Summary Report: Improving
characterization of reduced nitrogen at
IMPROVE and CSN monitoring sites

Christopher Rogers, Kevin Mishoe, Marcus Stewart, Katherine Barry
Wood Environment & Infrastructure Solutions, Inc.

Joann Rice, Xi (Doris) Chen
U.S. EPA, OAR, OAQPS, AQAD

John T. Walker
U.S. EPA, ORD, CEMM, AESMD

Melissa Puchalski
U.S. EPA, OAR, OAP, CAMD

Ralph Baumgardner
U.S. EPA, ORD, CPHEA, PHESD

Bret Schichtel
NPS, ARD

October 7, 2020


-------
Table of Contents

I.0	Introduction	1

1.1	Overview	1

1.2	Discussion	1

1.3	Study Objectives	3

2.0 Measurement Locations	4

3.0 Sampling Methods & Procedures	8

4.0 Laboratory Procedures	10

4.1	Preparation of Sampling Media	10

4.2	Chemical Analysis	10

4.3	Field and Laboratory Blanks	13

4.4	Chain of Custody and Field Forms	13

4.5	Shipping	13

5.0 Data Analysis and Management	15

5.1	Data Analysis, Interpretation, and Management	15

5.2	Data Reporting Requirements	15

5.3	Data Analysis	15

5.4	Data Storage Requirements	16

6.0 Quality Metrics (QA/QC Checks)	17

6.1 Calibration and Auditing of Field Equipment	17

7.0 Results	17

7.1	Results from the Summer 2017 Monitoring	17

7.2	Results from the Supplemental Study Conducted at EPA Research Triangle Park
Facility	25

8.0 Conclusions	32

9.0 References	34

10.0 List of Acronyms and Abbreviations	36

II.0	Appendix A: Preparation and Extraction of Sampling Media	38

11.1	ADFPS Phosphorous Acid Coated Denuder	38

11.2	ADFPS Sodium Carbonate Coated Denuder	39

11.3	Phosphorous Acid Impregnated Cellulose Filters	40

11.4	Nylon Filters	42

12.0 Appendix B: Chemicals for IC Analysis	43

13.0 Appendix C: Chain of Custody Forms	45

14.0 Appendix D: Characterization of NH3 Breakthrough on Phosphorous Acid Coated Annular

Denuders	49

14.1	Project description and objectives	49

14.2	Field site	49

14.3	Annular Denuder/Filter Pack System	50

14.4	Chemical analysis	51

14.5	Collection efficiency calculations	51

15.0 Appendix E: Validation Log for Study Samples	52

16.0 Disclaimer	59


-------
List of Figures

Figure 1. Monitoring Locations in the Southeastern US	4

Figure 2. Duke Forest, NC Monitoring Location	5

Figure 3. Duke Forest NHX Sampling Equipment	6

Figure 4. Gainesville, FL Monitoring Location	7

Figure 5. Flow Volumes measured at the Duke Forest, NC Site	19

Figure 6. Flow Volumes measured at the Gainesville, FL Site	19

Figure 7. Comparison of NHX Concentrations Measured by IMPROVE, CSN, and ADFPS

Methods at Duke Forest, NC	20

Figure 8. Scatter Plot of Primary ADFPS NHX Concentration compared with the Duplicate

ADFPS at Duke Forest, NC	20

Figure 9. Linear relationship between the ADFPS versus CSN NHX results and ADFPS versus

IMPROVE NHX results at Duke Forest	21

Figure 10. Box plot of Duke Forest, NC NHX concentrations	22

Figure 11. Comparison of NHX Concentrations Measured by IMPROVE, CSN, and ADFPS

Methods at Gainesville, FL	23

Figure 12. Scatter Plot of Primary ADFPS NHX Concentration compared with the Duplicate

ADFPS at Gainesville, FL	23

Figure 13. Linear relationship between the ADFPS versus CSN NHX results and ADFPS versus

IMPROVE NHX results at Gainesville, FL	24

Figure 14. Box plot of Gainesville, FL NHX concentrations	25

Figure 15. ADFPS with additional breakthrough denuders for determining if NH3 collected on
the backup denuder was a result of NH3 capture efficiency on the primary acid

denuder or volatilization of NH+4 from the nylon filter	25

Figure 16. Ammonium retained on the nylon filter as percent of total (nylon filter plus NH3

captured on the backup denuder from volatilized NH+4)	26

Figure 17. Time series from Duke Forest of (a) S024 concentrations and (b) NH+4 concentrations
from ADFPS (green), CASTNET (orange - Teflon), and CSN (green - nylon).

ADFPS and CSN concentrations are weekly averages matching the CASTNET

sampling period	28

Figure 18. Minimum fraction of NH* loss association with N03 and S024	30

Figure 19. Mean absolute relative percent difference (MARPD) between CSN samplers with

(orange) and without (blue) a cyclone at the inlet	31

Figure 20. EPA campus study site AIRS	50


-------
List of Tables

Table 1. Summary of measurement methods from both the Duke Forest and Gainesville

monitoring locations	8

Table 2. QC Procedures, Acceptance Criteria, and Corrective Actions for IC Chemical

Analyses	12

Table 3. Data Quality Indicators for Laboratory Analyses	13

Table 4. Flow Volume Targets for Each Sampling System	17

Table 5. Summary of Blanks Results	18

Table 6. Ratio of NH+4: S024 from filter extracts during the NHX sample period	27

Table 7. Cation concentration curves (|jg/mL)	44

iv


-------
1.0	Introduction

1.1	Overview

Total reduced nitrogen (NHX) measurements are not routinely performed in the US and currently
no regulatory requirements exist to measure NHX. However, scientific and policy interest in the
increasing trend in NHX concentrations in many regions of the US has evolved since NHX can
lead to atmospheric fine particulate matter (PM2.5) formation and visibility impairment and, once
deposited, can cause eutrophication, loss of species biodiversity and algal blooms. This study,
set in the southeastern US, was designed to measure reduced nitrogen and to assess
measurement methods for implementation in long-term monitoring networks, specifically the
Interagency Monitoring of Protected Visual Environments (IMPROVE) and Chemical Speciation
(CSN) networks. The IMPROVE network provides routine measurements of speciated
particulate matter (PM) that support the regional haze rule and source apportionment analyses.
The CSN provides data that support emission control strategies, health effects research, and
model evaluation. Both networks collect a 24-hour integrated sample every 3 days.

