nylon filter field blank statistics in jxg/filter (extraction volume 15 mL) for the analysis period
1/25/2019 through 3/26/2019 (samples collected 10/1/2018 through 12/31/2018).
Ions
Count
Median
(fig/filter)
Average
(fig/filter)
Min
(fig/filter)
Max
(fig/filter)
St. Dev.
(fig/filter)
CI
420
0.080
0.107
0.000
8.308
0.403
no3
420
0.092
0.428
0.000
91.517
4.469
SC>42-
420
0.022
0.157
0.000
23.837
1.184
Na+
420
0.058
0.075
0.000
1.024
0.084
nh4+
420
0.067
0.315
0.000
58.385
2.857
K+
420
0.017
0.103
0.000
9.217
0.470
Page 59 of 126
-------
4.2 UC Davis X-Ray Fluorescence Laboratory
The UC Davis X-Ray Fluorescence Laboratory received and analyzed PTFE filters from batches
39 through 50, which includes samples collected January 1, 2018 through December 31, 2018.
UC Davis performed analysis for 33 elements using energy dispersive X-ray fluorescence
(EDXRF) instruments. These analyses were performed during an analysis period from March 23,
2018 through April, 26, 2019. Three EDXRF instruments, XRF-1, XRF-4, and XRF-5 performed
all of the analyses during this period; see Table 4.2-1 for details.
Table 4.2-1: Sampling dates and corresponding EDXRF analysis dates during this reporting period. Analysis dates
include reanalysis - as requested during QA Level 1 validation - of any samples within the sampling year and
month.
Sampling
Year
Sampling
Month
XRF-1 Analysis Dates
XRF-4 Analysis Dates
XRF-5 Analysis Dates
2018
January
3/23/2018-4/17/2018
3/23/2018-4/16/2018
N/A
2018
February
4/16/2018-5/12/2018
4/16/2018-5/12/2018
N/A
2018
March
5/12/2018-6/14/2018
5/13/2018-6/12/2018
N/A
2018
April
6/13/2018-7/27/2018
6/19/2018-7/23/2018
N/A
2018
May
7/19/2018-9/12/2018
7/23/2018-8/30/2018
N/A
2018
June
8/20/2018 - 10/4/2018
8/17/2018-9/14/2018
N/A
2018
July
9/15/2018- 10/18/2018
9/14/2018- 10/19/2018
N/A
2018
August
10/18/2018- 11/15/2018
10/19/2018- 11/15/2018
N/A
2018
September
11/15/2018- 12/18/2018
11/15/2018- 12/17/2018
N/A
2018
October
12/29/2018 - 1/20/2019
12/29/2018 - 1/20/2019
12/26/2018 - 1/24/2019
2018
November
1/20/2019-2/6/2019
1/20/2019-2/19/2019
1/24/2019-2/19/2019
2018
December
3/27/2019-4/2/2019
2/19/2019-3/31/2019
2/19/2019-4/26/2019
2018
All Months
3/23/2018-4/2/2019
3/23/2018-3/31/2019
12/26/2018-4/26/2019
4.2.1 Summary of QC Checks and Statistics
Samples are received by the UC Davis XRF Laboratory following the chain-of-custody
procedures detailed in the UCD CSN TI302B. Samples are analyzed using Malvern-Panalytical
Epsilon 5 EDXRF instruments following UCD CSN SOP #302. Calibration of the EDXRF
instruments is performed annually and as needed to address maintenance or performance issues
(e.g. an X-ray tube or detector is replaced). Quality control procedures are described in UCD
CSN TI 302D and are summarized in Table 4.2-2.
Page 60 of 126
-------
Table 4.2-2: UC Davis quality control measures for element analysis by EDXRF.
Analysis
l'lV(|IK'IK\\
( rik'rinn
( ปnvc(i\c Action
Detector
Calibration
Weekly
None (An automated process done
by XRF software)
XRF software automatically adjusts
the energy channels
PTFE Blank
Daily
< acceptance limits with
exceedance of a single element
allowed for a maximum of two
consecutive days
Change/clean blank if
contaminated/damaged
Clean the diaphragm, if necessary
Further cross-instrumental testing
UC Davis Multi-
element sample
Daily
ฑ10% of reference mass loadings
for Al, Si, S, K, Ca, Ti, Cr, Mn,
Fe, Ni, Cu, Zn and Pb
Check sample for
damage/contamination
Further cross-instrumental testing
Replace sample if necessary
Micromatter
Al&Si sample
Weekly
ฑ10% of reference mass loadings
UC Davis Multi-
element sample
Weekly
ฑ10% of reference mass loadings
for Al, Si, S, K, Ca, Ti, Cr, Mn,
Fe, Ni, Cu, Zn and Pb
Reanalysis
samples
Monthly
z-score between ฑ1 for Al, Si, S,
K, Ca, Ti, Mn, Fe, Zn, Se and Sr
SRM 2783
Monthly
Bias between ฑ1 for Al, Si, S, K,
Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn and
Pb
Daily QC checks include a laboratory blank (PTFE blank) and a multi-elemental reference
material (ME-RM) to monitor contamination and stability/performance of the instruments. A
Micromatter Al&Si ME-RM and a UC Davis-made ME-RM are also analyzed weekly to check
the instrument performance. Inter-instrumental comparability is monitored by analyzing the bias
and precision between instruments of the weekly UC Davis ME-RM. Long-term inter-
instrumental comparability is monitored using a set of reanalysis filters which are reanalyzed
monthly on each instrument. Long-term reproducibility is monitored using the reanalysis filters
and by analyzing a NIST SRM 2783 standard monthly and comparing the EDXRF error from the
certified/reference mass loadings to acceptance limits.
4.2.2 Summary of QC Results
QC tests conducted over the course of the analysis period showed good overall control of the
instruments and process. There were sporadic failures of the QC criteria, which were
investigated promptly and corrected with minimal impact on sample analysis. The following
summarizes the QC issues which occurred during the analysis period reported here.
Random occasional zinc contamination was observed on QC filters for XRF-1, XRF-4, and
XRF-5. This sporadic zinc contamination appears to be related to the design of the instrument
and is unavoidable. Samples analyzed during this period were monitored closely for any
contamination and were reanalyzed if there was any question of contamination. The reported
data are not impacted. See Section 2.3.2, Section 3.2.1.1, and Section 4.2.2.1 for further detail.
XRF-1, XRF-4, and XRF-5 also exhibited some failures of the acceptance criteria for all QC
checks of Ca. Investigation is ongoing, but initial findings suggest that gradual increase in Ca
concentrations on QC filters might be caused by environmental deposition during extended
residence in the instruments. Samples are only exposed to the environment for a day or two
during routine analysis, thus are not susceptible to gradual Ca contamination. However, samples
are carefully monitored for atypical and abrupt calcium contamination events and reanalyzed as
Page 61 of 126
-------
necessary. The reported data are not impacted. See Section 2.3.3, Section 3.2.1.2, and Section
4.2.2.1 for further detail.
In addition to the QC results above, the conditions under which the elements are analyzed by
EDXRF (e.g. secondary targets and integration times) were changed during the annual
calibration of all EDXRF systems in December 2018. These changes were made to help reduce
variability, detection limits, and bias in some elements. See Section 2.3.1 and Section 4.2.2.5 for
further details.
4.2.2.1 Results of Daily QC Checks
Possible contamination and instability issues are monitored by analyzing a PTFE blank daily.
The EDXRF results are compared to acceptance limits, which are calculated as three times the
standard deviation plus the mean of a set of laboratory PTFE blanks. Figure 4.2-la and Figure
4.2-lb show the results of daily analyses of laboratory blanks for each instrument. The
acceptance limits shift in late December 2018 because of changes made to the XRF applications,
(see Section 4.2.2.5 for details). The application changes included modification of the
deconvolution parameters effecting the background in the spectra that necessitated the changes in
the blank acceptance limits. If the mass loading exceeds the limit for more than two consecutive
days, the blank is replaced to distinguish between blank contamination and instrument
contamination. Some occasional exceedance of the acceptance limits is expected but not
continuous or repeated exceedances. In all cases of exceedance, the other QC filters are checked
to determine if the problem is instrumental or strictly contamination of a blank. Sample analysis
results are reviewed during QA Level 1 validation (UCD CSN TI801C), and elements associated
with occasional contamination (Zn and Ca; see Section 2.3.2, Section 2.3.3, Section 3.2.1.1, and
Section 3.2.1.2) are monitored closely. When contamination is suspected, filters are reanalyzed
and the reanalysis result is reported if contamination was present in the original analysis. A total
of nine samples from 2018 were reanalyzed for suspected Zn contamination (three from XRF-1,
six from XRF-4, and none from XRF-5). Of those, one was found to have Zn contamination and
the reanalysis result was reported. For the rest the original valid result was reported. One sample
was reanalyzed for suspected Ca contamination (from XRF-5). The sample was found to have no
Ca contamination and the original valid result was reported.
Both XRF-1 and XRF-4 had sporadic elevated measurements of Zn on laboratory blanks
throughout the analysis period (as discussed in Section 2.3.2 and Section 3.2.1.1). These elevated
levels were not measured over consecutive days thus did not fail the acceptance criteria;
however, these occurrences are monitored closely. Zn contamination likely comes from wear on
the sample changer; Zn is a common contaminant in elemental analysis systems.
XRF-1, XRF-4, and XRF-5 all show gradual increases in Ca (as discussed in Section 2.2.3 and
Section 3.2.1.2), which is reduced immediately after the blank filter is changed. This indicates
contamination of the blank filter likely from atmospheric deposition and/or instrument wear. The
cause of Ca increases on QC filters with long, multi-day residences in the instrument is being
investigated.
Lastly, CI had few exceedances on XRF-1 and XRF-4 instruments during the analysis period. For
the larger exceedances, laboratory blanks were replaced which corrected the exceedance; for
others the signal decreased without correction. The cause of the CI exceedances is unknown; as a
volatile element, CI has a highly variable signal from QC filters. These exceedances are caused
by variability in the CI measurement, not contamination in the EDXRF instruments.
Page 62 of 126
-------
Figure 4.2-la: Results of daily analyzed PTFE laboratory blanks for the analysis period 3/23/2018 through
4/26/2019 (samples collected 1/1/2018 through 12/31/2018). Elements Na through Zn shown. Acceptance limits
were recalculated for the December 2018 calibrations resulting in the shifts seen in the plots, see text for details.
XRF-1
XRF-4
XRF-5
Accpeptance Limit
Na
Si
CI
Ti
Mn
Ni
0.080
0.060
0.040'
0.020
o.ooo-
0.004
0.003
0.002
0.001'
o.ooo-
0.020-
0.015-
0.010-
0.005-
0.002-
0.001
0.000
0.050-
0.040-
0.030-
0.020
0.010-
0.020-
0.015
0.010-
0.005
Mg
Fe
Cu
00
CO
CO
00
cr>
O)
cn
00
00
CO
00
O)
o
CM
o
CM
O
CM
o
CM
o
CNJ
O
CM
o
CM
O
CM
o
CM
O
CM
o
CM
o
CM
o
CN
>>
CD
"5
Q.
QJ
CO
>
o
Z
c
CD
D
CD
CD
>
CD
2
"5
3
Q.
CU
CO
>
o
ฃ=
CD
-ฆ>
ID
5
XRF Analysis
Date
0.100-
0.080
0.060
0.040
0.020
Al
0.010
0.005 H
0.000
0.040 H
0.030
0.020 H
0.010
0.000
0.009
0.007
0.005-
0.003-
0.005
0.004-|
0.003
0.002
0.001
0.000
0.040-
0.030
0.020-
0.010
0.000
o
CM
>>
CD
I ,l
Ca
Cr
Co
Zn
,
1 -aJ
00
00
CO
00
CT)
O O O O O
CM CM CM CM CM
>s "5 Q. > C
CD
-------
Figure 4.2-lb: Results of daily analyzed PTFE laboratory blanks for the analysis period 3/23/2018 through 4/26/2019
(samples collected 1/1/2018 through 12/31/2018). Elements As through Pb shown. Acceptance limits were recalculated
for the December 2018 calibrations resulting in the shifts seen in the plots, see text for details.
XRF-1 XRF-4
XRF-5
Accpeptance Limit
As
Rb
Ag
Sn
Ba
0.010
0.008-
0.005-
0.002
0.000
0.012-
0.008
0.004-
0.040-
0.030-
0.020
0.010-
0.075-
0.050
0.025-
Se
Sr
Cd
Sb
Br
0.010-
0.008-
0.005-
0.002-
o.ooo-
0.080-
0.060-
0.040-
0.020
o.ooo-
0.060-
0.040-
0.020 i
0.200-
0.150
0.100-
0.050
0.010-
Zr
Cs
Pb
00
CO
CO
00
CD
CD
CD
00
00
00
00
CD
CD
CD
00
CO
00
00
CD
CD
CD
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
CM
OJ
OJ
CM
OJ
OJ
CM
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
OJ
>%
Q.
>
ฃZ
CD
>.
Q.
>
ฃZ
CO
>N
>*
CL
>
c
CO
CO
^
CD
o
CO
CD
CD
CD
o
CO
CO
CO
J
cd
o
CO
CO
C0
5
03
z
)
2
C/)
XRF Analysis Date
Page 64 of 126
-------
Daily operational performance of the instruments is monitored by a multi-element reference
material (ME-RM). Each instrument had its own daily ME-RM produced by UC Davis. The
acceptance limits are set to +/- 10% RSD of the reference values for the relevant elements, as
listed in Table 4.2-2. When more than two consecutive measurements exceed these limits, the
results are marked unacceptable. Corrective actions for unacceptable QC results include
checking the sample for damage or contamination, checking the results for the affected element
on other QC samples, cross-instrumental testing if necessary to determine if the unacceptable
result is due to the instrument or the QC sample, and further investigations as necessary. Sample
analysis is halted or samples analyzed after the unacceptable QC result are noted for possible
reanalysis depending on the outcome of the investigation. When a problem with the instrument is
found the affected samples are reanalyzed on a different instrument or the same instrument after
the issue is corrected and once it has been demonstrated to be within control again. QC samples
which have been found to be damaged or contaminated will be replaced (UCD CSN TI302D).
Tables 4.2-3, 4.2-4, and 4.2-5 show the results of the UC Davis ME-RMs. The UC Davis ME-
RM QC samples were replaced in December 2018 (at time of calibrations). The new QC samples
have different mass loadings relative to those previously used. Si, Ti, Cu, and Pb have lower
loadings which may result in a higher number of exceedances relative to previously reported QC
results. A small number of criteria exceedances are expected statistically, but not more than a
few percent of the total number of measurements. Investigations of other QC filters and
laboratory blanks following these exceedances did not show any contamination or instrumental
issues, so no corrective actions were taken. Unacceptable QC results for Ca and Zn are expected
to be from the same source as discussed for laboratory blank contamination (see Section 2.3.2,
Section 2.3.3, Section 3.2.1.1, and Section 3.2.1.2). The laboratory blanks were replaced when
contamination occurred; however, the ME-RM samples were not replaced in response to
contamination.
Table 4.2-3: Descriptive statistics of XRF-1 results (|ig/cm2) of the daily UC Davis ME-RM for the analysis period
3/23/2018 through 4/26/2019 (see Table 4.2-1 for corresponding sampling dates), N = 540.
Element
A1
Average
1.918
Lower Limit
1.724
Upper Limit
:.107
% Execcdanee
0.0
% Unacceptable
0
RSD %
1.7
Si
1.786
1.605
1.961
0.0
0
1.5
S
13.726
12.426
15.188
0.0
0
0.7
K
2.008
1.813
2.216
0.0
0
0.7
Ca
2.129
1.884
2.303
0.0
0
2.0
Ti
0.094
0.082
0.100
4.3
0
4.5
Cr
0.816
0.735
0.899
0.0
0
0.8
Mn
0.403
0.362
0.442
0.0
0
1.7
Fe
2.283
2.039
2.492
0.0
0
0.8
Ni
0.141
0.128
0.156
0.0
0
2.1
Cu
0.428
0.383
0.469
0.0
0
1.3
Zn
0.361
0.319
0.390
0.2
0
1.6
Pb
0.364
0.327
0.400
3.5
0
4.8
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
Page 65 of 126
-------
Table 4.2-4: Descriptive statistics of XRF-4 results (|ig/cm2) of the daily UC Davis ME-RM for the analysis period
3/23/2018 through 4/26/2019 (see Table 4.2-1 for corresponding sampling dates), N = 667.
Element
A1
Average
1.656
Lower Limit
1.488
Upper Limit
1.818
% Exeeedanee
0
% Unacceptable
0
USD %
2.3
Si
1.923
1.741
2.128
0.3
0
2.3
S
13.536
12.239
14.959
0
0
0.6
K
1.987
1.797
2.196
0
0
0.5
Ca
2.171
1.955
2.390
10.0
5.8
2.9
Ti
0.098
0.089
0.108
1.6
0
4.4
Cr
0.822
0.744
0.910
0
0
0.7
Mn
0.401
0.363
0.443
0
0
1.8
Fe
2.263
2.034
2.486
0
0
0.9
Ni
0.143
0.129
0.157
0
0
2.0
Cu
0.426
0.386
0.472
0
0
1.3
Zn
0.356
0.313
0.383
2.4
0.4
3.4
Pb
0.358
0.327
0.400
5.5
0
5.6
Limits are +/-10% of the reference loading (see UCD CSN TI302D).
Table 4.2-5: Descriptive statistics of XRF-5 results (|ig/cm2) of the daily UC Davis ME-RM for the analysis period
12/25/2018 through 4/26/2019 (see Table 4.2-1 for corresponding sampling dates), N = 143.
Elemenl A\eraiie l.oซer l imit I pper l imit '!.ฆ Hxceedance "<ฆ I nacceplable USD "<ฆ
A1
2.233
1.988
2.430
0
0
1.4
Si
0.750
0.644
0.787
0
0
2.2
S
16.603
14.852
18.153
0
0
0.5
K
2.370
2.132
2.606
0
0
0.3
Ca
2.267
2.022
2.471
0
0
0.8
Ti
0.053
0.0463
0.566
9.8
2.1
4.9
Cr
0.954
0.855
1.045
0
0
0.6
Mn
0.469
0.418
0.511
0
0
1.5
Fe
2.602
2.330
2.848
0
0
0.6
Ni
0.166
0.150
0.183
0
0
1.5
Cu
0.337
0.302
0.370
0
0
1.2
Zn
0.357
0.319
0.390
0
0
1.3
Pb
0.076
0.068
0.083
17.5
2.1
7.8
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
4.2.2.2 Results of Weekly QC Checks
Weekly QC checks include analysis of a UC Davis produced ME-RM (different than the daily
ME-RM) and a ME-RM purchased from Micromatter containing only A1 and Si. The UC Davis
weekly ME-RM was replaced in December 2018. Weekly results are compared to acceptance
limits of +/- 10% of the reference values for the relevant elements, as listed in Table 4.2-2. When
more than two consecutive measurements exceed these limits, the results are marked
unacceptable. Corrective actions for unacceptable results are described in section 4.2.2.1 and can
Page 66 of 126
-------
be found in the UCD XRF SOP and Technical Instructions (UCD CSN TI302D). A weekly QC
report is generated internally, which includes checks of the laboratory blanks and the daily and
weekly ME-RMs.
Table 4.2-6, Table 4.2.7, and Table 4.2-8 show the EDXRF statistics of the weekly UC Davis
ME-RM run until March 2019.
Table 4.2-6: Descriptive statistics of XRF-1 results (|ig/cm2) of the weekly UC Davis ME-RM for the analysis
period 3/28/2018 through 1/25/2019 (see Table 4.2-1 for corresponding sampling dates), N = 41.
Element
U
Average
1.574
Lower Limit
1.403
Upper Limit
1.715
% Excecdanee
0
% Unacceptable
0
USD %
3.4
Si
2.597
2.313
2.828
0
0
1.8
S
9.692
8.632
10.551
0
0
0.8
K
1.483
1.316
1.608
0
0
0.7
Ca
1.618
1.426
1.743
0
0
1.8
Ti
0.121
0.106
0.130
0
0
2.8
Cr
0.604
0.537
0.658
0
0
0.8
Mn
0.291
0.258
0.315
0
0
2.1
Fc
1.667
1.479
1.807
0
0
1.1
Ni
0.100
0.089
0.109
0
0
2.7
Cu
0.287
0.257
0.314
0
0
1.3
Zn
0.233
0.203
0.249
4.9
0
3.7
Pb
0.537
0.486
0.594
0
0
2.2
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
Table 4.2-7: Descriptive statistics of XRF-4 results (|ig/cm2) of the weekly UC Davis ME-RM for the analysis
period 3/23/2018 through 3/5/2019 (see Table 4.2-1 for corresponding sampling dates), N = 47.
Element A\eraiie Lower l.imil I pper l.imil "ฆป Exceedance "ฆป I nacceplable USD";.
A1
1.388
1.242
1.518
0
0
3.7
Si
2.629
2.358
2.881
0
0
2.4
S
9.640
8.664
10.589
0
0
1.2
K
1.460
1.312
1.603
0
0
0.7
Ca
1.600
1.436
1.755
0
0
2.2
Ti
0.123
0.108
0.132
0
0
3.5
Cr
0.603
0.541
0.661
0
0
0.8
Mn
0.287
0.259
0.317
0
0
2.4
Fc
1.626
1.459
1.783
0
0
1.0
Ni
0.100
0.089
0.109
0
0
2.6
Cu
0.286
0.258
0.315
0
0
1.7
Zn
0.23 1
0.203
0.249
2.1
0
3.9
Pb
0.542
0.484
0.592
0
0
3.0
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
Page 67 of 126
-------
Table 4.2-8: Descriptive statistics of XRF-5 results (|ig/cm2) of the weekly UC Davis ME-RM for the analysis
period 1/3/2019 through 2/28/2019 (see Table 4.2-1 for corresponding sampling dates), N = 9.
Element
A1
Average
1.465
Lower Limit
1.319
Upper Limit
1.612
% Exeeedanee
0
% Unacceptable
u
RSD %
1.3
Si
2.675
2.410
2.946
0
0
1.1
S
9.598
8.648
10.570
0
0
0.6
K
1.475
1.329
1.625
0
0
0.3
Ca
1.616
1.455
1.779
0
0
0.9
Ti
0.123
0.112
0.136
0
0
2.8
Cr
0.605
0.544
0.665
0
0
0.6
Mn
0.291
0.263
0.321
0
0
1.4
Fe
1.645
1.481
1.810
0
0
0.5
Ni
0.100
0.090
0.110
0
0
2.4
Cu
0.290
0.261
0.319
0
0
1.1
Zn
0.241
0.217
0.265
0
0
1.2
Pb
0.549
0.496
0.606
0
0
1.7
Limits are +/-10% of the reference loading (see UCD CSN TI302D).
Table 4.2-9, Table 4.2-10, and Table 4.2-11 show results of the new weekly UC Davis ME-RM,
used beginning March 2019.
Table 4.2-9: Descriptive statistics of XRF-1 results (|ig/cm2) of the new weekly UC Davis ME-RM for the analysis
period 4/4/2019 through 4/23/2019 (see Table 4.2-1 for corresponding sampling dates), N = 2).
Llemenl
U
A\era tic
1.196
Lower Limit
1.075
I pper Limit
1.313
Vii r.xceeilance
0
"" I nacccplablc
u
RSI) "i.
0.4
Si
1.153
1.037
1.268
0
0
0.6
S
9.635
8.672
10.599
0
0
0.1
K
1.289
1.160
1.418
0
0
0.4
Ca
1.161
1.045
1.277
0
0
0.6
Ti
0.049
0.044
0.054
0
0
13.9
Cr
0.461
0.415
0.508
0
0
0.4
Mn
0.236
0.212
0.260
0
0
0.8
Fe
1.290
1.161
1.419
0
0
1.2
Ni
0.085
0.077
0.094
0
0
1.2
Cu
0.353
0.318
0.388
0
0
0.5
Zn
0.352
0.317
0.388
0
0
0.5
Pb
0.227
0.205
0.250
0
0
1.3
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
Page 68 of 126
-------
Table 4.2-10: Descriptive statistics of XRF-4 results (|ig/cm2) of the new weekly UC Davis ME-RM for the analysis
period 3/7/2019 through 4/24/2019 (see Table 4.2-1 for corresponding sampling dates), N = 8.
Element
A1
Average
1.027
Lower Limit
0.917
Upper Limit
1.120
%E\eec(lanee
0
% Unacceptable
u
RSD %
3.8
Si
1.189
1.067
1.304
0
0
2.7
S
9.600
8.599
10.510
0
0
1.0
K
1.300
1.166
1.425
0
0
0.9
Ca
1.167
1.044
1.277
0
0
1.3
Ti
0.048
0.040
0.049
25
0
13.0
Cr
0.464
0.416
0.509
0
0
1.0
Mn
0.240
0.216
0.264
0
0
1.1
Fe
1.282
1.149
1.405
0
0
0.9
Ni
0.083
0.076
0.092
0
0
2.4
Cu
0.352
0.315
0.385
0
0
1.3
Zn
0.344
0.309
0.378
0
0
0.9
Pb
0.249
0.223
0.273
0
0
3.5
Limits are +/-10% of the reference loading (see UCD CSN TI302D).
Table 4.2-11: Descriptive statistics of XRF-5 results (|ig/cm2) of the new weekly UC Davis ME-RM for the analysis
period 3/6/2019 through 4/24/2019 (see Table 4.2-1 for corresponding sampling dates), N = 8.
I. lemon ( \\eraiie Lower l.imil I pper l.imil " \ceeriance "<> I nacccplablc RSI)"..
A1
1.209
1.094
1.337
0
0
1.5
Si
1.117
0.997
1.219
0
0
1.7
S
9.514
8.581
10.487
0
0
0.8
K
1.266
1.141
1.384
0
0
0.4
Ca
1.119
1.005
1.228
0
0
0.8
Ti
0.056
0.052
0.063
0
0
6.1
Cr
0.459
0.413
0.505
0
0
0.6
Mn
0.233
0.210
0.256
0
0
2.2
Fe
1.271
1.146
1.400
0
0
0.9
Ni
0.080
0.071
0.087
0
0
3.2
Cu
0.350
0.34
0.384
0
0
0.9
Zn
0.344
0.309
0.378
0
0
0.7
Pb
0.226
0.203
0.249
0
0
3.6
Limits are +/-10% of the reference loading (see UCD CSN TI 302D).
A Micromatter ME-RM containing A1 and Si is also analyzed weekly. The results from these
analyses are plotted in Figure 4.2-2. The acceptance limits are set as +/- 10% of the average of
the first five measurement results from each EDXRF, thus a shift in the reference value is usually
observed when the instrument is recalibrated. The large shift in the XRF-4 reference value
corresponds with changes made to the EDXRF measurement conditions in an effort to reduce the
inter-instrument bias (see Section 2.3.1 and Section 4.2.2.5 for further details). No issues were
observed other than one drop in A1 on XRF-4. However, no other QC checks for A1 showed any
Page 69 of 126
-------
issue with the instrument during this time and the next measurement returned to acceptable
levels, thus no corrective actions were taken.
Figure 4.2-2: EDXRF results of the weekly Micromatter ME-RM containing A1 and Si. Limits are +/-10% of the
reference loading.
XRF Weekly Micromatter ME-RM Performance
10.00
E
O
CD
O)
C
T3
03
O
9.00
8.00
10.50-
Al
XRF-1
"
"'u\A
(/)
to 10.00-
9.50-
9.00-
8.50-
Si
XRF-1
iiiiiiiriiiiir
co cooocoooooco coco ct> coco ct>ct>
oooooooooooooo
>
Q-co 13.2 => o ฎ9-ro
<5-5^ cna) a> o>
oooooooooooooo
CNJCM CMCN CN CNJCM CM CM CN CMCM CMCM
br >>C="^ CDQ-tt > O C.Q ฃ)=;>>
9-CO Z3-^ Z3 CD^ O
-------
1 71
BiaSi = ~nLdiJ
i=i
r
where n is the number of measurements, j, made of the weekly ME-RM by the EDXRF
instrument over the analysis period.
The precision is estimated by,
Precisioni =
1 -J n(n - 1)
The results from this analysis, for the elements listed for the weekly ME-RM in Table 4.2-2,
averaged over both UC Davis ME-RM samples used during the analysis period, are presented in
Table 4.2-12. Boxplots of the mass loading results from the instruments are presented in figures
4.2-3 and 4.2-4 for each weekly ME-RM sample.
Table 4.2-12: Precision and bias of all EDXRF instruments from the weekly UC Davis ME-RM calculated for the
analysis period 3/23/2018 through 4/24/2019 (see Table 4.2-1 for corresponding sampling dates). Only elements
listed in Table 4.2-2 for the weekly UC ME-RM are evaluated.
1*: lemon 1
\RI-I
liin\ "<>
XRI--4
liiiis " <ฆ
\RI-5
ISiiis "<ฆ
\RI-I
Precision "n
\RI-4
Precision "n
\RI-5
Precision "n
A1
6.4
-7.3
3.4
2.0
3.5
1.5
Si
-0.4
1.7
-0.5
1.2
2.6
1.4
S
0.5
0.1
-0.6
0.5
1.1
0.7
K
0.6
0.3
-0.6
0.6
0.8
0.4
Ca
1.0
0.7
-0.9
1.2
1.8
0.8
Ti
-3.1
-3.6
4.8
7.9
7.8
4.7
V
-0.1
0.6
-2.0
1.8
1.8
1.9
Cr
0.0
0.2
-0.2
0.6
0.9
0.6
Mn
0.2
0.5
-0.5
1.5
1.8
1.8
Fc
1.1
-0.4
-0.3
1.1
1.0
0.7
Ni
2.2
0.8
-1.4
2.0
2.5
2.7
Cu
0.3
0.0
0.4
0.9
1.5
1.0
Zn
1.1
-0.5
1.5
2.4
2.4
1.0
Pb
-2.2
3.3
-1.3
1.7
3.3
2.6
Page 71 of 126
-------
Figure 4.2-3: Instrumental comparison using the weekly UC Davis ME-RM. Bias shown in plot labels is the
maximum of the three instruments. XRF-1: 3/28/2018 to 1/25/2019, N = 41. XRF-4: 3/23/2018 through 3/5/2019, N
= 47. XRF-5: 1/3/2019 through 2/28/2019, N = 9. (See Table 4.2-1 for corresponding sampling dates.)
Ej3 XRF-1 Ejii] XRF-4 Ej3 XRF-5
Al, Bias= 6.8%
K, Bias= 0.8
V, Bias= O.f
Fe. Bias= 1.3%
Zn, Bias= 3.3%
2.700-
2.600-
2.500-:
1.680-'
1.640-
1.600-
1.560-
0.610-
0.600-
0.590-
0.104-
0.100-
0.096-
Si, Bias= 2.1%
Ca, Bias= 0.6
Cr, Bias= 0.2%
Ni, Bias= 0.2%
9.900-
9.800-
9.700-
9.600-
9.500
9.400
0.130
0.125
0.120-
0.115-
0.300-
0.290-
0.280-
0.295-
0.290-
0.285
0.280
Pb, Bias= 1.7%
'
I
I
1
S, Bias= 0.3%
Ti, Bias= 0.6%
1
1
Mn, Bias= O.f
Cu, Bias= 1.0
XRF-1
XRF-4 XRF-5
XRF-1
XRF-4 XRF-5
XRF-1
XRF-4 XRF-5
Page 72 of 126
-------
Figure 4.2-4: Instrumental comparison using the new weekly UC Davis ME-RM. Bias shown in plot labels is the
maximum of the three instruments. XRF-1: 4/4/2019 to 4/23/2019, N = 2. XRF-4: 3/7/2019 through 4/24/2019, N :
8. XRF-5: 3/6/2019 through 4/24/2019, N = 8. (See Table 4.2-1 for corresponding sampling dates.)
Ej3 XRF-1 Ej3 XRF-4 EjB XRF-5
K, Bias= 1.3%
V, Bias= 1.0%
Fe, Bias= 0.9%
XRF-1
Zn, Bias= 2.2%
1.175
1.150
1.125
1.100
0.470-
0.465-
0.460-
0.455-
0.085-
0.082-
0.080
0.078 H
XRF-4 XRF-5
Si, Bias= 3.1%
Ca, Bias= 1.9%
Cr, Bias= 0.5%
Ni, Bias= 4.1%
XRF-1
XRF-4 XRF-5
9.700-
9.600-
9.500-
9.400-
0.060-
0.055-
0.050
0.045-
0.240-
0.235-
0.230-
0.360
0.355-
0.350
0.345 H
S, Bias= 0.7%
Ti, Bias= 9.0%
Mn, Bias= 1.6%
Cu, Bias= 0.5%
XRF-1
XRF-4 XRF-5
4.2.2.4 Long-term Stability, Reproducibility, and Inter-instrument Performance
A set of filters are reanalyzed monthly to monitor the long-term instrument performance; the set
was changed once during 2018. For analyses performed through May 2018, the set consisted of
16 CSN samples and one UC Davis produced ME-RM. The samples were on MTL 47 mm PTFE
filters and covered a range of mass loadings representative of the CSN. The second set of 16
filters, used beginning June 2018, were UC Davis ME-RMs and covered a range of mass
Page 73 of 126
-------
loadings simulating the CSN and higher for trace elements. In order to compare multiple filters
with different mass loadings, the results of reanalysis are first converted to z-scores. For a given
month, the z-score for the ith element and jth filter is
Y- ฆ V. .