The goal of this study was to determine whether acid-coated filters could be deployed at existing
sites using existing network equipment to measure reduced nitrogen concentrations in hot,
humid environments. Incorporating an NHX measurement into the CSN and IMPROVE networks
would leverage existing infrastructure, reduce costs and provide national coverage. The addition
of NHX also offers the potential for providing NHX and particulate ammonium (NH4) estimates at
sites co-located with the National Atmospheric Deposition Program's (NADP) Ammonia
Monitoring Network (AMoN). Hourly measurements are ideal for evaluation of chemical
transport models (CTMs) and understanding atmospheric processes that vary diurnally. On the
other hand, the use of speciation samplers for 24-hour NHX samples, collected every 3 days, is
a cost-effective way to improve current understanding of the spatial and temporal variability of
NHxand to generate data useful for source apportionment, potential for PM formation, and more
general model evaluation. When co-located with AMoN, monthly, seasonal and annual
averages of NH4 can be generated which are of direct benefit to PM and oxides of nitrogen and
sulfur (NOx/SOx) NAAQS reviews and subsequent implementation. Chen etal. (2014) performed
a similar study in the Western and midwestern US using only the IMPROVE sampler. In that
study the acid-impregnated filter data compared well with Annular Denuder Filter Pack System
(ADFPS) (URG Corp) measurements.

1.2	Discussion

Over the past two decades, the chemical composition of the atmosphere has changed markedly
from one dominated by oxidized secondary particles including nitrate (N03), sulfate (S024), and
carbonaceous aerosols, and their associated precursor gases (NOx, SOx, VOCs, SVOCs), to an

1


-------
atmospheric mixture that includes a significant amount of reduced inorganic nitrogen (NHX =
NH4 + NH3) (Li eta!., 2016; Du eta!., 2014). Total atmospheric nitrogen compounds are still
primarily in inorganic forms but have significant contributions from organically bound nitrogen
(both reduced and oxidized forms) (Jickells eta!., 2013). Looking forward, the shift to larger
relative contributions of atmospheric NHX is projected to continue as existing rules and
technologies continue to decrease NOx emissions from the transportation and energy
generation sectors (U.S. EPA, 2016) with an expected modest increase in NH3 emissions (Ellis
eta!., 2013; Martin eta!., 2015). As atmospheric emissions and composition evolve in response
to rules, ambient and deposition monitoring networks must evolve to best inform decision
makers and address such questions as:

1.	Cause of adversity: What atmospheric species and emission sources contribute to PM,

regional haze and nitrogen deposition?

2.	Mitigate adversity: What options in emissions reductions are available to reduce PM,

regional haze and nitrogen deposition?

a.	How much remaining capacity is available in rules and NAAQS provisions that target
emission reductions in oxidized nitrogen species?

b.	How effective are potential reductions in reduced nitrogen?

To answer these questions, it is necessary to characterize reduced nitrogen emission sources
and atmospheric composition. Ammonia is the precursor emission while particulate NH+4 is
important as the controllable pollutant under the PM NAAQS. Current monitoring networks are
relatively well positioned to characterize inorganic oxides of nitrogen (NOy), but poorly
positioned to characterize NHX, the increasingly dominant reduced component of atmospheric
nitrogen. This limitation is affecting our ability to understand and mitigate the causes of excess
PM, regional haze and nitrogen deposition. Measurements of NHX would provide a metric for
evaluating emission inventories and validating chemical transport models and provide input into
nitrogen deposition estimates. Having the speciated measurements (NH3 + NH+4) is analogous to
the importance of NOy measurements.

Ammonia and NH+4 directly contribute to excess nitrogen deposition through dry and wet
deposition and together now dominate the inorganic nitrogen deposition budget across most of
the U.S. (Li eta!., 2016). Ammonium directly contributes to PM and the resulting haze, while
NH3 indirectly affects PM and haze levels. Ammonia also plays an important role in governing
aerosol acidity (Silvern et al., 2017; Pye et al., 2020) which effects its hygroscopicity (water
uptake). This in turn influences the aerosol water content and its contributions to haze and
formation of secondary organic aerosols from biogenic VOC emissions (Carleton eta!., 2010,
Carleton and Turpin, 2013). The lack of reduced nitrogen measurements hinders our ability to
assess these direct and indirect effects on air quality issues and the ability to develop, test and
refine the simulation of important atmospheric processes governing the roles of reduced
nitrogen compounds in particle formation and deposition. The importance of understanding

2


-------
trends and the spatial variability of reduced nitrogen is recognized; NH+4 concentrations are
measured at more than 150 CSN sites and more than 90 CASTNET sites and NH3
concentrations are measured at more than 90 sites as part of AMoN. However, the CSN
measurements underestimate NH+4 concentrations (Yi eta!., 2006) while CASTNET measures
weekly integrated samples and AMoN collects two-week samples, making these data most
suitable for seasonal and annual characterizations.

1.3 Study Objectives

Study objectives from the Southeastern US Reduced Nitrogen Study are:

1.	Evaluating the feasibility of deploying acid-impregnated filters to measure NHX in national
networks

2.	Characterizing any biases in existing NH+4 measurements collected at CSN and
CASTNET sites

3.	Provide a recommendation to the Office of Air Quality Planning and Standards (OAQPS)
who manages the CSN for measuring NHX using a cost-effective method on a timescale
that informs the PM NAAQS review, regional haze policies, and critical loads
exceedances linked to reduced nitrogen deposition

3


-------
2.0 Measurement Locations

In the pilot study by Chen et at. (2014), it was shown that NHX could be measured in the
IMPROVE network at several Western and Midwestern sites. Before deploying this technique in
national networks, this study expanded the Chen etal. (2014) work to the warm, humid eastern
U.S. Acid-impregnated filters were deployed in both CSN and IMPROVE samplers in North
Carolina and Florida during the summer of 2017 (Figure 1). In this study, a spare module [37
millimeter (mm) filter diameter for IMPROVE, 47 mm filter diameter for CSN] was fitted with a
phosphorous acid (H3PO3) impregnated cellulose filter to collect and sustain (no volatile losses)
both particulate NH4 and NH3. A URG annular denuder filter pack system (ADFPS) was used as
a reference method. The ADFPS included an NH3 denuder, followed by a nylon filter to collect
particles and a backup denuder to capture any NH3 volatilized from the nylon filter. The sum of
NH3 from both denuders and NH4 from the nylon filter constituted a reference NHX sample.