AA.,,
zij
U(X y) +
where Xij is that month's result, liis the reference value for element i in filter /, and UXij and
U(ฃy,\
^ y -'are the uncertainty of that month's result and the reference uncertainty respectively. The
instrument-specific reference values for the samples of the reanalysis set are determined as the
mean and standard deviation of five initial measurements, while the values for SRM 2783 are the
certified or reference loadings. Monthly z-scores for each element are then summarized across
the N filters in terms of
Bias; > Zij
and
y,, *
^Ij
Every month, two different reference values are used to calculate z-scores: (1) one reference
value is only based on the average response from the one instrument for which the z-score is
being calculated, and (2) the other reference value is based on the average response from all
instruments. The first z-score serves as long-term reproducibility of each instrument while the
second z-score is an inter-instrumental comparison. These two z-scores are plotted and checked
to be within -1 to 1 for elements which have mass loadings well above the MDL (Al, Si, S, K,
Ca, Ti, Mn, Fe, Zn, Se, and Sr). For further detail see UCD CSN TI302D.
Figure 4.2-5 shows the mean z-score plots over the analysis period. Issues observed include
increasing mean z-scores for Ca on XRF-1 and XRF-4 instruments and low XRF-4 mean z-
scores for Al. The increasing Ca z-scores relate to the previously mentioned Ca contamination on
QC filters (see Section 2.3.3, Section 3.2.1.2, Section 4.2.2.1), and are observed on both sets of
reanalysis filters, occasionally resulting in acceptance criteria exceedances. The XRF-4 low
mean z-score for Al is from bias between the XRF-4 and XRF-1 Al values (Table 4.2-9), which
drives the XRF-4 mean z-score down with respect to the mean reference. However, the XRF-4
mean z-score with respect to its own reference remains constant with only a slight decrease in
September 2018. This indicates the low z-score values are from an inherent bias in the XRF-4 Al
measurement, and are not indicative of instrument change during the analysis period. Changes
were made in the December 2018 calibration to the EDXRF analysis protocols to reduce the
inter-instrument bias and this can be seen in the improved z-score values beginning in January
2019. The reference values for the reanalysis samples were also recalculated in January 2019 to
reflect the changes to that calibration.
Page 74 of 126
-------
Figure 4.2-5: Inter-instrument comparison by z-score of reanalysis sample set. Vertical red line denotes change in
reanalysis set. Multiple measures of the new reanalysis set during the month of June 2018 (denoted by A, B, and C)
were made for determination of reference values.
o XRF-1 vs. mean Ref XRF-4 vs. mean Ref o XRF-5 vs. mean Ref
A XRF-1 vs. XRF-1 Ref A XRF-4 vs. XRF-4 Ref A XRF-5 vs. XRF-5 Ref
1.00
0.50
0.00
-0.50
-1.00
--ha
A o
<5aOq9 O . ~
^M_a2^ILg&g.g
aMoOoo
O o o o
O 0 o Q
1.00-
0.50-
0.00 +ฎ -
-0.50 -
-1.00-
Q O O ฐ 6 o ฐ
A o 0 O o $
1.00-
0.50-
o.oo
-0.50"
-1.00
Mn
mm
1.00
0.50
o.oo 4#%
-0.50 -
-1.00-L
Se
> & ฐ *0 <
1.00
0.50
0.00
-0.50 -
-1.00
9 o
[U
i Aa
Si
Q 6 Q ฎ ฐ
oOoAAฐaAA
tx^-Arr- sta iff
. 0 9 <
O ฉ 0 ฉ O ,
1.00
0.00
-1.00
Ca
a a i 8 A0'
-ฆ- - ;
o 8 o
o
COCOcOT~T~T~T~T~'T~T~'T~'r~T~
OOOt-^^OOOOOOOOOO
(NCNJCNIOOOCN,CNJO
-------
analysis period. XRF-I and XRF-4 underwent routine calibrations in January 2018 and all three
instruments were calibrated at the end of December 2018. XRF-4 also underwent calibration in
June 2018 due to a replacement of the CaF2 target in the secondary target wheel. The results
from the monthly NIST SRM 2783 analyses indicate that calibrations for all instruments are
stable over the calibration periods. The overall error for most elements is less than 20%.
However, the error in Zn is around 30%. Per Yatkin et al. (2016b), an XRF interlaboratory
comparison reported SRM 2783 Zn error varying from -15% to 30%; the results shown here fit
wi thin that range. There were no exceedances of the acceptance criteria during this analysi s
period.
Figure 4.2-6: Error of each EDXRF instalment from the NIST SRM standard run monthly for the analysis period
3/23/2018 through 4/26/2019.
ฐ XRF-1 o XRF-4 o XRF-5 Acceptance Limit
100%-
50% -
0%
-50% -
-100%
Ca
ง 8 o %
0 ^
Q O
CO
00
1 ฎ
CO
ฉ o ฃ
O O &
Q
0
Mn
O
G
ง q ฐ O
t '<%
$ O
0 Q ฐ
O o _
o
ง
O O ฐ
10%
5%
0%
-5%
-10%
Fe
cocooocococococococoa)a)cr>a>cr>
oococococococooocococncncncna)
O) Q. q
, =J <1> X
< c/) O
> U C ฃ ^ ^ >
O (D CD r >.
u_ ^ ^
COCO(X)COCOCOCOCOCOCOO)0>CJ)0)05
: >, e ฆ
! 9- (O zj -
: < ^ "5
O) Q-
< (0 O
> U C JD
Page 76 of 126
-------
Table 4.2-13: Dates for calibrations performed on each EDXRF instrument during this analysis period.
I-'.DXKI-' Calibration Reason lor
lusl I'll IllC'lll
Dale
( alibralion
Kanue of Sample Dales \ii;il>/c(l
XRF-1
1/19/2018
Annual calibration
1/14/2018-9/29/2018
XRF-4
1/19/2018
Annual calibraiton
1/2/2018-3/30/2018
XRF-4
6/15/2018
CaF2 secondary
target was replaced
4/2/2018 - 9/29/2018
XRF-5
12/17/2018
Annual calibration
10/2/2018- 12/31/2018
XRF-4
12/19/2018
Annual calibration
10/5/2018- 12/23/2018
XRF-1
12/21/2018
Annual calibration
10/11/2018- 12/28/2018
In addition to the calibration verification QC results shown in Figure 4.2-6, the conditions under
which the EDXRF instruments measure the elemental results were changed in December 2018
(see Section 2.3.1). The EDXRF measurement applications, which include the secondary targets
and integration times, were adjusted in an effort to reduce the variability and detection limits for
some elements. Also, efforts were made to improve the bias between instruments, especially for
low atomic weight elements such as A1 and Si. The measurement applications for calibrations
prior to the December 2018 calibration and after that calibration are compared in Table 4.2-14.
The results of the application change are still being analyzed to determine the effect on
variability and detection limits, although inter-instrument bias has improved for the low atomic
weight elements.
Table 4.2-14: EDXRF measurement condition changes.
( ;
ilibralion .lanuan 20IX
( alibralion December 20IX
Seeondan
1 ai'uel
Time
(seconds)
I'llemenls
Seeondan
lariicl
Time
(seeonds)
I'llemenls
CaF2
600
Na, Mg, Al, Si, P,
S, CI, K
CaF2
600
Na, Mg, Al, Si, P,
S, CI, K
Fe
400
Ca, Ti, V, Cr
Fe
400
Ca, Ti, V, Cr
Ge
300
Mn, Fe, Co, Ni,
Ge
400
Mn, Fe, Co, Ni,
Cu, Zn
Cu, Zn
KBr
300
As
SrF2
500
As, Se, Br
SrF2
300
Sc. Br
Mo
500
Rb, Sr
Mo
300
Rb, Sr, Pb
A1203
500
Zr, Ag, Cd, In,
Sn, Sb, Cs, Ba, Ce
A1203
200
Zr, Sn, Sb, Cs,
Ba, Ce
Zr
500
Pb
Csl
200
Ag, Cd, In
4.2.3 Determination of Uncertainties and Method Detection Limits
For discussion of Method Detection Limits (MDLs) see Section 3.1.3.2.
For discussion of analytical uncertainty and total uncertainty see Section 3.1.2 and Section 6.5,
respectively.
Page 77 of 126
-------
4.2.4 Audits, Performance Evaluations, Training, and Accreditations
4.2.4.1 System Audits
The EPA did not conduct any audits or performance evaluations of the UC Davis XRF
Laboratory during this reporting period.
4.2.4.2 Performance Evaluations
The UC Davis XRF Laboratory actively participates in interlaboratory comparisons.
In 2018 (during the analysis period for samples collected during 2018), UC Davis participated in
an interlaboratory comparison with Environment and Climate Change Canada. Q1SO4 and CuO
reference materials, generated at UC Davis, were analyzed by XRF, IC, and ICP-MS. Results
indicate agreement between the laboratories with less than 5% absolute difference.
4.2.4.3 Training
Training of all personnel who assist with or operate the EDXRF instruments is mandatory
through UC Davis. Personnel in the XRF laboratory are required to take the following UC Davis
safety trainings: UC Laboratory Safety Fundamentals, Radiation Safety for Users of Radiation
Producing Machines, Analytical X-ray Quiz, and Cryogen Safety.
Only personnel listed in UC Davis CSN Quality Assurance Project Plan (QAPP), trained on the
appropriate SOPs and Technical Instructions (CSN SOP 302 and CSN TI302A-D), and
authorized by the Laboratory Manager can perform EDXRF analysis on CSN samples.
4.2.4.4 A ccreditations
There are no accreditations for elemental analysis on aerosol filters by EDXRF.
4.2.5 Summary of Filter Field Blanks
Over the sampling period (January 1, 2018 through December 31, 2018) there were 1,665 valid
PTFE filter field blanks. Table 4.2-15 summarizes the field blank statistics.
Page 78 of 126
-------
Table 4.2-15: PTFE filter field blank statistics for the analysis period 3/23/2018 through 4/26/2019 (samples
collected 1/1/2018 through 12/31/2018).
Species
Count
Median
(Hg/cm:)
Average
(Hg/cm:)
Min
Qig/cnr)
Max
(Hg/cm:)
St. Dev.
Oig/cnr)
Ag
1665
0.014
0.015
0.002
0.049
0.006
A1
1665
0.078
0.079
0.023
0.258
0.015
As
1665
0.000
0.000
0.000
0.006
0.001
Ba
1665
0.089
0.090
0.000
0.183
0.026
Br
1665
0.002
0.002
0.000
0.008
0.002
Ca
1665
0.003
0.006
0.000
0.162
0.009
Cd
1665
0.014
0.015
0.003
0.040
0.006
Ce
1665
0.107
0.111
0.032
0.219
0.034
CI
1665
0.002
0.004
0.000
0.608
0.019
Co
1665
0.002
0.002
0.000
0.021
0.001
Cr
1665
0.004
0.005
0.002
0.072
0.003
Cs
1665
0.060
0.061
0.000
0.141
0.020
Cu
1665
0.006
0.007
0.000
0.029
0.003
Fe
1665
0.022
0.023
0.000
0.262
0.012
In
1665
0.033
0.033
0.010
0.063
0.009
K
1665
0.008
0.009
0.002
0.257
0.007
Mg
1665
0.012
0.014
0.000
0.091
0.015
Mn
1665
0.006
0.007
0.000
0.017
0.002
Na
1665
0.000
0.014
-0.002
0.270
0.024
Ni
1665
0.001
0.001
0.000
0.022
0.001
P
1665
0.000
0.000
0.000
0.028
0.001
Pb
1665
0.014
0.014
0.003
0.032
0.004
Rb
1665
0.004
0.004
0.000
0.014
0.002
S
1665
0.000
0.001
0.000
0.286
0.013
Sb
1665
0.038
0.040
0.009
0.093
0.014
Se
1665
0.003
0.003
0.000
0.009
0.001
Si
1665
0.012
0.015
0.000
0.438
0.017
Sn
1665
0.039
0.040
0.010
0.086
0.013
Sr
1665
0.005
0.005
0.002
0.012
0.002
Ti
1665
0.001
0.001
0.000
0.065
0.002
V
1665
0.000
0.000
0.000
0.002
0.000
Zn
1665
0.003
0.003
0.000
0.049
0.002
Zr
1665
0.022
0.023
0.000
0.074
0.010
Page 79 of 126
-------
4.3.A DRI Thermal Optical Analysis Laboratory
The DRI Thermal Optical Analysis Laboratory, as a subcontractor to UC Davis, received and
analyzed quartz filters from batches 39 through 47 covering the sampling period January 1, 2018
through September 30, 2018. Analysis of these samples was performed May 23, 2018 through
December 17, 2018. All analyses were performed using the DRI Model 2015 multi-wavelength
carbon analyzer with the IMPROVE A protocol and analysis results were reported to UC Davis.
Thirteen DRI Model 2015 Thermal Optical Carbon Analyzers (designated as units # 21, 31, 32,
34-38, 40-43, and 47) were used for these CSN analyses.
4.3.A. 1 Summary of QC Checks and Statistics
Samples were received by the DRI Thermal Optical Analysis Laboratory following the chain-of-
custody procedure specified in DRI CSN SOP # 2-231 (specific to CSN). Samples are analyzed
using DRI Model 2015 analyzers following DRI CSN SOP 2-226r3. Quality control (QC)
measures for the DRI carbon analysis are included in the SOP and summarized in Table 4.3.A-1.
The table specifies the frequency and standards required for the checks, along with the
acceptance criteria, and corrective actions for the carbon analyzers. More detail on individual
control measures is provided in specific subsections.
Page 80 of 126
-------
Table 4.3.A-1 DRI quality control measures for carbon analysis by TOA (Model 2015 analyzer).
QA/QC Activity
Calibration Standard and
Range
Calibration
Frequency
Acceptance Criteria
Corrective Action
System Blank Check
NAa
Once per week
<0.2 \ig C/cm2. See Table 4.3.A-2
and Figure 4.3.A-1.
Check instrument.
Laboratory Blank
Check
NA
Beginning of
analysis day
<0.2 ng C/cm2. See Table 4.3.A-3
and Figure 4.3.A-2.
Check instrument and filter punch
and rebake
Calibration
Peak Area Check
NIST 5% CH4/He gas standard;
20 \ig C (6-port valve injection
loop, 1000 fj.1)
Every analysis
Typical counts 15,000-25,000 and 95-
105% of average calibration peak area
of the day. See Figure 4.3.A-4.
Void analysis result; check flowrates,
leak, and 6-port valve temperature;
conduct an auto-calibration; and
repeat analysis with second filter
punch.
Auto-Calibration
Check
NIST 5% CH4/He gas standard;
20 \ig C (Carle valve injection
loop, 1000 fj.1)
Alternating
beginning or
end of each
analysis day
Relative standard deviation of the
three injection peaks <5% and
calibration peak area 90-110% of
weekly average. See Table 4.3.A-4
and Figure 4.3.A-3.
Verify if major maintenance has
occurred. Troubleshoot and correct
system before analyzing samples.
Manual Injection
Calibration
NIST 5% CH4/He or NIST 5%
CCh/He gas standards; 20 \ig C
(Certified gas-tight syringe, 1000
Hi)
Four times a
week (Sun.,
Tue., Thu., and
Sat.)
95-105%) recovery and calibration
peak area 90-110%o of weekly
average. See Figure 4.3.A-5a.
Troubleshoot and correct system
before analyzing samples.
Sucrose Calibration
Check
10|xL of 1800 ppm C sucrose
standard; 18 \ig C
Thrice per
week (began
March, 2009)
17.1-18.9 \ig C/filter. See Figure
4.3.A-5b.
Troubleshoot and correct system
before analyzing samples.
Potassium Hydrogen
Phthalate (KHP)
Calibration Check
10|xL of 1800 ppm C KHP
standard; 18 \ig C
Twice per
week (Tue. and
Thu.)
17.1-18.9 \ig C/filter. See Figure
4.3.A-5c.
Troubleshoot and correct system
before analyzing samples.
Multiple Point
Calibrations
1800 ppm C Potassium hydrogen
phthalate (KHP) and sucrose;
NIST 5% CH4/He, and NIST 5%
CCh/He gas standards; 9-36 \ig
C for KHP and sucrose; 2-30 \ig
C for CH4 and CO2
Every six
months or after
major
instrument
repair
All slopes ฑ5%o of average. See Table
4.3.A-5.
Troubleshoot instrument and repeat
calibration until results are within
stated tolerances.
Sample Replicates
(on the same or a
different analyzer)
NA
Every 10
analyses
ฑ10%o when OC and TC >10 |ig
C/cm2
ฑ20%o when EC > 10|ig C/cm2 or
<ฑ1 (rg/cm2 when OC and TC <10 \ig
C/cm2
<ฑ2 (rg/cm2 when ECR <10|ig C/cm
See Table 4.3.A-8 and Figure 4.3.A-
6, and Table 4.3.A-9 and Figure
4.3.A-7.
Investigate instrument and sample
anomalies and rerun replicate when
difference is > ฑ10%o.
Temperature
Calibrations
NIST-certified thermocouple
Every six
months, or
whenever the
thermocouple
is replaced
Linear relationship between analyzer
and NIST thermocouple values with
R2>0.99. See Table 4.3.A-6.
Troubleshoot instrument and repeat
calibration until results are within
stated tolerances.
Oxygen Level in
Helium Atmosphere
(using GC/MS)b
Certified gas-tight syringe; 0-
100 ppmv
Every six
months, or
whenever leak
is detected
Less than the certified amount of He
cylinder. See Table4.3.A-7.
Replace the He cylinder and/or O2
scrubber.
a NA: Not Applicable.
b Gas chromatography/mass spectrometer (Model 5975, Agilent Technology, Palo Alto, CA, USA).
4.3.A.2 Summary of QC Results
Detailed results of the carbon QC are presented in the subsections below. All system blanks
(Table 4.3-2) or laboratory blanks (Table 4.3-3) that did not meet the acceptance criteria were
reanalyzed and if they did not pass the second analysis, instrument maintenance was performed
Page 81 of 126
-------
and additional blanks were ran before the analyzer was placed on-line. Exceedance in multipoint
calibrations (Table 4.3-5) result in verification of individual calibration points, troubleshooting
the instrument, and repeating calibrations. Exceedances in auto-calibrations (Table 4.3-4),
internal calibrations (Figure 4.3-4), as well as CO2 (Figure 4.3-5a), sucrose (Figure 4.3-5b), and
KHP (Figure 4.3-5c) calibrations result in reanalysis and/or instrument maintenance. For cases
where CSN samples were analyzed after an exceedance, data were flagged with the QX (Does
Not Meet QC Criteria) qualifier in files delivered to AQS by UC Davis (see Section 2.4.2 and
Section 3.2.3.1).
4.3. A. 2.1 System and Laboratory Blanks
Table 4.3.A-2 lists the number of system blanks analyzed during the report period and their
concentration statistics. The system blank control chart is shown in Figure 4.3.A-1. System
blanks are used to ensure that the system is not introducing bias in the carbon analysis. Most
system blanks were below the limit of 0.2 |igC/cm2. When an exceedance is observed, possible
contamination is checked, suspicious parts are cleaned, the sample oven is baked, and a second
system blank is rerun to ensure that it passes the criterion. Two system blanks did not pass and
nine CSN samples were analyzed after this system blank failure. These cases were flagged with
the QX (Does Not Meet QC Criteria) qualifier in files delivered to AQS by UC Davis (see
Section 2.4.2 and Section 3.2.3.1). As a corrective action, software tools are being developed to
generate QC control charts and summaries to ensure QC exceedances are captured and corrected
immediately.
Table 4.3.A-2: Statistics of system blanks ran on the Model 2015 analyzer for the analysis period 5/23/2018 through
12/17/2018 (samples collected 1/1/2018 through 9/30/2018). Elemental carbon (EC) fractions are indicated as (1)
through (3), organic carbon (OC) fractions are indicated as (1) through (4). Organic pyrolyzed (OP), elemental
carbon (EC), and organic carbon (OC) are shown by reflectance (R) and transmittance (T).
Parameter
( mini
Median
(uii/cnr I
A\era lie
(.uii/enr)
Min
(.uii/cnr)
Max
(uii/cnr I
Si. I)e\.
(uii/cnr I
# r.xeeedanee
OC1
308
0.000
0.000
0.000
0.050
0.003
0
OC2
308
0.000
0.001
0.000
0.048
0.004
0
OC3
308
0.000
0.008
0.000
0.229
0.023
1
OC4
308
0.000
0.001
0.000
0.089
0.006
0
OCR
308
0.000
0.011
0.000
0.286
0.028
1
OCT
308
0.000
0.012
0.000
0.318
0.032
2
OPR
308
0.000
0.000
0.000
0.012
0.001
0
OPT
308
0.000
0.002
0.000
0.126
0.009
0
EC1
308
0.000
0.000
0.000
0.028
0.003
0
EC2
308
0.000
0.002
0.000
0.122
0.008
0
EC3
308
0.000
0.000
0.000
0.033
0.002
0
ECR
308
0.000
0.002
0.000
0.126
0.010
0
ECT
308
0.000
0.000
0.000
0.038
0.003
0
TC
308
0.001
0.013
0.000
0.318
0.033
2
Page 82 of 126
-------
Figure 4.3.A-1: Control chart of system blank total carbon concentrations on the DRI Model 2015 carbon analyzers.
The red dash lines indicate the limit of 0.2 |ig C/cm2.
O
o>
c
0)
o
c
o
o
o
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
System Blanks; N=308
Analysis Date
Table 4.3.A-3 lists the number of laboratory blanks analyzed during the report period and their
concentration statistics. The laboratory blank control charts are shown in Figure 4.3.A-2.
Laboratory blank analyses are performed daily to check for system contamination and evaluate
laser response. All 3,308 laboratory blanks were below the limit of 0.20 |igC7cm2 during this
reporting period. When an exceedance is observed, the sample oven is baked and a second
laboratory blank is run. If the second blank still exceeds the limit, the analyzer is taken offline
for cleaning and maintenance.
Page 83 of 126
-------
Table 4.3.A-3: Statistics of laboratory blanks run on the Model 2015 analyzer for the analysis period 5/23/2018
through 12/17/2018 (samples collected 1/1/2018 through 9/30/2018). Elemental carbon (EC) fractions are indicated
as (1) through (3), organic carbon (OC) fractions are indicated as (1) through (4). Organic pyrolyzed (OP), elemental
carbon (EC), and organic carbon (OC) are shown by reflectance (R) and transmittance (T).
Parameter
Count
Median
frig/cm2)
Average
frig/cm2)
Min
frig/cm2)
Max
frig/cm2)
St. Dev.
frig/cm2)
# Exceedance
OC1
3308
0.000
0.000
0.000
0.079
0.002
0
OC2
3308
0.000
0.000
0.000
0.068
0.003
0
OC3
3308
0.000
0.001
0.000
0.130
0.005
0
OC4
3308
0.000
0.001
0.000
0.078
0.004
0
OCR
3308
0.000
0.003
0.000
0.187
0.012
0
OCT
3308
0.000
0.004
0.000
0.190
0.015
0
OPR
3308
0.000
0.001
0.000
0.184
0.006
0
OPT
3308
0.000
0.002
0.000
0.184
0.009
0
EC1
3308
0.000
0.000
0.000
0.078
0.003
0
EC2
3308
0.000
0.002
0.000
0.150
0.009
0
EC3
3308
0.000
0.000
0.000
0.095
0.002
0
ECR
3308
0.000
0.002
0.000
0.150
0.010
0
ECT
3308
0.000
0.001
0.000
0.138
0.007
0
TC
3308
0.000
0.005
0.000
0.190
0.017
0
Figure 4.3.A-2: Control chart of daily laboratory blank total carbon concentrations run on the DRI Model 2015
carbon analyzers for the analysis period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through
9/30/2018). The red dash lines indicate the limit of 0.2 |igC7cnr.
Laboratory Blanks; N=3308
1 n
0.9
-S 0.8
O)
ฆ3 07
ง 0-6
^ 0.5
ฆ| 0.4
S 0.3
O 0.2
o ฐ-i
1- 0.0
~ TP
tit!
ULi
5/22/2018
6/21/2018
7/21/2018
ฃ
|r 8/20/2018
"
D
jjf 9/19/2018
10/19/2018
11/18/2018
12/18/2018
4.3.A.2.2 Auto-Calibration and Internal Calibration Peak Area Check
Once per day each analyzer runs an auto-calibration protocol. Using the Carle valve, an aliquot
of methane standard is injected once in a He-only atmosphere (organic carbon stage), once in a
He/02 atmosphere (elemental carbon stage), and finally as the normal internal calibration peak.
The three peaks should have similar peak areas if the catalysts are in good condition. The
similarity of the three peaks are measured by the relative standard deviation (RSD), which is the
standard deviation divided by the average of the three peak areas. The acceptance limit is RSD
Page 84 of 126
-------
<5% and ฑ10% from weekly average. Table 4.3.A-4 summarizes the RSD of the three methane
injection peaks during the analysis period and the control chart is shown in Figure 4.3.A-3. There
were three exceedances during this reporting period. When an exceedance is observed, the
analyzer is checked and the auto-calibration is rerun. The calibration peak areas of previous runs
are examined and/or manual injections are done to ensure the analyzer is working properly. A
total of 106 CSN samples were analyzed during auto-calibration peak area exceedances; these
cases were flagged with the QX (Does Not Meet QC Criteria) qualifier in files delivered to AQS
by UC Davis (see Section 2.4.2 and Section 3.2.3.1). As a corrective action, software tools are
being developed to generate QC control charts and summaries to ensure QC exceedances are
captured and corrected immediately.
Table 4.3.A-4: Statistics of the relative standard deviation (RSD) of the three methane injection peaks from auto-
calibration checks for the analysis period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through
9/30/2018).
Statistic
Auto-Calibration
Count
2343
Median
0.71%
Average
0.93%
Min
0.01%
Max
5.5%
Standard deviation
0.71%
Exceedance
3
Figure 4.3.A-3: Control chart of the relative standard deviation of the three methane injection peaks from daily
auto-calibration ran on the DRI Model 2015 carbon analyzers for the analysis period 5/23/2018 through 12/17/2018
(samples collected 1/1/2018 through 9/30/2018). The red dash lines indicate the limit of 5% RSD.
Auto Calibration; N=2343
g
C
o
.5
>
o
D
T3
k.
TO
T3
C
TO
*->
CO
d)
>
2
0)
q:
Analysis Date
At the end of each run, a fixed amount of methane is injected via a Carle valve as an internal
calibration standard. The internal calibration peak area is examined for each sample. Significant
changes in calibration peak area counts are monitored and instruments are checked for
performance against daily calibrations. Typical ranges for the internal calibration peaks fall
Page 85 of 126
-------
between 15,000 and 25,000 counts for Model 2015. In addition to peak area ranges, the peak
areas are also compared to the daily averages. Sudden changes or atypical counts result in
instrument maintenance. Metadata concerning QC measures and instrument maintenance are
reported to UC Davis quarterly. Figure 4.3.A-4 shows the daily internal calibration peak area
during the reporting period for all analyzers. During this reporting period, 12,583/12,595
(99.9%) passed both peak area and daily average criteria. There were 12 CSN samples analyzed
during exceedance of internal calibration QC limits; these cases were flagged with the QX (Does
Not Meet QC Criteria) qualifier in files delivered to AQS by UC Davis (see Section 2.4.2 and
Section 3.2.3.1). However, other QC analyses (e.g. replicates, auto-calibration, and internal
calibration peak area check) within the time period indicated acceptable values. As a corrective
action, software tools are being developed to generate QC control charts and summaries to
ensure QC exceedances are captured and corrected immediately.
Figure 4.3. A-4: Control chart of the internal calibration peak area for the DRI Model 2015 carbon analyzers. The
red dash lines indicate the typical internal calibration peak area between 15,000 and 25,000 for Model 2015. Twelve
samples were run after there was an exceedance outside of the daily average (but within the typical peak area).
Internal Calibration; N=12595
35000
30000
wT
g 25000
omamtliM:
0)
Q_
o 10000
TO
ฃ 5000
TO
o
0
00 00 00 00 00 00 00
o o o o o o o
CM CM CM CM CM CM CM
CN ^ ^ O O) O) CO
CM (N CM CM ^ ^
in
-------
provides summary statistics for full multipoint calibrations by analyzer for the period during
which the project samples were analyzed. All multipoint calibrations met the QC criterion during
this reporting period.
Table 4.3.A-5: Multipoint calibration statistics for the analysis period 5/23/2018 through 12/17/2018 (samples
collected 1/1/2018 through 9/30/2018). Units for the slope are |ig carbon per ratio of standard injection peak
count/calibration gas peak count.
( iii'hon
Anali/or
( alihralion
Dale
Slope
r
Difference from
An;il\/cr A\era;ie
# of Samples
Haiiiicd
21
5/30/2018
19.348
0.998
2%
0
7/25/2018
19.155
0.996
1%
0
11/20/2018
18.548
0.992
-3%
0
12/20/2018
19.069
0.998
0%
0
31
8/22/2018
19.627
0.996
5%
0
11/16/2018
17.851
0.996
-5%
0
32
6/29/2018
17.870
0.998
3%
0
8/16/2018
16.611
0.995
-4%
0
9/3/2018
17.489
0.995
1%
0
10/11/2018
17.419
0.996
1%
0
12/2/2018
17.021
0.988
-2%
0
34
8/10/2018
18.919
0.995
-2%
0
10/12/2018
19.560
0.995
2%
0
35
9/6/2018
19.134
0.997
0%
0
10/11/2018
19.304
0.992
0%
0
36
5/30/2018
18.853
0.995
-2%
0
6/26/2018
19.547
0.996
2%
0
7/11/2018
19.658
0.9966
2%
0
7/27/2018
19.027
0.9955
-1%
0
9/4/2018
19.302
0.987
0%
0
9/28/2018
18.885
0.998
-2%
0
37
6/1/2018
19.180
0.998
0%
0
6/26/2018
19.242
0.998
0%
0
12/17/2018
19.208
0.999
0%
0
38
6/26/2018
18.964
0.996
-1%
0
12/20/2018
19.167
0.998
1%
0
40
7/25/2018
19.636
0.998
0%
0
8/10/2018
19.596
0.995
0%
0
41
7/27/2018
19.538
0.997
2%
0
8/27/2018
19.332
0.995
0%
0
11/8/2018
18.518
0.994
-4%
0
12/11/2018
19.584
0.999
2%
0
42
5/17/2018
19.090
0.997
0%
0
Page 87 of 126
-------
Carbon
Analyzer
Calibration
Date
Slope
r2
Difference from
Analyzer Average
# of Samples
Flagged
9/5/2018
18.911
0.998
0%
0
43
6/5/2018
18.022
0.971
-3%
0
7/26/2018
18.984
0.998
3%
0
47
7/24/2018
18.435
0.996
1%
0
9/25/2018
18.234
0.998
-1%
0
C02 calibrations are performed on each analyzer four times per week, sucrose calibration checks
are done on each analyzer three times per week, and KHP calibrations are done twice per week.
Calibration control charts for the Model 2015 analyzers are shown in Figures 4.3.A-5a through
4.3.A-5c. For this reporting period, 4 out of 1,083 CO2 calibrations, 6 out of 1,106 sucrose
calibration, and 10 out of 944 KHP calibrations exceeded the criteria. When an exceedance is
observed, the analyzer is checked and the calibration is rerun. In some cases, a full carbon
calibration was conducted. There were 9, 29, and 49 CSN samples analyzed after CO2, sucrose,
and KHP exceedances, respectively; these cases were flagged with the QX (Does Not Meet QC
Criteria) qualifier in files delivered to AQS by UC Davis (see Section 2.4.2 and Section 3.2.3.1).
However, for all samples that were run after an exceedance calibration, other QC analyses (i.e.,
replicates, auto-calibration, and internal calibration peak area checks) within the time period
indicate acceptable values. As a corrective action, software tools are being developed to generate
QC control charts and summaries to ensure QC exceedances are captured and corrected
immediately.
Figure 4.3.A-5: Control chart of manual calibration checks for: (a) CO2, (b) sucrose, and (c) KHP injections for the
analysis period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through 9/30/2018). The red dash lines
indicate the total carbon limits of 17.1 and 18.9 |igC per injection for sucrose and KHP and 19.57 and 21.63 |iC per
injection for CO2.
(a)
(b)
(c)
C02 Calibration; N=1083
Sucrose Calibration; N=1106
KHP Calibration; N=944
>!/___
Analysis Date
4.3.A.2.4 Temperature Calibrations
Table 4.3.A-6 provides summary statistics for the multi-point temperature calibrations of each
Model 2015 carbon analyzer. The temperature calibrations are performed every six months or
after a major instrument repair. Criteria for an acceptable calibration is linear regression
coefficient of determination (r2) > 0.99. Separate linear regressions are used for the lower
temperatures and higher temperature ranges. These two ranges are separated with a toggle point
typically around 200-300 ฐC, which is set to the temperature at which the two regression lines
intercept. All calibrations met the acceptable r2 criteria (r2 > 0.99) during this report period.
Page 88 of 126
-------
Table 4.3.A-6: Multi-point temperature calibration statistics on the Model 2015 carbon analyzer for the analysis
period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through 9/30/2018).