Figure 1. Monitoring Locations in the Southeastern US

Duke Forest

Gainesville

4


-------
The Duke Forest research site in the NC Piedmont area (Figure 2) was managed by John
Walker of EPA's Office of Research and Development (ORD).. Measurements were conducted
in an unfertilized 15 ha grass field in the Blackwood Division of Duke Forest, Orange County,
North Carolina, USA (35.9745 latitude,

-79.0990 longitude). A variety of air quality and meteorological measurements (N.C. Forest
Service Fire Forecast, National Oceanic and Atmospheric Administration (NOAA) Climate
Reference Network, and AMoN) were performed at this site, which is periodically also used for
special studies (Rumsey and Walker, 2016). Instruments were operated in the middle of the
field adjacent to the NC Forest Service and NOAA meteorological towers and collocated with
the AMoN sampler (Figure 3). In addition to the routine measurements conducted in the grass
field, weekly CASTNET filter pack measurements (DUK008) and meteorological data were
collected at the forest flux tower located immediately adjacent to the grass field.

Figure 2. Duke Forest, NC Monitoring Location

Measurement
Location

Chapel Hill

5


-------
Figure 3. Duke Forest and Gainesville NHX Sampling Equipment

Duke Forest NHx study site

URG den uder/filter pack

•	Separates NH3 and NH-*

*	Acid coated denuder (NH,)
Nylon filter (NH4+)

*	Backup denuder (volatile NHj)

•	Duplicates

•	PM2 $ inlet @> 10 Lpm

CSN

•	One module collecting MH4* on
nylon filter

•	2n(J module collecting total NHx on
acid impregnated cellulose filter

•	PM2 s inlet at 6,7 Lpm

Gainesville NHx study site

The Gainesville, FL site was served by Wood Environment & Infrastructure Solutions (Wood),
which operates CASTNET and several of the NADP network sites. The Gainesville site
(29.6497 latitude, -82.4914 longitude) served as the 2nd study site. Instrumentation was
mounted on the roof of the CASTNET shelter used for testing equipment/methods (Figure 3). In
addition to the IMPROVE, CSN, and ADFPS measurements, weekly CASTNET filter pack,
biweekly AMoN NH3 samples, and hourly meteorological data were also collected in Gainesville.

6


-------
Figure 4. Gainesville, FL Monitoring Location

In addition to the two study sites, additional QA testing was performed at the Research Triangle
Park ambient air innovation research site (AIRS) located on the EPA campus in late 2017
through early 2018 using the same ADFPS system deployed at Duke Forest. Equipment was
operated by staff in EPA ORD/AEMD.

7


-------
3.0 Sampling Methods & Procedures

A description of the measurements is provided in this study's Quality Assurance Project Plan
(QAPP) (U.S. EPA, 2018). Table 1 summarizes the samplers operated at the two sites.

Table 1. Summary of measurement methods from both the Duke Forest and Gainesville
monitoring locations

Sampler

Channels
(measured)

Cutpoint

Flow Rate

Sample Period

URG
denuder/filter pack
system (ADFPS) -
in duplicate

Na2CC>3-coated
denuder (HNO3 -
not analyzed)

Teflon coated
cyclone - 2.5
|jm

10 Ipm

1 in 3 day, 24-
hour samples

H3PC>3-coated
denuder (NH3)

Nylon filter (NH+4)

H3PC>3-coated
backup denuder
(volatile NH+4)

CSN MetOne
SuperSASS

MgO denuder (not
analyzed) + nylon
filter (NH4)

Cyclone - 2.5
|jm

6.7 Ipm

1 in 3 day, 24-
hour samples

47 mm H3PO3-
impregnated
cellulose filter (NHX)

Cyclone - 2.5
|jm

6.7 Ipm

IMPROVE PM
Sampler

37 mm H3PO3-
impregnated
cellulose filter (NHX)

2.5 |jm

22.8 Ipm

1 in 3 day, 24-
hour samples

CASTNET

Teflon filter (NH+4,

so24, no3)

N/A

3 Ipm

Weekly

Nylon filter (HNO3,
S02)

K2CO3 impregnated
cellulose filter (S02)

AMoN

Radiello (NH3)

N/A

Passive

Bi-weekly

Notes:

Na2CC>3: sodium carbonate
H3PO3: phosphorous acid
MgO: magnesium oxide
K2CO3: potassium carbonate

8


-------
The acid-impregnated cellulose filters eliminated off-gassing of ammonia that can occur in
routine sampling operations when collecting ions on nylon filters with base properties. The nylon
and cellulose filters were subsequently analyzed by the Wood laboratory for NH+4 by Ion
Chromatography (IC). With the exception of preparing the H3PC>3-impregnated filters, described
in section 10.3 of this report, all sampling and analysis protocols were identical to those in the
routine operations of IMPROVE (http://vista.cira.colostate.edu/lmprove/particulate-monitoring-
network/) and CSN (https://www.epa.gov/amtic/chemical-speciation-network-field-qapps-and-
sops). The ADFPS protocols were adapted from EPA's Inorganic compendium Method I04.2
(https://www3.epa.gov/ttn/amtic/inorq.htmn.

Filters and denuders were prepared and shipped to Duke Forest by the Wood laboratory. The
ORD laboratory extracted the denuders and shipped the extracts to Wood for analysis. Wood
prepared and shipped travel blanks to the Duke Forest site. Laboratory blanks were routinely
run for each method.

Nylon extracts from CSN and the ADFPS were re-analyzed in 2018 for S024 and N03 following
the CASTNET QAPP v9.2 (U.S. EPA, 2019).

9


-------
4.0	Laboratory Procedures

4.1	Preparation of Sampling Media

Procedures for preparation and extraction of sampling media are summarized in Appendix A.

4.2	Chemical Analysis

Laboratory IC quality control procedures are fully described in section 3.3.2 of the CASTNET
QAPP (U.S. EPA, 2019). Details relevant to this study are summarized below. It is noted that
while CASTNET routinely analyzes for NH+4 using colorimetry, IC analysis was used in this
project for consistency with IMPROVE and CSN laboratory methods. The description and
methods for IC analysis for the CASTNET anions in the CASTNET QAPP Appendix 4
Laboratory SOP (GLM-3180-001) is the same procedure that was followed here for NH+4.

Denuder and filter extracts were analyzed for NH+4 by Wood using a Thermo/Dionex ICS-1600
ion chromatograph (Thermo/Dionex Corporation, 2009) equipped with a Thermo/Dionex CG16
cation guard column, Thermo/Dionex CS16 cation analytical column, and Thermo/Dionex CERS
500 self-generating suppressor. Samples were injected onto the analytical column from a 50 |jL
loop and eluted in 30 mM methanesulfonic acid at a flow rate of 1 mL min-1.

Procedures for chemical analysis and quality assurance are described below. Details of
chemical preparation for chemical analyses are included in Appendix B.