Carbon
Analvzcr
Calibration
Date*
Low T
Slope
Low T
Intercept
Low T
1*2
High T
Slope
High T
Intercept
High T
1*2
21
5/25/2018
1.038
11.702
0.999
0.989
25.596
0.999
11/16/2018
1.066
3.619
**
1.008
19.441
**
31
8/17/2018
1.048
7.391
0.999
0.984
24.259
0.999
32
9/26/2018
1.033
4.141
0.999
1.018
11.423
0.999
34
7/16/2018
1.069
2.360
1.000
1.001
20.822
1.000
9/10/2018
1.035
4.474
0.999
0.977
20.427
0.999
35
10/4/2018
1.047
1.344
0.999
1.014
11.078
0.999
36
6/5/2018
0.998
14.056
1.000
0.986
19.062
1.000
7/6/2018
1.039
12.666
0.999
0.984
27.854
0.999
7/23/2018
1.091
12.163
0.998
0.989
39.248
0.999
9/25/2018
1.048
9.961
0.999
0.993
26.625
0.999
11/7/2018
1.068
4.743
0.999
1.008
21.873
0.999
37
5/29/2018
1.060
4.289
1.000
0.978
27.330
1.000
6/13/2018
1.049
8.293
0.999
0.975
27.800
0.999
12/9/2018
1.039
0.227
0.999
1.004
9.502
0.999
38
6/18/2018
1.096
2.330
1.000
1.006
27.547
1.000
12/14/2018
1.142
-1.087
0.999
1.010
36.195
0.999
40
8/1/2018
0.999
8.688
0.999
0.989
11.705
0.999
10/16/2018
1.068
2.923
0.999
1.025
15.200
0.999
12/14/2018
1.094
1.094
0.999
1.021
20.659
0.999
41
7/23/2018
1.025
9.735
1.000
1.005
16.973
1.000
42
8/28/2018
0.981
11.410
0.999
0.999
7.769
0.999
11/26/2018
1.053
9.056
0.999
1.037
15.267
0.999
43
5/18/2018
1.050
5.926
1.000
1.005
19.394
1.000
5/29/2018
1.060
4.289
1.000
0.978
27.330
1.000
8/1/2018
1.122
-9.090
0.998
0.989
27.277
0.999
9/2/2018
1.112
2.344
1.000
0.998
33.237
1.000
10/23/2018
1.144
-0.606
1.000
1.014
35.429
1.000
47
7/18/2018
1.037
9.090
1.000
0.988
23.588
1.000
* Includes both regular maintenance and semi-annual calibration data
** Calibration point data were deleted from file, therefore r2 data not available
4.3.A.2.5 Oxygen Level Check
Table 4.3.A-7 provides a summary of the Model 2015 oxygen leak test results that are performed
every six months or after major instrument repairs. The results are considered acceptable if the
O2 concentration is < 100 ppm. The O2 contents were well below 100 ppm, in the range of 17-73
ppm.
Page 89 of 126
-------
Table 4.3.A-7: Model 2015 oxygen test statistics for the analysis period 5/23/2018 through 12/17/2018 (samples
collected 1/1/2018 through 9/30/2018).
Analvzcr
O2 Statistics
Feb 2018
Aug 2018
No.
(ppm)
140 (ฐC)
580 (ฐC)
140 (ฐC)
580 (ฐC)
21
Mean O2
17.9
16.7
32.6
30.1
Std Dev
4.7
4.6
9.9
9.8
31
Mean O2
19.8
18.3
28.7
23.9
Std Dev
4.6
4.6
10.2
9.9
32
Mean O2
24.8
26.5
47.2
48.0
Std Dev
4.7
4.8
10.5
10.1
34
Mean O2
39.1
50.5
Offline
Offline
Std Dev
5.6
5.6
Aug 2018
Aug 2018
35
Mean O2
22.6
26.6
35.1
32.5
Std Dev
4.7
4.8
10.0
9.7
36
Mean O2
20.0
22.7
21.9
24.3
Std Dev
4.7
4.7
9.7
9.7
37
Mean O2
18.8
16.6
37.4
33.7
Std Dev
4.9
4.6
10.0
9.7
38
Mean O2
31.2
28.1
35.9
29.2
Std Dev
4.9
4.7
9.8
9.8
40
Mean O2
24.3
25.3
42.2
47.3
Std Dev
4.7
4.8
9.9
10.5
41
Mean O2
23.8
20.9
25.3
22.7
Std Dev
4.8
4.7
9.8
9.7
42
Mean O2
17.5
16.7
66.4
72.6
Std Dev
4.8
4.7
12.4
10.7
43
Mean O2
26.7
24.8
33.6
32.3
Std Dev
5.0
4.8
9.8
9.9
47
Mean O2
17.8
16.8
46.6
45.0
Std Dev
4.7
4.8
9.9
9.8
4.3.A. 2.6 Replicate and Duplicate Analyses
Replicate analysis results are from two or more punches from the same sample filter analyzed on
different instruments. Duplicate analysis results are from two punches from the same sample
filter analyzed on the same instruments. A replicate or duplicate analysis was performed
randomly on one sample from every group of 10 samples. Table 4.3.A-8 and 4.3.A-9 give the
criteria and summary statistics for replicate and duplicate carbon analyses, respectively, for this
reporting period. Control charts for replicate and duplicate analyses are shown in Figure 4.3.A-6
and Figure 4.3.A-7, respectively.
Replicate analysis results for total carbon (TC), organic carbon (OCR), and elemental carbon
(ECR) by reflectance agree well, with only 24/3876 data points (0.6%) for OCR, ECR, and TC
exceeding the criteria. Duplicate analysis results for total carbon (TC), organic carbon (OCR),
and elemental carbon (ECR) by reflectance agree well, with zero data points exceeding the
criteria. Samples not meeting replicate/duplicate criteria (i.e., for TC, OCR, or ECR < 10 ng
C/cm2, TC, OCR < ฑ 1.0 jag C/cm2 and ECR < ฑ 2.0 jag C/cm2; and for TC, OCR or ECR > 10
Page 90 of 126
-------
[j,g C/cm2, TC or OCR < 10% RPD and ECR < 20% RPD) are reanalyzed, typically on a third
analyzer. However, the small size (25 mm) of the filter does not permit more than three punches
(each -0.5 cm2) to be taken from the filter. Filter inhomogeneity, which is flagged prior to first
analysis, is also examined.
Table 4.3.A-8: Replicate analysis criteria and statistics for the analysis period 5/23/2018 through 12/17/2018
(samples collected 1/1/2018 through 9/30/2018). Total carbon (TC), organic carbon (OCR), and elemental carbon
(ECR) are shown by reflectance.
Replicates
Range
Criteria
Statistic
TC
oc
EC
TC, OCR, & ECR
TC, OCR < ฑ1.0 ng C/cm2
Count
69
72
294
<10 |ig C/cm2
ECR < ฑ2.0 |ig C/cm2
No. Fail
1
1
21
%Fail
1.4
1.3
6.8
Units: |ig C/cm2
Mean
0.30
0.30
0.72
StdDev
0.29
0.26
0.73
Max
1.73
1.25
3.90
Min
0.00
0.00
0.00
Median
0.22
0.23
0.44
TC, OCR, & ECR
TC, OCR %RPD < 10%
Count
1223
1220
998
> 10 |ig C/cm2
ECR %RPD < 20%
No. Fail
0
1
0
%Fail
0
0.10
0
Units: %
Mean
1.77
2.12
5.03
StdDev
1.22
1.52
3.63
Max
5.94
15.22
17.86
Min
0.00
0.00
0.00
Median
1.53
1.86
4.41
Note: RPD= 100 x absolute value [original sample-duplicate sample]/[(original sample+ duplicate sample)/2]
Page 91 of 126
-------
Figure 4.3.A-6: Replicate (two punches from the same sample filter analyzed on different instruments) analysis
results for the analysis period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through 9/30/2018). The
limits are ฑ1.0 |ig/cnr for TC and OCR <10 |ig/cm2. ฑ2.0 |ig/cm2 for ECR <10 |ig/cm2. ฑ10% relative percent
difference for TC and OCR >10 |ig/cnr. and ฑ20% relative percent difference for ECR >10 |ig/cm2.
F
1.5
o
O)
=L
1.0
0)
0.5
a>
a)
0.0
J-
n
-0.5
a>
-1 0
TO
O
Q.
-1.5
a>
a:
TC Replicates: 0-10 jjg/cm2; N=69
rt i
V
10 20 30 40 50 60
Sample #with increasing TC (0-10 pg/cm2)
TC Replicates: >10 |jg/cm2; N=1223
20
15
10
5
0
-5
-10
-15
-20
200 400 600 800 1000 1200
Sample #with increasing TC (>10pg/cm2)
Q
O
O
OC Replicates: 0-10jjg/cm2; N=72
OC Replicates: >10jjg/cm2; N=1220
\
15 30 45 60
Sample #with increasing OC (0-10 jjg/cm2)
* 8
O c
a: o
O ai
o fc
o Q
Cm
200 400 600 800 1000
Sam pie #with increasing OC (>10 pg/cm2)
EC Replicates: 0-10 |jg/cm2; N=294
EC Replicates: >10 |jg/cm2; N=998
ST 3
1=
ฃ 2
jnton jfr
JWi
# v
50 100 150 200 250
Sample #with increasing EC (0-10 pg/cm2)
200 400 600 800
Sample #with increasing EC (>10 pg/cm2)
Page 92 of 126
-------
Table 4.3.A-9: Duplicate analysis criteria and statistics for the analysis period 5/23/2018 through 12/17/2018
(samples collected 1/1/2018 through 9/30/2018). Total carbon (TC), organic carbon (OCR), and elemental carbon
(ECR) are shown by reflectance.
Duplicates
Range
Criteria
Statistic
TC
oc
EC
TC, OCR, & ECR
TC, OCR < ฑ1.0 ng C/cm2
Count
4
4
14
<10 |ig C/cm2
ECR < ฑ2.0 |ig C/cm2
No. Fail
0
0
0
%Fail
0
0
0
Units: |ig C/cm2
Mean
0.29
0.23
0.40
StdDev
0.14
0.08
0.42
Max
0.43
0.31
1.25
Min
0.07
0.10
0.01
Median
0.33
0.25
0.18
TC, OCR, & ECR
TC, OCR %RPD < 10%
Count
62
62
52
> 10 |ig C/cm2
ECR %RPD < 20%
No. Fail
0
0
0
%Fail
0
0
0
Units: %
Mean
1.61
1.72
3.70
StdDev
1.15
1.32
3.13
Max
4.36
6.36
12.52
Min
0.01
0.07
0.08
Median
1.24
1.64
2.63
Note: RPD= 100 x absolute value [original sample-replicate sample]/[(original sample+ replicate sample)/2
Page 93 of 126
-------
Figure 4.3.A-7: Duplicate (two punches from the same sample filter analyzed on the same instrument) analysis
results for the analysis period 5/23/2018 through 12/17/2018 (samples collected 1/1/2018 through 9/30/2018). The
limits are ฑ1.0 |ig/cm2 for TC and OCR <10 |ig/cm2. ฑ2.0 |ig/cm2 for ECR <10 |ig/cm2. ฑ10% relative percent
difference for TC and OCR >10 |ig/cnr. and ฑ20% relative percent difference for ECR >10 |ig/cnr.
TC Duplicates: 0-10 jjg/cm2; N=4
TCDuplicate:>10 jjg/cm2; N=62
?T"
1 5
{-
o
CD
1 0
0)
O
05
c
0)
it
0.0
Q
C)
-O.b
-1.0
o
Q.
-1.5
ฃJ
1 2 3
Sample #with increasing TC (0-10 pg/cm2)
OC Duplicates: 0-10 |jg/cm2; N=4
ST
1 5
i-
o
CD
1.0
O
Oh
0)
it
0.0
Q
o
-0 5
O
-1.0
o
Q.
-1.5
ฃJ
2 3
Sample#with increasing OC (0-10 pg/cm2)
20
c
0)
o
15
0)
CL
10
s
6
5
ro
0)
O
n
a:
0)
-5
it
0)
Q
-10
re
o
-15
Q.
-?0
LI
V
ป
V-
Vi
5 10 15 20 25 30 35 40 45 50 55 60 65
Sample #with increasing TC (>10 |jg/cm2)
OC Duplicates: >10 [jg/cm2; N=62
20
15
10
5
0
-5
-10
-15
-20
f.
h'
*
ฆr-
>
0 5 10 15 20 25 30 35 40 45 50 55 60 65
Sample #with increasing OC (>10 |jg/cm2)
EC Duplicates:0-10 |jg/cm2; N=14
EC Duplicates: >10 |jg/cm2; N=52
> g.
40
30
20
10
t 2 0
a: o
O ซ -10
w t
aj Q -20
re
.2 -30
Q.
3 -40
'
M
*1
%
3 4 5 6 7 8 9 10 11 12 13 14
Sample #with increasing EC (0-10 pg/cm2)
10 15 20 25 30 35 40 45 50 55
Sample #with increasing EC (>10 |jg/cm2)
4.3.A.3 Determination of Uncertainties and Method Detection Limits
For discussion of Method Detection Limits (MDLs) see Section 3.1.3.2.
For discussion of analytical uncertainty and total uncertainty see Section 3.1.2 and Section 6.5,
respectively.
4.3.A.4 Audits, Performance Evaluations, Training, and Accreditations
4.3.A. 4.1 System Audits
UC Davis contracted a third-party auditor (Technical & Business Systems; Placerville, CA) to
perform a Laboratory Systems Audit of the DRI Thermal Optical Analysis Laboratory. The audit
was conducted on September 19, 2018. No issues were identified that affected data quality;
Page 94 of 126
-------
auditors provided minor recommendations for improved documentation and tracking, and
assured QA/QC documentation agreed with existing procedures.
4.3.A. 4.2 Performance Evaluations
No performance evaluations were conducted during this reporting period.
4.3.A.4.3 Training
All new laboratory staff receive training in performing the tasks in the SOPs for their assigned
work.
4.3. A. 4.4 A ccreditations
There are no accreditation programs for analysis of carbon on aerosol filters by TOA.
4.3.A.5 Summary of Filter Blanks
Over the sampling period (January 1, 2018 through September 30, 2018) there were 1,253 valid
quartz filter field blanks. Table 4.3.A-10 summarizes the field blank statistics.
Table 4.3.A-10: Quartz filter field blank statistics for the analysis period 5/23/2018 through 12/17/2018 (samples
collected 1/1/2018 through 9/30/2018). Elemental carbon (EC) fractions are indicated as (1) through (3), organic
carbon (OC) fractions are indicated as (1) through (4). Organic pyrolyzed (OP), elemental carbon (EC), and organic
carbon are shown by reflectance (R) and transmittance (T).
Species
( Oil III
Median
(fig/cnr)
A\era tic
(11 li/C 111")
Min
(uii/cnr I
Max
(uii/cnr I
Si. I)c\.
(uii/cnr)
EC1
1253
0.000
0.016
0.000
0.728
0.050
EC2
1253
0.028
0.054
0.000
0.396
0.065
EC3
1253
0.000
0.000
0.000
0.000
0.000
ECR
1253
0.016
0.053
0.000
0.860
0.086
ECT
1253
0.000
0.024
0.000
0.478
0.060
OC1
1253
0.147
0.148
0.000
1.697
0.101
OC2
1253
0.293
0.323
0.028
1.445
0.147
OC3
1253
0.513
0.566
0.151
3.555
0.262
OC4
1253
0.080
0.100
0.000
0.784
0.089
OCR
1253
1.075
1.154
0.244
4.879
0.469
OCT
1253
1.099
1.184
0.244
5.715
0.500
OPR
1253
0.000
0.017
0.000
0.591
0.058
OPT
1253
0.005
0.046
0.000
0.860
0.086
4.3.B UC Davis Thermal Optical Analysis Laboratory
The UC Davis Thermal Optical Analysis Laboratory received and analyzed quartz filters from
batches 48 through 50, covering the sampling period October 1, 2018 through December 31,
2018. Analyses of these samples were performed January 2, 2019 through April 3, 2019. Five
Thermal Optical Carbon Analyzers (Sunset Laboratory Model 5L; designated as Alpha, Beta,
Delta, Gamma and Zeta) were used for analysis during this period using the IMPROVE A
protocol.
Page 95 of 126
-------
Table 4.3.B-1: Sampling dates and corresponding TOA analysis dates covered in this reporting period. Analysis
dates include reanalysis - as requested during QA level 0 and level 1 validation - of any samples within the
sampling year and month.
Sampling Year
Sampling Month
Analysis Bateh #
TOA Analysis Date
2018
October
48
1/2/2019-2/15/2019
2018
November
49
2/7/2019-3/5/2019
2018
December
50
3/4/2019-4/3/2019
2018
All Months
48-50
1/2/2019-4/3/2019
4.3.B.1 Summary of QC Checks and Statistics
Samples are received by the UC Davis Thermal Optical Analysis Laboratory following the
chain-of-custody procedures specified in the UCD CSN TI402A. Samples are analyzed using
Sunset Laboratory Model 5L OCEC analyzers following UCD CSN SOP #402. Daily and weekly
quality control (QC) checks are implemented to ensure data quality. Calibrations of the analyzers
are performed semi-annually or as needed (e.g. when the CHVHe mixture gas cylinder is
replaced). Maintenance is performed as needed by trained laboratory staff. Quality control
procedures are described in UCD CSN SOP #402 and are summarized in Table 4.3.B-2.