4.2.1	Sample Analysis

The IC analysis follows the procedures described in the CASTNET QAPP Appendix 4
Laboratory SOP (GLM3180-001 section 6.2) with the exception of cation columns and the use
of the methanesulfonic acid solution as the eluent.

4.2.2	Data Analysis

1.	Data files were processed using algorithms contained in the data collection software
(Chromeleon 7.2 from Thermo/Dionex). Parameters were adjusted as dictated by
instrument performance.

2.	Chromatograms were examined visually. Any anomalies in the data batch were noted in
the narrative.

3.	Responses were exported to an Excel spreadsheet.

4.	Data were assembled in a batch folder, including copies of all extraction worksheets, run
logs, certificates of analyses and processing methods, hard copies of each
chromatogram, and other necessary documentation.

10


-------
4.2.3 Quality Control

1.	Filter blanks (BLK): If any analyte from an extracted filter exceeded 1 |jg/filter, the box
that the filter came from was marked "Failed", removed from the lab and not used for
sample collection.

2.	One BLK was analyzed during each extraction. The BLK for extracted samples was the
applicable volume of extraction solution followed by the appropriate extraction
procedure. The BLK results were required to be less than or equal to two times the
reporting limit for the analytes of concern.

3.	A CCV was analyzed at a frequency of 10 percent for every analytical batch, as well as
at the beginning and end of each run. The measured value of the CCV was required to
be within ±10 percent of the certified value.

4.	An SRM was used for an initial and a final calibration verification. The measured value of
the reference sample was required to be within ±10 percent of the certified value.

5.	All calibration curves were required to contain a minimum of five points for quadratic
calculations and have a correlation coefficient greater than or equal to 0.995. See
Appendix B for a complete description.

6.	Approximately 5 percent of samples from each batch were analyzed in duplicate (DUP1,
DUP2) to monitor within-run precision. Sample extracts were selected at random and
reinjected for analysis. For samples greater than five times the reporting limit (0.02 |jg-N
mL_1), the relative percent difference (RPD) of the replicate samples was required to be
within ± 20 percent. For samples with concentrations less than or equal to five times the
reporting limit, the absolute difference between sample and replicate was required to be
less than the reporting limit.

7.	An internal system monitoring spike (rubidium bromide) was used in the IC analysis to
assess shifts in retention time and sample injection volume.

8.	All sample responses were within the standard calibration range (0.00 to 5.00 |jg mL-1).
Any samples with responses above the calibration curve high standard were diluted and
reanalyzed.


-------
4.2.4 Corrective Actions

Table 2 summarizes QC procedures and the corrective actions taken when the QC samples
were not within acceptance criteria.

Table 2. QC Procedures, Acceptance Criteria, and Corrective Actions for IC Chemical
Analyses

Quality Control

Acceptance Criteria

Corrective Action

Calibration curve

Correlation coefficient > 0.995

Rerun calibration standards. If still
out of control, prepare new
calibration standards and
recalibrate the instrument, or
document why data were
acceptable.

Calibration curve
responses

Bracket all samples

Dilute and reanalyze samples
exceeding the calibration range, or
document why data were
acceptable.

Reference standard
(SRM) [Accuracy
indicator; brackets
all samples in run]

± 10 % of the certified true value

Rerun standard. If still out of
control, recalibrate instrument and
reanalyze samples, or document
why data were acceptable.

Control standard
(CCV) [Accuracy
indicator; analyzed
every 10 injections}

± 10% of the certified true value

Rerun standard. If still out of
control, recalibrate instrument and
reanalyze samples since last
acceptable CCV, or document why
data were acceptable.

MB (BLK)

< 2 times the RL

Determine the cause of blank
problem.

Notes: RL = Reporting limit

RPD = Replicate percent difference

Source: Wood

Laboratory precision was estimated by analyzing replicate injections of randomly selected
extracts. Approximately 5% of the IC samples from each batch were reanalyzed. Differences
between the original and replicate concentrations were calculated as relative percent difference
(RPD). The data quality indicator (DQI) precision goals are summarized in Table 3. Laboratory
accuracy was determined by analyzing reference samples and control standards (CCV). An
independent National Institute of Standards and Technology (NIST)-traceable reference
standard was analyzed at the beginning and end of each analytical run. One midlevel CCV,
produced by an independent laboratory and NIST-traceable, was analyzed every ten IC
samples. The responses relative to the CCV and reference samples were required to be within
±10% (the accuracy DQI criterion) of the certified target values.

12


-------
Table 3. Data Quality Indicators for Laboratory Analyses



Acceptance criteria

Analyte

Method

Precision
(RPD)

Accuracy

(%)

Nominal
Reporting
Limits

Lower Limit
of Detection

Ammonium
(NH4)

IC

20

90-110

0. 020 |jg-
N/mL

0. 020 |jg-
N/mL

4.2.5 Calculations

All calculations were performed with data reduction algorithms that reside in the instrument
software. Separate calibration curves were prepared for NH4 by plotting the response (peak
area) of standards against concentration values using quadratic regression in the instrument
software. Sample concentrations were calculated using the quadratic equation for the curve.
The analyst was allowed to eliminate points to improve accuracy throughout the range of
calibration but at least 5 points plus a blank had to remain.

4.3	Field and Laboratory Blanks

Blanks were assessed to characterize contamination on denuders and filters that may have
occurred in the laboratory and field. Weekly laboratory blanks were prepared and analyzed on
H3PO4 denuders, H3PO4 filters, and nylon filters. Field or trip blanks for ADFPS, IMPROVE, and
CSN were prepared and analyzed biweekly at both field sites. For IMPROVE and CSN, field
blanks consisted of filters placed in the normal configuration within the sampler but unexposed.
For the ADFPS, field blanks consisted of a standard sample train, including the cyclone inlet,
placed in the ADFPS enclosure in the normal sampling configuration but left unexposed. The
ADFPS field blank was deployed and collected with the exposed sample for the corresponding
sample period. Trip blanks were not unpacked from the shipping container. Note that the
ADFPS samplers were changed out two or three times per week, depending on the 1:3
schedule, as opposed to weekly change out of the IMPROVE and CSN samplers.

4.4	Chain of Custody and Field Forms

Chain of custody forms accompanied all samples during shipping from the analytical laboratory
through field sampling and until receipt and analysis. Field notes (including sample volume, time
and sample module identifiers) were recorded on the forms and archived at the analytical
laboratory. Examples of the chain of custody and field forms are included in Appendix C.

4.5	Shipping

All samples and extractions were shipped in insulated coolers with ice blocks to maintain a
temperature below 4°C. Filter based samples were shipped to the EPA facility and transported
to and from the Duke Forest site each Monday by the site operators, to be installed during the

13


-------
following week. Upon receipt, the receiver recorded the temperature of the samples, time of
receipt, and initials on the Shipping Temperature Log (See Appendix C) included with each
cooler and transferred the samples to a refrigerator until deployment. The samples exposed
during the previous week, along with denuder extractions, were returned to the analytical
laboratory using fresh ice blocks. Upon receipt at the laboratory, the temperature of the
samples, time of receipt, and initials were recorded on the Shipping Temperature Log and all
samples are transferred to a cold room for storage.

All samples collected at the Gainesville site were stored in the laboratory cold room at < 4°C
until deployment and immediately following sample collection.

14


-------
5.0 Data Analysis and Management

5.1	Data Analysis, Interpretation, and Management

Laboratory, flow, concentration, and QC data were managed and reported to EPA by Wood
under EP-W-16-015 Task Orders 1012 and 2012.

5.2	Data Reporting Requirements

Laboratory data were stored in Wood's laboratory information management system (LIMS)
Element Data System. Element is used to organize and schedule the analyses performed by the
CASTNET laboratory. Laboratory procedures, including setting up projects in the Element Data
System are described in the CASTNET QAPP (U.S. EPA, 2019). Calculations for converting the
mass of an analyte (NH+4) in each extract to atmospheric concentrations are described in section
4.4.1 of the CASTNET QAPP. Laboratory data, including mass, extract volumes, and blank
results were transferred to SQL Server for calculating atmospheric concentrations. The
CASTNET QA Manager verified the concentration data before the results were transferred to
EPA as described in the CASTNET QAPP Appendix 6 (Data Operations SOPs).

5.3	Data Analysis

Duplicate URG ADFPS systems were deployed to collect collocated samples to determine
precision of the reference method. Precision was calculated by aggregated mean relative
percent difference (MRPD):

where:

51	= The value for the primary measurement

52	= The value for the co-located duplicate measurement

k = The number of pairs of duplicate measurements (approximately 44 for this study)

Accuracy is calculated as percent recovery (%R) between the IMPROVE or CSN NHX or NH+4
measured concentration and the URG ADFPS concentration:

1

MRPD = -
k

sr (Si - si\

/	*200

Zj \S1 + 52/

where:

Y = The measured value (CSN or IMPROVE)
X = The reference value (ADFPS)

15


-------
Bar charts were created using R, SAS, or Tableau software to compare the 24-hour
concentrations between the three measurement systems for each site. Additionally, average
concentrations were reported for each measurement system for NHX and NH+4.

5.4 Data Storage Requirements

Electronic copies of the data are stored by Wood and EPA/Clean Air Markets Division (CAMD)
in an Oracle database. Data tables were created to store the CSN, IMPROVE, and URG
ADFPS ambient concentration data. QA, flow, and temperature data are stored in a separate
table. Wood will maintain the data records in accordance with the CASTNET QAPP. The
CASTNET database is maintained by Wood's Data Management Center (DMC) and Data
Management, Analysis, Interpretation and Reporting Manager (DMAIRM). The CASTNET
database is backed up following the procedures and schedules described in the CASTNET
QAPP.

16


-------
6.0 Quality Metrics (QA/QC Checks)

6.1 Calibration and Auditing of Field Equipment

Continuous data (flow and temperature) were validated according to the validation procedures
described in section 4.3 of the CASTNET QAPP (U.S. EPA, 2019), following calibration
procedures for the SuperSASS sampler (https://www.epa.gov/amtic/chemical-speciation-
network-quality-assurance). Flow data are typically flagged as invalid if the value is outside the
nominal flow by > 10% (https://www.epa.gov/amtic/chemical-speciation-network-data-reporting-
and-validation). Ambient temperature data are flagged if the value is < -40 or > 50 ° C. However,
for this study, criteria for data review were established based on flow volume criteria, and no
samples were invalidated based solely on nominal flow in order to improve data capture. Table
4 summarizes the target flow volumes per sample for each system.

Table 4. Flow Volume Targets for Each Sampling System

Sample System

Target (m3)

CSN SuperSASS

9.6

IMPROVE PM Sampler

31.3

URG ADFPS

14.6

Flow calibrations were performed at the start of the sampling period, approximately at the mid-
point of the study, and at the end of the study period. Flow rates for the SuperSASS and ADFPS
samplers were verified using a NIST-traceable MesaLabs Definer 220 dry piston flow meter
(MesaLabs, Lakewood, CO). Calibration of the SuperSASS sampler is described by EPA:
(https://www.epa.gov/amtic/chemical-speciation-network-gualitv-assurance). Calibration of the
IMPROVE sampler is described in the IMPROVE SOP 226

(http://airgualitv.crocker.ucdavis.edu/files/6614/5808/2356/TI226H Calibration of Flow Check
Devices.pdf). A Magnehelic differential pressure meter (Dwyer Instruments, Michigan City, IN)
was used to calibrate pressure sensors in the IMPROVE sampler. The flow rate through the
sampler was calculated using the pressure drop across a flow restriction and the density of
ambient air. The Magnehelic flow check device was calibrated using a Definer 220 NIST-
traceable flow meter. Calibration results were used to review the concentration measurements
from each sampler-type. During flow audits, temperature sensors in the SuperSASS and
IMPROVE samplers were calibrated at one point using a NIST-traceable reference
thermometer. The temperature criterion for each system is ± 1 degree C. The SuperSASS and
IMPROVE pressure sensor criteria were ± 10%.

7.0	Results

7.1	Results from the Summer 2017 Monitoring

Field, trip and laboratory blanks were routinely monitored for each sampling system. A summary
of the blank results is shown in Table 5. Median concentrations are shown in |jg-NH4 mL_1 and

17


-------
trip blanks were only shipped to Duke Forest because the laboratory was located at the
Gainesville monitoring site. The number of samples are shown in parentheses.