Page 96 of 126
-------
Table 4.3.B-2: UC Davis quality control measures for carbon analysis by TOA (Sunset Laboratory OCEC
analyzer).
QA/QC Activity
Frequency
Acceptance Criteria
Corrective Action
Laboratory Blank
Check
Beginning of analysis day
<1.0 ng C/cm2
Repeat analysis. If same result,
check filter lots for possible
contamination and perform pre-
firing
Instrument Blank
Check
Beginning of analysis day
<0.3 |ig C/cm2
Repeat analysis. If same result,
check instrument and gas lines for
possible contamination
Single-point Sucrose
Standard Check
Beginning of analysis day
Within ฑ7% of the calculated value
Repeat analysis. If same result, run
a different sucrose solution to
determine if the problem is with the
solution or instrument. If former,
make new sucrose solution. If latter,
perform full 5-point calibration to
determine new calibration constant.
Calibration Peak
Area Check
Every analysis
Within ฑ10% of the average value
for a specific instrument
Void analysis result; Repeat
analysis with second filter punch
Laser Performance
Check
Beginning of analysis day
Laser Transmittance signal for
Instrument blank >5000
Check laser alignment and/or
examine oven for frosting
Network Sample
Replicates
Every 20th network
sample analyses
ฑ10% when TC >10 |ig /cm2
ฑ20% when ECR >2.5 ng /cm2
or
<ฑ1 |ig/cm2 when TC <10 |ig /cm2
<ฑ0.5 |ig/cm2 when ECR <2.5
Ug/cm2.
Investigate instrument and sample
anomalies. Analyze the third punch
on a difference analyzer
Inter-instrument
Comparison Check
Weekly
Measurement bias for a given
analyzer should be < 10% for TC
and OC and < 20% for ECR.
Investigate instrument and sample
anomalies and rerun replicate when
criterion is not met
Multi-point Sucrose
Standard Check
Every six months or after
major instrument repair or
change of calibration gas
cylinder
NAa
Calculate new calibration constant
based on calibration slope and
update in the parameter file
Temperature
Calibrations
Every six months or after
major instrument repair
NA
Change the temperature offset
values in IMPROVEA.par files
accordingly
a NA: Not Applicable.
4.3.B.2 Summary of QC Results
Detailed results from the carbon quality control checks are presented in the subsections below. In
addition to performing routine daily and weekly QC activities, readings of oven pressure, back
oven and methanator oven temperatures, FID baseline and initial laser are verified to be within
the acceptable range specified for each analyzer before starting sample analysis. After analysis,
each thermogram is reviewed for the following: 1) correct peak identification and integration, 2)
correct laser response, 3) system pressure stability, and 4) to ensure data quality objectives are
met. Individual samples with unusual laser response, baseline shift, low system pressure,
Page 97 of 126
-------
erroneous split point, or samples impacted by failure to meet QC criteria outlined in Table 4.3.B-
2 are reanalyzed.
4.3.B.2.1 Laboratory and Instrument Blanks
At the beginning of the analysis day, following the clean oven procedure, a quartz filter
laboratory blank and an instrument blank are analyzed to check for system contamination and
evaluate laser response. Results are reviewed immediately upon analysis completion and are
compared against the acceptance limits. Table 4.3.B-3 lists the number of blanks analyzed during
the report period and their areal density statistics.
Table 4.3.B-3: Statistics of daily quartz filter laboratory blank and instrument blank analyses on all carbon analyzers
for the analysis period 1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Blank Type
Count
Median
(jig/cm2)
Average
(jig/cm2)
Min
(jig/cm2)
Max
(jig/cm2)
St.Dev.
(jig/cm2)
# Exceedance
Laboratory Blank
283
0.249
0.309
-0.078
1.819
0.255
10
Laboratory Blank - R*
10
0.222
0.251
0.086
0.660
0.171
0
Instrument Blank
283
-0.052
-0.050
-0.239
0.146
0.075
0
*Laboratory Blank - R: repeated laboratory blank when original fails the QC criterion
For laboratory blanks, if the TC areal density exceeds 1.0 |ig C/cm2, a second punch taken from
the same blank filter lot is analyzed (Laboratory Blank - R). If the original and repeated blank
analyses on more than one instruments exceeds the limit, a new lot of quartz blank filters is used
to distinguish filter lot contamination from system contamination. Figure 4.3.B-1 shows the
results of daily laboratory and instrument blank analyzed on all five analyzers during this
reporting period.
Figure 4.3.B-1: Results of daily quartz filter laboratory blanks for the analysis period 1/2/2019 through 4/3/2019
(samples collected 10/1/2018 through 12/31/2018). Red dashed line indicates the acceptance limit of 1.0 |ig C/cm2
for total carbon areal density. For cases when the acceptance limit was exceeded (red points), a repeated analysis
(blue points) was performed.
1.5
N
E
_o
* i.o
C
o
-Q
TO
o
~CC
O 0.5
0.0
Instrument blank analysis is performed following the laboratory blank analysis by reusing the
sample punch. The instrument blank acceptance limit is 0.3 |ig C/cm2 of total carbon. Figure
4.3.B-2 shows the results of daily analyses of instrument blanks on all five analyzers. The TC
ซ :
t .
##
K*
:: v
t
-9 4
jjuป
> s
.*<.
? *!
iV :
*
>
V
2019-01-07 2019-01-21 2019-02-04 2019-02-18 2019-03-04 2019-03-18 2019-04-01
AnalysisDate
Page 98 of 126
-------
values center around 0 |ig C/cm2, with small fluctuations due mostly to the slight baseline shift
during the analysis. There were no instrument blank exceedances during this report period.
Figure 4.3.B-2: Results of daily instrument blanks for the analysis period 1/2/2019 through 4/3/2019 (samples
collected 10/1/2018 through 12/31/2018). Red dash line indicates the acceptance limit of 0.3 |ig C/cm2for total
carbon areal density.
0.50-
CM 0 25 "
E
_o
D>
I 0.00-
ro
o
H -0.25-
-0.50-
2019-01-07 2019-01-21 2019-02-04 2019-02-18 2019-03-04 2019-03-18 2019-04-01
AnalysisDate
4.3.B.2.2 Single-Point Sucrose Standard Check
Following the daily blank analyses, a single-point sucrose calibration check is performed to
evaluate FID response by injecting 10 |iL of sucrose standard solution onto a clean filter punch
and analyzing for total carbon content. Table 4.3.B-4 summarizes the concentrations of the
sucrose standard solutions generated for calibrating the analyzers. They cover a full range of the
TC levels typically seen on the CSN network samples. Sucrose #5 and #15 are chosen for daily
single-point calibration check because their concentrations are most comparable to the CSN
network median TC value.
Table 4.3.B-4: Lookup table of sucrose solution standard concentration in |igC7cnr.
Sucrose ID
Concentration
(U2 C/cm2)
Sucrose 1
41.97
Sucrose|2
25.27
Sucrose|3
36.01
Sucrose 4
5.05
*Sucrose|5
10.10
Sucrose 6
55.00
Sucrose|7
2.00
Sucrose 11
210.50
Sucrose 12
105.25
Sucrose|13
42.10
Sucrose 14
21.05
*Sucrose|15
10.53
Sucrose 16
2.11
* Sucrose #5 and #15 are chosen for the daily single-point sucrose calibration check.
Page 99 of 126
a c* Cv
f 2A
iZ
. :
ซ & C V
. ,ซp ?ซ i T* *5
-------
Upon completion of the sucrose analysis, the measured TC is processed and compared against
the true value (i.e. calculated TC). The % error between the measured and calculated TC is
derived using Equation 4.3.B-1. If the error exceeds the ฑ 7% acceptance limits, a second
analysis is performed before any network samples are analyzed on that instrument. If the second
analysis still exceeds the limit, or if a consistent one-sided bias (with error within ฑ 7%) is
observed on multiple instruments, a different sucrose solution is analyzed to determine if the
problem is with the solution or with the instrument. If the former, a new sucrose solution is made
and verified; if the latter, a full five-point calibration is performed to determine the new
calibration constant for that instrument. Table 4.3.B-5 summarizes the statistics of the daily
sucrose check. There were 25 exceedances out of the 298 sucrose runs during the report period.
All second analyses of the sucrose solution showed acceptable results (Figure 4.3.B-3).
Error (%) =
/Measured TC Calculated TC>
Calculated TC
x 100%
(Eq. 4.3.B-1)
Table 4.3.B-5: Statistics of daily single-point sucrose standard analyses on all carbon analyzers for the analysis
period 1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Count
Median
Error (%)
Average
Error (%)
Min
Error (%)
Max
Error (%)
St.Dev.
Error (%)
# Exceedance
298
0.387
-0.108
-24.938
12.560
4.467
25
Figure 4.3.B-3: Results of daily single-point sucrose calibration standard check for the analysis period 1/2/2019
through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018). Red dashed lines indicate the acceptance limit
of ฑ7% error. For cases when original measured sucrose value (red points) exceeded the acceptance limit, a repeated
analysis was performed (blue points).
3i
V
.1
/
r r
ซ
n
.al
*
* ซ j ^
* -
#
2019-01-07 2019-01-21 2019-02-04 2019-02-18 2019-03-04 2019-03-18 2019-04-01
AnalysisDate
4.3.B.2.3
Calibration Peak Area Check
At the end of each analysis, a fixed amount of methane (CH4) from a cylinder containing 5%
CH4 in Helium is injected into the system as an internal gaseous standard. The CH4 peak area is
quantified and compared to the average peak area of all analyses performed on that instrument.
If the error (calculated using Equation 4.3.B-2) exceeds ฑ 10% acceptance limits, the analysis
result is voided; the flowrate of the calibration gas and sample oven pressure are verified;
corrective actions (if applicable) are taken immediately after the problem is identified; and the
analysis is
Page 100 of 126
-------
repeated using a second filter punch usually after the completion of analysis for that batch
(not on the same day). Table 4.3.B-6 summarizes the statistics of the calibration peak area check.
There were 13 exceedances during this reporting period, most of which occurred when the clamp
that connects the oven ball joint was not sufficiently tightened, resulting in a leak in the system.
Eight exceedances occurred on Alpha over a short period of time due to a worn-out O-ring
(Figure 4.3.B-4). The problem was eliminated after the O-ring was replaced. All reanalyses of
the affected samples had acceptable results.
/Cal. Peak Average Cal. Peak \
Error (%) = ( - ) x 100%
V Average Cal. Peak / (Eq 4 3 B-2)
Table 4.3.B-6: Statistics of internal calibration peak area check on all carbon analyzers for the analysis period
1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Annh/i'i'
( Olllll
Median
llrror (Vii)
A\c ratio
llrror (Vii)
Mill
llrror (Vii)
Max
llrror (Vii)
S(.l)c\.
llrror (Vii)
#
r.xcc'c'daiicc
Alpha
758
0.499
0.000
-39.197
6.827
3.711
8
Beta
759
-0.034
0.000
-99.998
6.680
6.667
2
Delta
802
-0.153
0.000
-15.532
7.795
2.738
1
Gamma
785
-0.360
0.000
-10.766
9.195
3.142
1
Zeta
786
0.087
0.000
-14.361
5.133
2.232
1
Page 101 of 126
-------
Figure 4.3.B-4: Results of internal calibration area check for the analysis period 1/2/2019 through 4/3/2019
(samples collected 10/1/2018 through 12/31/2018). Blue solid line indicates the mean calibration area for the
specific instrument. Red dashed lines indicate the acceptance limit of ฑ10% error from the mean value. For cases
when calibration area exceeded the acceptance limit, a repeated analysis (blue points) was performed and the
original analysis was voided (red points).
alpha
it
1$
ป W M, I
fc
L(
re Kfe - !. '
V 'N*
*aป ฆ r
&
I
2019-01-07 2019-01-21 2019-02-04 2019-02-18 2019-03-04 2019-03-18 2019-04-01
2019-01-07 2019-01-21 2019-02-04 2019-02-18 2019-03-04 2019-03-18 2019-04-01
AnalysisDate
4.3.B.2.3 Laser Performance Check
Laser signals (both reflectance and transmittance) are monitored throughout the TOA analysis
and are examined for stability during post-analysis thermogram review. Any unusual laser
response results in reanalysis of the sample. In addition, before starting the instrument blank
analysis each day, the reading of laser reflectance and transmittance signals is checked to make
sure it is above the initial laser acceptance criterion (i.e. 5000 a.u.). Figure 4.3.B-5 shows the
initial readings of the reflectance and transmittance signals for instrument blank analysis. The
laser signals show good overall stability during this analysis period. There was a step increase in
the laser signal for most of the five analyzers on 3/14/2019 resulting from laser fine-tuning and
signal optimization during a major instrument maintenance. There were no exceedances within
during this analysis period.
Page 102 of 126
-------
Figure 4.3.B-5: Initial laser readings (top: Reflectance; bottom: Transmittance) of the instrumental blank analysis
for the analysis period 1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018). Red dashed
line indicates the acceptance limit of 5000 a.u. of the laser signal. Black vertical line indicates date of instrument
maintenance (3/14/2019). Points are colored by analyzer.
30000-
20000ฆ
TO
O
CD
5=
CD
cr
q3
CO
03
10000-
0-
tua,
SB eป M <
f"X sk ป7 8- as ii
t J
<
20000
-------
Figure 4.3-6: Statistical distribution of the elemental, organic and total carbon areal densities measured from the
weekly performance check samples by each analyzer for the analysis period 1/2/2019 through 4/3/2019 (samples
collected 10/1/2018 tlirough 12/31/2018). Elemental carbon (EC) and organic carbon (OC) are shown by reflectance
(R) and transmittance (T); total carbon is indicated as TC. The thick horizontal line in each box represents the
median value.
ECR
10-
15-
10-
5-
ECT
T
J
alpha beta delta gamma zeta
alpha beta delta gamma zeta
120
E
90
CO
CO
60
c
(1)
Q
"ro
0)
30
<
OCR
120
90-
60-
30-
OCT
alpha
beta
delta gamma zeta
alpha
beta
delta gamma zeta
Name
alpha
S beta
$ delta
^ gamma
zeta
125-
100-
75-
50-
25-
alpha beta delta gamma zeta
Analyzer
The measured carbon areal density from each analyzer (Ax) is compared against the average
value derived from all analyzers available. The bias for each carbon parameter (Biasi) is
calculated for each analyzer each week as:
Ax - averaqe
Bias. = ( ) * 100%
average (Eq 4 33.3)
Page 104 of 126
-------
The acceptance criteria for inter-instrumental bias is 10% for TC and 20% for ECR. Exceeding
the acceptance limits results in further investigation of the instrument and sample anomalies. A
second performance check sample is run on all analyzers once the issue is resolved. Table 4.3.B-
7 summarizes the statistics of the instrument bias for ECR and TC. There was no exceedance
during the report period.
Table 4.3.B-7: Statistics of the instrument bias from the weekly performance check for the analysis period 1/2/2019
through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
r.k'UK'iihil ('iirl)iin l>\ RdU'ikiiui' (l".CK)
Tiihil ( ;irl)iin (TC)
An;il>/.i-r
Cllll 111
Mi'dhin
Awr;iปi-
Min
M;i\
Mi'dhin
Awr;iปi-
Min
M;i\
Alpha
19
4.097
4.325
-5.870
16.074
6.233
-1.697
-1.899
-6.571
2.068
2.751
Beta
19
0.899
1.105
-7.375
16.229
5.893
1.047
1.796
-1.702
7.249
2.741
Delta
19
-6.275
-5.861
-10.723
1.070
3.028
0.353
0.834
-5.294
8.046
3.454
Gamma
19
-1.344
-1.587
-15.927
9.478
6.307
0.007
-0.489
-7.356
3.803
2.566
Zeta
19
2.484
2.019
-7.137
14.511
5.028
-0.296
-0.242
-6.202
2.788
1.982
4.3. B. 2.5 Network Sample Replicates
Replicate analyses are performed on approximately 5% of samples, where replicate analysis
results are obtained from a second punch from the same filter analyzed on a randomly selected
analyzer. Table 4.3.B-8 lists the acceptance criteria for replicate analysis and the summary
statistics from this reporting period. A total of 194 replicate analyses were performed out of the
3,851 samples. For cases that exceeded the acceptance limits, a third punch (if available) was
analyzed on a different analyzer. All three sets of results (routine, replicate, and reanalysis) from
the same sample are compared to determine analysis validity. Instrument anomaly and/or sample
inhomogeneity are also examined. Figure 4.3.B-7 shows the results of the replicate analysis.
There were 12 TC exceedances and seven ECR exceedances during this reporting period.
Affected samples were reanalyzed on a third analyzer. Three samples (F128113, F132000, and
F129430) failed the replicate analysis criteria for TC and were not reanalyzed because there was
insufficient deposit area remaining (see Figure 4.3.B-7, panel d). All other reanalyses had
satisfactory results.