Table 5. Summary of Blanks Results

Blank

Duke Forest

Gainesville

Field Blanks





ADFPS Nylon Filter

0.000 (3)

0.0002 (7)

CSN Nylon Filter

0.000 (3)

0.0031 (7)

Acid-Coated Denuder

0.0576 (3)

0.0032 (7)

CSN Acid-Impregnated Filter

0.0152 (3)

0.0048 (7)

IMPROVE Acid-Impregnated Filter

0.0162 (3)

0.002 (7)

Laboratory Blanks





37 mm Cellulose Filter

0.000 (30)

47 mm Cellulose Filter

0.000 (30)

Nylon Filter

0.000 (30)

Acid-Coated Denuder

0.0046 (13)

0.0054 (15)

Trip Blanks





ADFPS Nylon Filter

0.00235 (4)



CSN Nylon Filter

0.00165 (4)



Acid-Coated Denuder

0.00665 (4)



CSN Acid-Impregnated Filter

0.0078 (4)



IMPROVE Acid-Impregnated Filter

0.00715 (4)



Flow was recorded continuously for the ADFPS, IMPROVE and CSN samplers. Figures 5 and 6
show flow volume from Duke Forest and Gainesville, respectively. Actual flows were used to
calculate the ambient concentrations. At Duke Forest the CSN flow rate was stable and the
IMPROVE flow rate was stable until 10/1/2017. The ADFPS at Duke Forest performed well. At
the Gainesville site all three sampling types experienced high variability in the flow.

18


-------
Figure 5. Flow Volumes measured at the Duke Forest, NC Site

40

35 -
30 -
25 -
E 20 -
15 -
10 |
5

fr



i i





5/25/17	6/25/17	7/25/17	3/25/17	9/25/17

¦ ' D-IMPROVE	D-ADSFP

-—-D-ADSFP Duplicate	—•— D-CSN (Nylon Filter Module)

D-CSN (Phosphorus Acid Filter Module)

10/25/17

Figure 6. Flow Volumes measured at the Gainesville, FL Site

6/25/17	7/25/17

G-IM PROVE

-G-ADSFP Duplicate (Dry Gas Meter)
-G-C5N (Nylon Filter Module)

3/25/17	9/25/17	10/25/17

- «1SFP

)5FP Duplicate (MFC)
3N (Phosphorus Acid Filter Module)

Ambient concentrations of NHX from each of the sampling systems are shown in Figure 7 from
the Duke Forest site. The ADFPS NHX value was reported as the nylon NH4 concentration plus
the NH3 collected on the primary denuder and the volatilized NH4 collected on the backup acid-
coated denuder. There was good agreement between the primary ADFPS and the duplicate
system. The correlation between the two showed a slope of 0.94 and an r2 of 0.9 (Figure 8).

19


-------
Figure 7. Comparison of NHX Concentrations Measured by IMPROVE, CSN, and ADFPS
Methods at Duke Forest, NC

5/25/17	6/25/17	7/25/17	8/25/17	9/25/17	10/25/17

D-IMPROVE -*-D-ADSFP NHx (Denuder+Filter Pack) -< D-ADSFP NHx Duplicate -«-D-CSN

Figure 8. Scatter Plot of Primary ADFPS NHX Concentration compared with the Duplicate
ADFPS at Duke Forest, NC

1.4

1.2 -

E 1.0 -

3-

S: 0 8

w
Q
<

0

1	0.6 H

"q.

3

Q

x 0.4

• •

v •.

0.94*NHx_Pri_ADSFP + 0.03
R2 = 0.89

0.2 -

0.0

0.0	0.2	0.4	0.6	0.8	1.0

NHx Primary ADSFP (jig N m"3)

1.2

1.4

20


-------
The linear relationship between the ADFPS and the CSN SuperSASS sampler and the
IMPROVE sampler from Duke Forest are shown in Figure 9. There was good agreement
between the ADFPS and IMPROVE systems with a slope of 0.95 and r2 equal to 0.89. These
results were similar to results presented in the western US in the NHX pilot network (Chen et al.,
2014). Agreement between the ADFPS and the CSN system was fair with a slope of 0.69 and
an r2 equal to 0.71. The CSN was biased low as compared to the ADFPS. There seems to be
less NHX retained on the acid-impregnated filter in the CSN system at higher concentrations.

Figure 9. Linear relationship between the ADFPS versus CSN NHX results and ADFPS

1,6

1.4

iSP	•

E 1 2

~ZL	¦ *

S	improve;	• •

iH 1 0	0.95,NHx ADSFP + 0.11	«

§	PP = 0,39	.	•

£	mm* 	•

5 0,3	. *	* * 		""

1	•

O 0.6	«

2	• » V"" -r t

o	•	•• • •

5 0.4

z	¦"'*# t •	CSN

.	0,89*NHx_AD8FP + 0,16

02	R2 = 0,71

0,0 	

0,0	0,2	0,4	0,6	0,8	1,0	1,2	1,4

NHx ADSFP (ng N rrr3)

9. Linear relationship between the ADFPS versus CSN NHX
versus IMPROVE NHX results at Duke Forest

IMPROVE
0.95,NHx_ADSFP + 0.11

R? 0„8g

®	s,- •	-

#	J • f,--i

5 * .>-•"" *

* •

# .9'' *	• • m

• •

CSN

.	0,89*NHx_AD8FP + 0,16

R2 = 0.71

21


-------
Figure 10 shows the distribution of NHX concentrations measured during the study at the Duke
Forest, NC site. The boxes represent the 25th and 75th percentile, the solid and dashed lines
within the box represent the median and mean, respectively, the whiskers represent the 10th
and 90th percentiles, and dots represent the 5th and 95th percentiles. Median concentrations over
the 6-month period were similar for each sampling method.

Figure 10. Box plot of Duke Forest, NC NHX concentrations

1.2
1.0
0.8

CD

~ 0.6

CO

X

0.4
0.2

ADFPS

CSN IMPROVE

The time series from Gainesville, FL for each of the sampling systems is shown in Figure 11.
The CSN pump began to fail following removal of the sample installed on 6/6/2017.

Performance slowly degraded, and it was repaired prior to the sample installed on 7/15/2017.
This issue, along with several other operational problems with the SuperSASS, resulted in a
loss of almost 2 months of concentration data during June and July. The NHX results from the
IMPROVE and CSN samplers showed a negative bias. There was good agreement between the
primary and duplicate denuder systems with a slope of 0.81 and an r2 equal to 0.69 (Figure 12),
although precision was not as good compared with results from Duke Forest.

22


-------
Figure 11. Comparison of NHX Concentrations Measured by IMPROVE, CSN, and ADFPS
Methods at Gainesville, FL

5/25/17	6/25/17	7/25/17	8/25/17	9/25/17	10/25/17

-G-IMPROVE -*-G-ADSFP NHx (Denuder+Filter Pack) G-ADSFP NHx Duplicate -«-G-CSN

Figure 12. Scatter Plot of Primary ADFPS NHX Concentration compared with the Duplicate
ADFPS at Gainesville, FL

1.4

1.2 -

'E 1.0 -

S

S: o.s

CO
~

<

0

ro 0.6 -

Q

x 0.4
z

0.2 -
0.0







•







•







•





•







• *• ...







• •

•









•





•

0.81*NHx Pri ADSFP + 0.15
R2 = 0.69

0.0

0.2

0.4	0.6	O.i

NHx Primary ADSFP (jig N rrr3)

1.0

1.2

23


-------
Agreement between the ADFPS and the CSN SuperSASS sampler and the IMPROVE sampler
from Gainesville are shown in Figure 13. At Gainesville the range of concentrations as
measured by the ADFPS was narrower (0.17-1.13 |jg rrr3) than the NHX concentrations
measured in Duke Forest (0.087-1.29 |jg rrr3). The agreement in NHX concentrations was
relatively poor compared to Duke Forest for both CSN and IMPROVE with a slope = 0.43 and r2
= 0.44 and slope = 0.72 and r2 = 0.42, respectively. While the cause of the poor performance is
not obvious, differences in meteorology between the two sites may be a factor.

Figure 13. Linear relationship between the ADFPS versus CSN NHX results and ADFPS

1,6

1.4 -

£ 1.2 -
z

3

ty 1.0 -	if • •

§	0,?2'*NH; »+ 0.07

QC	f t .*

a.

5 0,8 -	•

z:	«	*		

o 0.0 -	•

-r	mm	•	•

§

x 0,4 -	»

z	* 		• •	*

^ 		 » «

02-		""	*	CSN

0.43*NHx_ADSFP + 0.20

R2 = 0.44

0,0 -I	

0,0	0,2	0,4	0,6	0,8	1,0	1,2

NHx ADSFP (ng N rrr3)

Figure 14 shows the distribution of NHX concentrations measured during the study at the
Gainesville, FL site. As in Figure 10, boxes represent the 25th and 75th percentile, the solid and
dashed lines within the box represent the median and mean, respectively, the whiskers
represent the 10th and 90th percentiles, and dots represent the 5th and 95th percentiles. Both the
CSN and IMPROVE samplers collected less NHX than the ADFPS.

13. Linear relationship between the ADFPS versus CSN NHx
versus IMPROVE NHX results at Gainesville, FL

» + 0.07

• #

I.-.*'	 • V

• • „

• •

CSN

Q,43*NHx_ADSFP + 0.20

K2 = 0.44

24


-------
Figure 14. Box plot of Gainesville, FL NHX concentrations

O)

~

ADFPS

CSN IMPROVE

7.2 Results from the Supplemental Study Conducted at EPA Research Triangle Park
Facility

It was noted during the 6-month study period that there was a significant fraction of NHX being
collected on the backup acid coated denuder. The performance of the ADFPS was challenged
by collecting 24-hour samples with a second primary denuder as well as a second backup
denuder (Figure 15). The additional breakthrough denuders were deployed three times each at
Duke Forest and Gainesville. For more information about the field set up during the
supplemental study refer to Appendix D. The results indicated there was breakthrough of NH3 on
the primary denuder (the second primary denuder was collecting NH3). Additional sampling was
performed by EPA/ORD at the EPA Research Triangle Park (RTP) campus following the 6-
month study to further test the capture efficiency of the primary acid denuder.

Figure 15. ADFPS with additional breakthrough denuders for determining if NH3 collected
on the backup denuder was a result of NH3 capture efficiency on the primary
acid denuder or volatilization of NH4 from the nylon filter

Base denuder

Air In

first Primary acid
denuder

Second primary Nylon First backup
acid denuder filter acid denuder

-£

HN03, S02,
HO

NH3

NH3

breakthrough

NH4

NH3
volatilized MH4

Second backup
acid denuder

Air out



NH3

breakthrough from
backup denuder

25


-------
Testing at RTP showed that the collection efficiency of NH3 on the primary acid coated denuder
was > 95% (N=32), in contrast to the more limited assessment during the 6-month field study.
The NH4 collected on the nylon filter versus the total NH+4 collected from the nylon filter and
backup denuder was also assessed. The NH+4 retained on the nylon filter was < 30% of the total
NH* retained by the nylon filter plus the NH3 captured by the backup denuder. The backup
denuder captured NH, breakthrough from the primary denuder and NH* volatilized from the
nylon filter. Figure 16 shows the variability in the amount of NH4+ retained on the nylon filter (as
a % of the filter NhV plus NH3 on backup denuder) at the three sites. Similar losses of NH+4 from
nylon filters have been reported in the literature (Solomon etai., 2000; Yu etai., 2006).

Figure 16. Ammonium retained on the nylon filter as percent of total (nylon filter plus NH3
captured on the backup denuder from volatilized NH4). Boxes represent the
25th and 75th percentile, the solid and dashed lines within the box represent the
median and mean, respectively, the whiskers represent the 10th and 90th
percentiles, and dots represent the 5th and 95th percentiles.

(D

c
o

>
c

c
o

"O
CD
c

CO

-I—»

CD

RTP Duke Gainesville
Forest

The nylon filter had poor capture efficiency of NH4. On the other hand, the total NH+4 was
represented by the particulate fraction of NHX and that value was used to compare to the NH+4
concentrations measured by CASTNET. For the NHX comparisons described in section 7.1 the
ADSFP NHX concentration was calculated using the concentration of NH3 from the primary
denuder and the volatilized NH* captured as NH3 on the breakthrough denuder plus particulate
NH* that was retained on the nylon filter. This value should be an accurate representation of
total NHX concentration measured by the ADSFP, given the results of this breakthrough study.

Wood re-analyzed the nylon extracts for N03 and S04 to determine if the nylon filter was
ineffective at capturing particles in general or if the poor retention was specific to NH4. The ratio

26


-------
of NH+4: SO4 from the ADFPS extracts were compared to CASTNET and CSN. CASTNET
measures NH+4 and S024 using a Teflon filter. Results from Duke Forest and Gainesville
supported the Research Triangle Park results that the NH4 was not being retained on the nylon
filter. The results are summarized in Table 6.

Figure 17 shows the comparison between CASTNET, ADFPS and CSN for S024 and NH4.
Notably, S024 compared very well among the three methods but CSN was biased low for NH4
indicating the bias was not due to the overall nylon filter particle capture efficiency but rather the
ability of the filter to retain NhV after initial collection. Similar results were noted at existing co-
located CASTNET/CSN sites in Arendtsville, PA and Perkinstown, Wl. The median ratio of NH+4
/SO4 from CASTNET (Teflon/Teflon) was approximately twice that of CSN (nylon/nylon)
indicating lower capture efficiency of NH+4 on the nylon filter with good correlation for S024 from
the Teflon and nylon filters.