Table 4.3.B-8: Acceptance criteria and the summary statistics of the replicate analyses for the analysis period
1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Parameter
Acceptance Criteria
#
Replicate
#
Exceedance
# Reanalysis
passed
TC
*RPD < ฑ10% when TC >10 ng /cm2
or
Absolute difference <ฑ1 |ig/cnr when TC <10 |ig /cm2
194
12
9
ECR
*RPD < ฑ20% when ECR >2.5 |ig /cm2
or
Absolute difference <ฑ0.5 |ig/cnr when ECR <2.5
Ug/cm2
194
7
7
*RPD: Relative Percentage Difference = (Replicate-Routine)/Average *100%
Page 105 of 126
-------
Figure 4.3.B-7: Results of CSN replicate analysis for ECR (Panel a and b) and TC (Panel c and d) for the analysis
period 1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018). The red dashed lines in each
panel represents the acceptance limits.
0.50-
" 0.25-
O)
=L
b
o
(a)
'
. #
0.0 0.5 1.0 1.5
Average Areal Density, n g/crrr
2.0
2
0.0 2.5 5.0 7.5
Average Areal Density, j.i g/cm'
(C)
>e 20-
0 10
ฃ-10
en
o
111 -20
2.5
20-
o 10
c
0)
ฃ
b o
0
J5
(U
Of -10--
o
-20-
.(b)
% .
10
20
30
40
Average Areal Density, ^ g/cm
(d)
* * ป
r/ r*
10.0
2
25
50
75
100
Average Areal Density, g/cm
4.3.B.2.6 Multi-point Sucrose Standard Check
A multi-point calibration is performed every six months, when the calibration gas cylinder is
replaced, or if a consistent one-sided bias is observed with the daily single-point sucrose standard
check, whichever comes first. The calibration uses sucrose standards at five different
concentration levels (see Table 4.3.B-4 for details). The least-square correlation coefficient (r2)
of measured versus calculated mass of carbon, force-fit through the origin (0,0), should be higher
than 0.995. The calibration constant for each analyzer is updated if the measured and calculated
sucrose concentrations deviate from the 1:1 line by more than 1% (i.e. calibration slope > 1.01 or
< 0.99). Table 4.3.B-9 summarizes the multi-point sucrose calibrations performed during this
reporting period.
Table 4.3. B-9: Summary of multi-point sucrose standard checks performed for the analysis period 1/2/2019 through
4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Analyzer
Calibration Date
Slope
r2
Calibration Constant
Beta
2/5/2019
1.0009
0.9999
20.8463
Alpha
2/5/2019
0.9634
0.9999
20.6972
Gamma
2/5/2019
0.9993
0.9999
20.3398
Delta
2/5/2019
1.0125
0.9988
20.0899
Zeta
2/5/2019
0.9694
0.9993
20.7421
Page 106 of 126
-------
4.3.B.2.7 Temperature Calibration
A temperature calibration is performed every six months (usually along with a multi-point
sucrose calibration) or after a major instrument repair (e.g., replacement of main oven or heating
coils). The sample temperature probe is calibrated using a manufacturer-provided temperature
calibration device, inserted into the sample oven so that the external temperature probe sits at
where a filter punch would be during a routine analysis. The oven temperature cycles through the
IMPROVEA protocol temperature set points (from 140ฐC to 840ฐC). The differences in
temperature readings by the calibration probe and the sample temperature probe (i.e. temperature
offsets) are calculated and updated in the instrument parameter file. The system then goes
through the IMPROVE A temperature cycle again to verify that the temperature readings from
the two probes are within 10ฐC at all temperature steps. Table 4.3.B-10 summarizes the
temperature calibrations performed on each analyzer during this reporting period. Differences in
the temperature offsets between the two calibrations are expected if the heating coils are replaced
or re-wrapped around the main oven.
Table 4.3.B-10: Summary of the temperature calibrations performed on each analyzer for the analysis period
1/2/2019 through 4/3/2019 (samples collected 10/1/2018 through 12/31/2018).
Analyzer
Calibration
Date
Oven Re-
NYrapped?
Temperature Offsets (ฐC)
140ฐC
280ฐC
480ฐC
580ฐC
740ฐC
840ฐC
Beta
2/1/2019
N
-12
-27
-33
-32
4
-3
3/14/2019
Y
8
7
-21
-55
-4
-16
Alpha
2/1/2019
N
16
31
30
28
-5
-18
3/14/2019
Y
12
31
30
28
-5
-18
Gamma
1/31/2019
N
8
22
32
30
30
18
3/14/2019
Y
-24
-53
-50
-65
-37
-40
Delta
1/31/2019
N
30
36
36
29
32
19
3/14/2019
Y
0
-15
-21
-21
-5
-11
Zeta
1/31/2019
N
11
23
31
30
39
18
3/14/2019
Y
-34
-63
-71
-60
-8
-20
4.3.B.3 Determination of Uncertainties and Method Detection Limits
For determination of Method Detection Limits (MDLs) see Section 3.1.3.2.
For uncertainty estimates see Section 6.5.
4.3.B.4 Audits, Performance Evaluations, Training, and Accreditations
4.3.B. 4.1 System Audits
The EPA did not conduct any audits or performance evaluations of the UC Davis Carbon
Laboratory during this reporting period.
4.3.B.4.2 Performance Evaluations
The UC Davis Thermal Optical Analysis Laboratory participated in an interlaboratory
comparison study organized by the European Centre for Aerosol Calibration (ECAC) in March
2019. Eight quartz filter samples and one solution sample were received and analyzed for OC,
Page 107 of 126
-------
EC and TC. UC Davis passed the evaluation with good data repeatability and no systematic bias.
The full report is available at https://www.actris-ecac.eu/files/OCEC-2Q19-l-REPORT final.pdf.
4.3. B. 4.3 Training
All new laboratory staff and student assistants working in the UC Davis Thermal Optical
Analysis Laboratory receive mandatory UC Laboratory Safety Fundamentals training. Personnel
who operate the TOA analyzers receive additional training on the CSN SOP 402 and relevant
Technical Instructions.
4.3. B.4.4 A ccreditations
There are no accreditations for analysis of carbon on aerosol filters by TOA.
4.3.B.5 Summary of Filter Blanks
Over the sampling period (October 1, 2018 through December 31, 2018) there were 423 valid
quartz filter field blanks. Table 4.3.B-11 summarizes the field blank statistics.
Table 4.3.B-11: Quartz filter field blank statistics for the analysis period 1/2/2019 through 4/3/2019 (samples
collected 10/1/2018 through 12/31/2018). Elemental carbon (EC) fractions are indicated as (1) through (3), organic
carbon (OC) fractions are indicated as (1) through (4). Organic pyrolyzed (OP), elemental carbon (EC), and organic
carbon (OC) are shown by reflectance (R) and transmittance (T).
Species
( Oil III
Median
(.11 li/CIll")
A\era tic
Uiii/cnr I
Min
uiii/cnr)
Max
(fiii/cnr)
S(.l)c\.
(iiii/cnr)
EC1
423
0.025
0.065
-0.050
0.989
0.121
EC2
423
0.065
0.095
-0.010
0.675
0.089
EC3
423
0.013
0.017
-0.025
0.287
0.028
ECR
423
0.000
0.000
-0.002
0.098
0.007
ECT
423
0.000
0.000
-0.002
0.001
0.000
OC1
423
0.225
0.230
0.017
0.689
0.096
OC2
423
0.310
0.343
0.122
1.949
0.176
OC3
423
0.367
0.562
0.121
4.672
0.529
OC4
423
0.140
0.261
0.022
1.580
0.273
OCR
423
1.250
1.572
0.455
7.078
0.963
OCT
423
1.250
1.573
0.455
7.078
0.963
OPR
423
0.107
0.176
-0.080
1.593
0.206
OPT
423
0.107
0.177
-0.080
1.593
0.206
5. Data Management and Reporting
5.1 Number of Events Posted to AQS
Table 5.1-1 summarizes dates that data were delivered to AQS for samples collected January 1,
2018 through December 31, 2018. Data are expected to be delivered to AQS within 120 days of
receipt of filters by the analytical laboratories. Laboratory analysis delays resulted in later
deliveries to AQS (see Section 2.1.1, Section 2.2.1, Section 2.4.1, and Section 2.5.1).
Page 108 of 126
-------
Table 5.1-1: Summary of data deliveries to AQS, January 1, 2018 through December 31, 2018.
Data (Month Samples Collected)
Filter Receipt Date
AQS Delivery Date
Days
January 2018
March 7, 2018
August 31, 2018
177
February 2018
April 4,2018
September 14, 2018
163
March 2018
May 2, 2018
September 28, 2018
149
April 2018
June 7, 2018
October 16, 2018
131
May 2018
July 10. 2018
November 3, 2018
115
June 2018
August 8, 2018
December 4, 2018
118
July 2018
September 6, 2018
January 3, 2019
119
August 2018
October 9, 2018
January 30, 2019
113
September 2018
November 7, 2018
February 28, 2019
113
October 2018
December 11, 2018
April 22, 2019
132
November 2018
January 9, 2019
May 9, 2019
120
December 2018
February 6, 2019
June 11,2019
125
6. Quality Assurance and Data Validation
6.1 QAPP Revisions
The UC Davis Quality Assurance Project Plan (QAPP) for Laboratory Analysis and Data
Processing/Validation for Chemical Speciation of PM2.5 Filter Samples was accepted by the EPA
on November 29, 2017. Updated versions were delivered to the EPA on November 30, 2018 and
July 31, 2019.
6.2 SOP Revisions
The UC Davis Standard Operating Procedures (SOPs) for Laboratory Analysis and Data
Processing/Validation for Chemical Speciation of PM2.5 Filter Samples were accepted by the
EPA on November 29, 2017. Updated versions were delivered to the EPA on November 30,
2018 and July 31, 2019.
6.3 Summary of Internal QA Activities
Following laboratory analysis all analytical results are assembled by UC Davis for processing
and initial validation. Data processing involves calculating ambient concentration, uncertainty,
and MDL for each analyte using the laboratory result plus the sample volume determined from
the field data. The calculated concentrations undergo two levels of validation at UC Davis: (1)
Level 0 validation to examine the fundamental information associated with each measured
variable, such as chain of custody, shipping integrity, sample identification, and damaged
samples, and (2) Level 1 review for technical acceptability and reasonableness based on
information such as routine QC sample results, data quality indicator calculations, performance
evaluation samples, internal and external audits, statistical screening, internal consistency
checks, and range checks. Further detail regarding the UC Davis data processing and validation
Page 109 of 126
-------
can be found in UCD CSN SOP #801: Processing and Validating Raw Data, and in the
associated Technical Information (TI) documents as follows:
1) UCD CSN TI 801 A, Data Ingest. Sample event information (including filter IDs, flow
rates, flags, and comments) are received from the Sample Handling Lab (Wood PLC)
via email and uploaded to the UC Davis CSN database. UC Davis EDXRF results are
transferred into the UC Davis CSN database through an automated service. For
samples collected through September 30, 2018, IC and TOA analysis result files were
received via email from DRI. Beginning October 1, 2018, IC analysis result files are
received via email from RTI; and, UC Davis TOA results are transferred into the UC
Davis CSN database through an automated service. Result files received via email
(DRI and RTI) are ingested to the UC Davis CSN database.
2) UCD CSN TI 801C, Level 0 Validation: Data and metadata are reviewed through
several visualizations to identify oddities such as inconsistent dates that appear to be
data transcription and/or data entry errors. These are resolved through communication
with the Sample Handling Lab.
3) UCD CSN TI 80IB, Data Processing: Sample volume and analysis results are
combined to calculate concentrations. Blank values are used to derive MDLs. MDLs
and concentrations are used to estimate uncertainty.
4) UCD CSN TI 801C, Level 1 Data Validation: Several statistical and visual checks are
applied and examined. Reanalyses are requested as needed. Data are flagged with
qualifier or null codes.
5) UCD CSN TI 80ID, Data Posting: Initially validated concentration data and metadata
are posted to DART for SLT (State, Local, and Tribal) review. After the specified 30-
day review period, changed or unchanged data are re-ingested to the UC Davis CSN
database.
6) UCD CSN TI 80ID, AQS Delivery : SLT initiated changes and comments are
reviewed and resolved. Data are formatted for delivery to AQS and posted.
6.4 Data Validation and Review
The validation graphics shown in this section are a small subset of the many QC evaluations that
UC Davis performs on a routine basis. They are selected to illustrate the nature and use of the
QC tools, and provide an overview of the review process.
Additional information and detail regarding analytical and validation procedures can be found in
the standard operation procedure (SOP) documents, UC Davis CSN Quality Assurance Project
Plan (QAPP), and the Data Validation for the Chemical Speciation Network guide, all available
at the UC Davis CSN site: https://aqrc.ucdavis.edu/csn-documentation.
6.4.1 Summary of Monthly Data Validation Review Results
6.4.1.1 Comparisons Across Years
Multi-year time series plots are used to examine large-scale trends and/or analytical problems.
Comparisons to historical network data provide context for validation and review of more recent
data.
Page 110 of 126
-------
Figures 6.4-1 and 6.4-2 show time series for the network-wide 90!h percentile, median (50th
percentile), and 10th percentile concentrations of organic carbon by reflectance (OCR) and
elemental carbon by reflectance (ECR). These figures show raw data without blank correction to
enable comparison across a wider timeframe. The carbon fractions OCR and ECR are
determined by thermal optical analysis (TOA) with a correction for pyrolysis based on optical
monitoring as the sample is heated. Measurements for samples collected from 2005 through 2015
were made at DRI with DRI Model 2001 analyzers monitoring at the single wavelength 633 nm;
samples collected from January 2016 through September 2018 were analyzed at DRI using their
DRI Model 2015 analyzers monitoring seven wavelengths centered at 635 nm; and, beginning
with samples collected from October 2018 analysis was performed at UC Davis using the Sunset
Laboratory Model 5L analyzer monitoring at the single wavelength 658 nm. The OCR 2018
concentrations at the median and 90th percentile were elevated during July and August, but
otherwise trended lower than previous years for most months, particularly at the 90'1' percentile.
The ECR concentrations for 2018 trend similarly to previous years with modestly elevated
concentrations during the summer months.
Figure 6.4-1: Multi-year time series of network-wide organic carbon by reflectance concentrations (OCR; raw data
without blank correction). Symbols denote laboratory and type of analyzer: DRI Model 2001 (circle), DRI Model
2015 (triangle), and UCD Sunset Laboratory Model 5L (square).
90th Percentile
Year
Median
2010
2011
2012
2013
2014
2015
2016
2017
2018
Year
E 2.5
O)
2.0
0ฃ
O
O
1.5
1.0
Jan
10th Percentile
Feb
Mar
Apr May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
2010
2011
2012
2013
2014
2015
2016
2017
2018
Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
Analyzer
DRI Model 2001
DRI Model 2015
UCD Sunset Laboratory Model 5L
Page 111 of 126
-------
Figure 6.4-2: Multi-year time series of network-wide elemental carbon by reflectance concentrations (ECR; raw
data without blank correction). Symbols denote laboratory and type of analyzer: DRI Model 2001 (circle), DRI
Model 2015 (triangle), and UCD Sunset Laboratory Model 5L (square).
2017
2018
Year
2010
2011
2012
2013
2014
2015
*- 2016
90th Percentile
O)
a
tr
o
LU
Median
0.8
0.7
0.6
0.5
0.4
0.3
Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
10th Percentile
/-v 0.25
CO
0.20
CT>
3 0.15
cc
Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
Analyzer DRI Model 2001 A DRI Model 2015 ฆ UCD Sunset Laboratory Model 5L
During TOA analysis some of the OC pyrolyzes upon heating under the inert environment. The
organic pyrolyzed carbon (OPR) is combusted with the EC collected on the filter, and is
accounted for by monitoring the laser signal and i dentifying an OC/EC split point based on
return of the last signal to its initial value. To some extent, the split point - and thus the amount
of OPR - is operationally defined based on instrument parameter settings. As seen in Figure 6.4-
3, corresponding with the change in analyzers from DRI Model 2001 to DRI Model 2015 that
occurred on January 1, 2016, the OPR concentrations at the median and 90th percentile
decreased; and, corresponding with the laboratory transition from DRI (DRI Model 2015) to UC
Davis (UCD Sunset Laboratory Model 5L) the OPR concentrations at the 10th percentile,
median, and 90th percentile increased. The October, November, and December 2018 OPR
concentrations at the median and 90 percentile are in closer alignment with those from DRI
made prior to the instrument transition (DRI Model 2001 to DRI Model 2015).
Page 112 of 126
-------
Figure 6.4-3: Multi-year time series of network-wide organic pyrolyzed carbon by reflectance concentrations (OPR;
raw data without blank correction). Symbols denote laboratory and type of analyzer: DRI Model 2001 (circle), DRI
Model 2015 (triangle), and UCD Sunset Laboratory Model 5L (square).
90th Percentile
Jan Feb Mar
Median
Apr May
Aug Sep
Year
10th Percentile
0.10
0.05
0.00
Analyzer DRI Model 2001 A DRI Model 2015
UCD Sunset Laboratory Model 5L
Page 113 of 126
-------
Similar to 2016 and 2017, the 2018 sulfur concentrations generally continue to be low (Figure
6.4-4), with reduced seasonal variability.
Figure 6.4-4: Multi-year time series of network-wide sulfur (S) concentrations.
90th Percentile Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
Median Year
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2018
2010
2011
2012
2013
2014
2015
2016
2017
10th Percentile Year
0.25-
cT
ฃ 0.20-
ง 0.15-
2010
2011
2012
2013
2014
2015
2016
2017
2018
Page 114 of 126
-------
The nitrate concentrations show strong seasonality with elevated winter concentrations; however,
2016, 2017, and 2018 concentrations are generally lower relative to previous years (Figure 6.4-
5).
Figure 6.4-5: Multi-year time series of network-wide nitrate concentrations.
90th Percentile Year
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2018
10th Percentile Year
2010
2011
2012
2013
2014
2015
2016
2017
2018
Year
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Median
2010
2011
2012
2013
2014
-* 2015
-~ 2016
2017
ฃ 0.5
^ฐ-4
=S-
^ 0.3H
-------
Figure 6.4-6: Scatter plot of (3 xS) versus SO4, samples collected January 1, 2018 through December 31, 2018.
Number of observations (complete pairs) is 12,826. Dotted black horizontal and vertical lines indicate MDLs. Solid
gray line indicates 1:1. Solid red line indicates regression.
O
0
ฉ
0
0 o 0 0 0
/ฃ--
O
O ^
0
SiiP ฐ0
0
y = 1 Olx + 0.08
r2 = 0.9
SoaomKm
ฐ ซ
*:
0.0
2.5
5.0
7.5
Sulfate (ng/m3)
10.0
Capitol
Deer Park
Jefferson Elementary (10th and Vine)
Louisville - Cannon's Lane
Rutgers (collocated)
South Dekalb
Potassium versus Potassium Ion
PTFE filters are analyzed for elemental potassium using EDXRF, and nylon filters are analyzed
for potassium ion using IC. Similar to the S/SO4 ratio relationship, the potassium/potassium ion
ratio can be used to identify outliers as well as atmospherically unusual events. In a scenario
where all the particulate potassium is present as water-soluble potassium ion, the
potassium/potassium ion ratio is expected to be near one. This expectation is generally met, with
greater variability at low concentrations (Figure 6.4-7). A known exception to this expectation is
for soil-borne potassium, which is not water soluble; high soil contributions are thus expected to
result in ratios greater than one.
Page 116 of 126
-------
Figure 6.4-7: Scatter plot of potassium versus potassium ion, samples collected January 1, 2018 through December
31, 2018. Number of observations (complete pairs) is 12,826. Dotted black horizontal and vertical lines indicate
MDLs. Solid gray line indicates 1:1. Solid red line indicates regression.
Potassium Ion (|.ig/m3)
PM2.5 versus Reconstructed Mass (RCM)
Gravimetric data are compared to RCM, where the RCM composite variable is estimated from
chemical speciation measurements, to test many different aspects of overall data quality. The
formulas used to estimate the mass contributions from various chemical species are detailed in
UCD CSN TI801B. In the simple case where valid measurements are available for all needed
variables, reconstructed mass is the following sum:
RCM = (4.125 x S) + (1.29 x NO3" ) + (1.4 x OC) + (EC) +
(2.2 x A1 + 2.49 x Si + 1.63 x Ca + 2.42 x Fe + 1.94 x Ti) + (1.8 x chloride)
The parenthesized components represent the mass contributions from, in order, ammonium
sulfate, ammonium nitrate, organic compounds, elemental carbon, soil, and sea salt.
Gravimetric analysis is not routinely performed using CSN filters. Thus, for comparison
purposes 24-hour average gravimetric PM2.5 mass data from AirNow Tech is used as part of the
validation process in DART. The data provided by AirNow Tech is not final, so the data used
here is a snapshot, downloaded at the time the plots were generated.
If the RCM completely captures and accurately estimates the different mass components, the
RCM to AirNow Tech mass ratio is expected to be near one. The RCM and AirNow Tech mass
generally correlate (Figure 6.4-8), but RCM tends to underestimate AirNow Tech mass.
Page 117 of 126
-------
Figure 6.4-8: Scatter plot of RCM versus AirNow Tech PM2.5 mass data (Mass), samples collected January 1, 2018
through December 31, 2018. Number of observations (complete pairs) is 9,630. Solid gray line indicates 1:1. Solid
red line indicates regression.
Mass (|.ig/m3)
Page 118 of 126
-------
6.5 Uncertainty Estimates and Collocated Precision Summary Statistics
Several network sites are equipped with collocated samplers, where simultaneous samples are
collected on independent samplers and analyzed using the same analytical protocols. Differences
between the resulting data provide a measure of the total uncertainty associated with filter
substrates, sampling and handling in the field, and laboratory analysis.
Scaled relative difference between sample pairs collected at CSN collocated sites is calculated as
shown in Equation 6.5-1 and used to evaluate collocated precision (Figure 6.5.1, elements;
Figure 6.5-2, ions; Figure 6.5-3, carbon).
(collocated routine) / V2
Scaled Relative Difference = -7 : :,
(collocated + routine) / 2 (Eq 6 5-1)
The scaled relative differences are ฑV2 when one of the two measurements is zero, and vary
between these limits at concentrations close to the detection limit. They generally decrease with
increasing concentration, and are expected to converge to a distribution representative of
multiplicative measurement error when the concentration is well above the detection limit. This
convergence is not observed for many elements and carbon fractions that are rarely measured
above the MDL.
Page 119 of 126
-------
Figure 6.5-1: Scaled relative difference for element measurements at sites with collocated samplers across the
network (January 1, 2018 through December 31, 2018). Dotted vertical lines indicate MDL.
ฐ Bakersfield - California Ave. (06-029-0014) o G.T. Craig (39-035-0060)
Site ฐ Deer Park (48-201-1039) o Riverside - Rubidoux (06-065-8001)
o Dudley Square - Roxbury (Boston) (25-025-0042) o Rutgers (34-023-0011)
Page 120 of 126
-------
Figure 6.5-2: Scaled relative difference for ion measurements at sites with collocated samplers across the network
(January 1, 2018 through December 31. 2018). Dotted vertical lines indicate MDL.
Ammonium
o o CD
O
<9 O 8
30%o^ฐoฐ
cd "dQQSffli
o o J&vrrฎ
8o0^
oฐ
o
o
ฐ ฐ o
o o
Potassium Ion
o ^ :
Chloride
: o o0
o
c
o ฐ ฐ
gฃฐ jฐ 0
p
Nitrate
o o Q
ฐฐ.ฐ
^oo
o * o
o
o
o
Sodium Ion
ซ&ฆ<*,.
o
o
o *
Jh&o co
o <%
8 o : o
o ; ฐ
Sulfate
oo o
o
o
o
o
o
o
o
0.01
100
0.01 1 100
Average Concentration/MDL
0.01
100
o Bakersfield - California Ave.
Site Name 0 Deer Park
ฐ G.T. Craig
o Riverside - Rubidoux
o Dudley Square - Roxbury (Boston) o Rutgers
Page 121 of 126
-------
Figure 6.5-3: Scaled relative difference for carbon measurements at sites with collocated samplers across the
network (January 1, 2018 through December 31, 2018). Dotted vertical lines indicate MDL. Elemental carbon (EC)
fractions are indicated as (1) through (3), organic carbon (OC) fractions are indicated as (1) through (4). Organic
pyrolized (OP), elemental carbon (EC), and organic carbon (OC) are shown by reflectance (R) and transmittance
(T).
EC1
4
o
ฐ
a
Qb
o
ECT
o
ฐ %
o
OC4
o
#ฆ
o
OPT
c
o
D@GD<ง> O
O
000ฐ ฐ
o
00
0.1 1 10 100
EC2
OC1
OCR
ฉ O (^DGHDฎ O O
0.1
10 100
OC2
O
o
OCT
#
ฐo
o
ECR
OC3
o
o
OPR
c
m c
0
*
o
<3D
-------
Collocated precision is reported with CSN data delivered to AQS as fractional uncertainty.
Fractional uncertainty is calculated from scaled relative differences (Equation 6.5-1) between
sample pairs collected at CSN collocated sites, using the subset of observations with
concentrations at least three times the MDL. To limit uncertainty in determination of the
necessary percentiles, calculations are performed using multiple years of collocated data
(January 1, 2009 through December 31, 2014 for this reporting period) with a minimum of 60
collocated pairs per year. The calculation for fractional uncertainty is documented in UCD CSN
TI801B, and summarized in Equation 6.5-1, Equation 6.5-2, and Equation 6.5-3.
(B4th percentile of SRD) (16th percentile of SRD)
Collocated Precision (cp) = : -
(Eq. 6.5-2)
Fractional Uncertainty = 100 x
(Eq. 6.5-3)
Tables 6.5-1 (elements), 6.5-2 (ions), and 6.5-3 (carbon) list fractional uncertainties calculated
for this reporting period. Since many species are routinely measured at or below the MDL, there
are numerous instances where insufficient pairs were available, in which cases a fractional
uncertainty of 0.25 is assigned. Historical data (2009-2014) are used to calculate fractional
uncertainties for this reporting period. Beginning with the next reporting period (samples
collected January 1, 2019 through December 31, 2019), fractional uncertainties will be updated
annually and calculated using collocated data from the previous two years.
The network measurement quality objectives (MQOs) are based on the coefficient of variation
(CV) between collocated measurements, and are defined as CV of 10% for ions, 20% for
elements, and 15% for total carbon. As shown in Equation 6.5-4 and Equation 6.5-5, CV is
calculated from sample pairs collected at CSN collocated sites, using the subset of observations
with concentrations at least three times the MDL.
X, - Yj
Relative Percent Difference (RPD) = x 100
{Xi + YJ / I (ฃq 6 5_4)
| RPD |
rv = 1
V2
(Eq. 6.5-5)
where X and Y, are the measurements from routine and collocated sites, respectively, for the ith
pair of measurements. Tables 6.5-1 (elements), 6.5-2 (ions), and 6.5-3 (carbon) list CV
calculated from collocated samples collected during 2018.
Page 123 of 126
-------
Table 6.5-1: Fractional uncertainty (calculated from collocated samples collected 2009 through 2014) and median coefficient
of variation (CV; calculated from samples collected during 2018) for elemental species. Fractional uncertainty and CV values
not reported for species with less than 60 collocated pairs.
Species
Fractional Uncertainty (%)
2009-2014
Pairs
Coefficient of Variation (%)
2018
Pairs
Na
16.4
1,270
38
Mg
24.5
365
5
A1
25.2
1,209
55
Si
15.2
3,897
4.2
186
P
17.3
93
4
S
6.2
5,530
1.9
338
CI
34.2
1,740
11.9
98
K
10.6
4,825
2.6
321
Ca
16.8
4,067
4.3
141
Ti
17.4
697
55
V
12.8
499
0
Cr
38.9
83
1
Mn
15.4
623
11
Fe
17
5,520
4.4
184
Co
10
0
Ni
17.8
400
0
Cu
26.9
2,313
4
Zn
12.3
3,144
3.5
123
As
18.8
155
0
Se
43
0
Br
15
1,610
0
Rb
0
0
Sr
58
0
Zr
3
0
Ag
1
0
Cd
0
0
In
0
0
Sn
0
0
Sb
0
0
Cs
7
0
Ba
16.5
123
0
Ce
21
0
Pb
18.5
381
0
Page 124 of 126
-------
Table 6.5-2: Fractional uncertainty (calculated from collocated samples collected 2009 through 2014) and median coefficient
of variation (CV; calculated from samples collected during 2018) for ions. Fractional uncertainty and CV values not reported
for species with less than 60 collocated pairs.
Species
Fractional Uncertainty (%)
2009-2014
Pairs
Coefficient of Variation (%)
2018
Pairs
Ammonium
7.1
5,466
4.6
316
Chloride*
3.8
203
Nitrate
7.6
5,767
2.7
320
Potassium Ion
12.6
2,072
8
Sodium Ion
24.7
3,562
4.0
206
Sulfate
4.9
5,680
1.9
335
*Collocated chloride results were not available/reported until February 2017.
Table 6.5-3: Fractional uncertainty (calculated from collocated samples collected 2009 through 2014) and median coefficient
of variation (CV; calculated from samples collected during 2018) for carbon fractions. Fractional uncertainty and CV values
not reported for species with less than 60 collocated pairs. Elemental carbon (EC) fractions are indicated as (1) through (3),
organic carbon (OC) fractions are indicated as (1) through (4). Organic pyrolyzed (OP), elemental carbon (EC), and organic
carbon (OC) are shown by reflectance (R) and transmittance (T).
Species
Fractional Uncertainty (%)
2009-2014
Pairs
Coefficient of Variation (%)
2018
Pairs
Elemental Carbon (EC1)
12.9
1,948
4.3
337
Elemental Carbon (EC2)
36.8
992
9.0
235
Elemental Carbon (EC3)
4
12
Elemental Carbon (ECR)
15.5
1955
4.2
335
Elemental Carbon (ECT)
12.8
1,606
4.9
335
Organic Carbon (OC1)
32.9
1,039
10.4
219
Organic Carbon (OC2)
13.6
1,877
3.8
328
Organic Carbon (OC3)
17.8
1,860
4.4
286
Organic Carbon (OC4)
15.7
1,487
6.6
262
Organic Carbon (OCR)
11.6
2,033
2.9
315
Organic Carbon (OCT)
7.3
1,774
2.5
316
Organic Pyrolyzed (OPR)
25.1
919
8.4
197
Organic Pyrolyzed (OPT)
17.3
1,557
6.8
298
7. References
Chen, L.-W.A.; Chow, J.C.; Wang, X.L.; Robles, J.A.; Sumlin, B.J.; Lowenthal, D.H.; Watson,
J.G. (2015). Multi-wavelength optical measurement to enhance thermal/optical analysis for
carbonaceous aerosol. Atmos. Meas. Tech., 8:451-461. http://www.atmos-meas-
tech.net/8/451/2015/amt-8-451-2015.html.
Chen, L.-W.A.; Chow, J.C.; Watson, J.G.; Schichtel, B.A. (2012). Consistency of long-term
elemental carbon trends from thermal and optical measurements in the IMPROVE network.
Atmos. Meas. Tech., 5:2329-2338. http://www.atmos-meas-tech.net/5/2329/2012/amt-5-2329-
2012.pdf.
Chow, J.C., Watson, J.G. (2017). "Enhanced ion chromatographic speciation of water-soluble
PM2.5 to improve aerosol source apportionment." Aerosol Science and Engineering 1:7-24.
Page 125 of 126
-------
Chow, J.C.; Wang, X.L.; Sumlin, B.J.; Gronstal, S.B.; Chen, L.-W.A.; Trimble, D.L.; Kohl,
S.D.; Mayorga, S.R.; Riggio, G.M.; Hurbain, P.R.; Johnson, M.; Zimmermann, R.; Watson, J.G.
(2015). Optical calibration and equivalence of a multiwavelength thermal/optical carbon
analyzer. AAQR, 15(4): 1145-1159. doi:10.4209/aaqr.2015.02.0106.
http://aaqr.org/ArticlesInPress/AAQR-15-02-OA-0106_proof.pdf.
Chow, J.C.; Watson, J.G.; Robles, J.; Wang, X.L.; Chen, L.-W.A.; Trimble, D.L.; Kohl, S.D.;
Tropp, R.J.; Fung, K.K. (2011). Quality assurance and quality control for thermal/optical
analysis of aerosol samples for organic and elemental carbon. Anal. Bioanal. Chem.,
401 (10):3141 -3152. DOI 10.1007/s00216-011-5103-3.
Chow, J.C.; Watson, J.G.; Chen, L.W.; Chang, M.C..; Robinson, N.F..; Dana Trimble; Steven
Kohl. (2007). The IMPROVEA Temperature Protocol for Thermal/Optical Carbon Analysis:
Maintaining Consistency with a Long-Term Database. J. Air Waste Manage. Assoc., 57:1014-
1023.
EPA 40 CFRPart 58. Available at http://origin.www.gpo.gov/fdsys/pkg/FR-2010-12-
27/pdf/2010-32153.pdf#page=l
Yatkin, S., Amin, H. S., Trzepla, K., Dillner, A.M. (2016a). Preparation of Lead (Pb) X-ray
Fluorescence Reference Materials for the EPA Pb Monitoring Program and the IMPROVE
Network Using an Aerosol Deposition Method. Aerosol Sci. Technol. 50:309-320.
Yatkin, S., Belis, C.A., Gerboles, M., Calzolai, G., Lucarelli, F., Fabrizia, C., Trzepla, K. (2016b). An
Interlab oratory Comparison Study on the Measurement of Elements in PMio. Atmos. Environ. 125: 61-
68.
Page 126 of 126
------- |