Table 6. Ratio of NH4: SO4 mass concentrations from filter extracts during the NHX
sample period

Method

Duke Forest

Gainesville

CASTNET (Teflon)

0.35



ADFPS (nylon + backup denuder)

0.35

0.35

CSN (nylon)

0.14

0.10

27


-------
Figure 17. Time series from Duke Forest of (a) SO* concentrations and (b) NH4

concentrations from ADFPS (green), CASTNET (orange - Teflon), and CSN
(green - nylon). ADFPS and CSN concentrations are weekly averages matching
the CASTNET sampling period.

Measure Names

¦	S04_ADS_WK

¦	S04_CNTFP

¦	S04_CSN_WK

« 0.35

c\| 1-

o

CO

0.00

May 29 Jun 13 Jun 28 Jul 13 Jul 28 Aug 12 Aug 27 Sep 11 Sep 26 Oct 11 Oct26 Nov 10

2017

28


-------
(b)

0.60
0.55
0 50
0.45
0.40
E 0.35

O)

Zi.

Z

V)

+ 0 30

+I
Z

0.25

0.20

0.15

0.10

0.05
0.00

The nylon filter captured SO4 but showed losses of NH4. To verify that the loss of NH4 from the
nylon filter was not due only to volatilization of NH4N03, the maximum amount of MH4 from the
nylon filter that could be associated with NH4N03was calculated. The remaining NH4 that could
be associated with (NH4)2S04 was then calculated. Those results were used to determine the
minimum fraction of NH4 loss from both NH4N03 and (NH4)2S04. The results are summarized in
boxplots as the ratio of NH4 as measured on the breakthrough denuder/maximum amount of
NH4 associated with N03 or SO,. The mean and median minimum fraction of NH4 lost from NH4
N03 was greater than 1 at both sites indicating there was more NH4 lost from the nylon filter
than could be explained by 100% volatilization of NH4NO3. The mean and median minimum
fraction of NH4 lost from (NH4)2S04 was approximately 0.5 at both sites, indicating a substantial
amount of loss associated with (MH4)2S04.

Measure Names

I NH4_ADS_WK
I NH4.CNTFP
I NH4_CSN_WK

May 29 Jun 13 Jun28 Jul 13 Jul28 Aug12 Aug27 Sep11 Sep26 Oct11 Oct26 Nov10

2017

29


-------
Figure 18. Minimum fraction of NH4 loss association with NOs and SO"

CO

o

i

o

CO

I

Duke Forest Gainesville



2.00

o

C/}

1.75

C\l

1.50

X

z

1.25



1.00

o



CO

0.75

LL



I

0.50

c



0.25



0.00

Duke Forest Gainesville

30


-------
During the supplemental study in Research Triangle Park, NC the inlet was removed from the
SuperSASS to verify that the CSN inlet wasn't scrubbing NH3. There were very small differences
between the sampling systems with and without the inlet. The results are shown in Figure 19.
The mean absolute relative percent difference (MARPD) was < 7% for the 8 co-located
samples.

Figure 19. Mean absolute relative percent difference (MARPD) between CSN samplers
with (orange) and without (blue) a cyclone at the inlet

Duke Forest (5.7% MARPD)

Gainesville (6.6% MARPD)

11/9/17 11/12/17 11/15/17 11/17/17 11/9/17 11/12/17 11/15/17 11/17/17
¦ Original Config bNo Cyclone/lmpactor

31


-------
8.0 Conclusions

Even though final results of the study are presented in this report, there are several outstanding
questions that should be explored to understand the poor NH+4 capture efficiency of the nylon
filter (Yu etal., 2006). Answering these questions will reduce uncertainty in the reported CSN
NH4 concentrations. Also, analyzing meteorological data may help in determining why the
retention of NH+4 was not consistent over the sample period or across the sites.

In addition, different types of cellulose filters were used for the CSN and IMPROVE samplers.
The CSN 47 mm cellulose filters were procured from Whatman and listed as Type 41, which is
rated for coarse particle retention. The Whatman 47mm cellulose filter is also used by
CASTNET, where it is impregnated with potassium carbonate (K2CO3) for SO2 sample
collection. The IMPROVE 37 mm filters used were procured from SKC Omega Specialty
Division and listed as Type 40, which is rated as "medium." During the procurement process, a
37 mm Type 41 cellulose filter was not available for use in the IMPROVE system. The ratings
for Type 40 and 41 indicate there is a 98% retention of 8 nm and 20 nm particles, respectively.
Acid-impregnation procedures were identical. It is recommended that the two types of cellulose
filters be investigated further in a laboratory setting (chamber study) to determine if the different
ratings have any effect on the NH+4 collection efficiency.

Overall, the acid-impregnated filters performed well for NHX in the SuperSASS and IMPROVE
PM sampler at the Duke Forest site. The median NHX concentrations from the ADSFPS, CSN,
and IMPROVE from Duke Forest were reasonably close (0.59, 0.57, 0.68 ng rrr3, respectively).
The performance of the acid-impregnated filters at the Gainesville site was poor, with the filters
significantly underestimating the NHX concentrations. The median NHX concentrations from the
reference method (ADSFPS) and the CSN and IMPROVE samplers from the Gainesville site
were 0.66, 0.48, 0.51 ng rrr3, respectively. Besides the CSN pump failure, the cause of the poor
performance from the Gainesville site is unclear. Gainesville does have high humidity and
morning dew which could have impacted the sampling. We conclude that the acid-impregnated
filter method is not suitable for measuring NHX in environments similar to Gainesville and an
additional evaluation is needed to fully understand the negative biases and potential resolutions.
It is recommended that air quality and meteorological modeling results (i.e. CMAQ, WRF) be
used to determine if biases are greater during specific meteorological conditions to help
characterize the uncertainty and identify potential sampling locations for future studies. If further
evaluation can satisfy questions about the meteorological impacts to sample retention and
resolve the differences in pore size and associated particle retention, a next step could be to
deploy the acid-impregnated cellulose filters at a small subset of both CSN and IMPROVE sites
to determine if adding NHX as a measured parameter is feasible across a national network. This
would provide a better assessment across different regions, land use types, and climates. If a
small pilot is successful, deploying across existing networks would provide a novel, cost-

32


-------
effective method for measuring NHX with little burden to the monitoring agencies while filling a
significant gap in our understanding of the nitrogen budget.

33


-------
9.0 References

Carlton, A. G. and B. J. Turpin, 2013. Particle partitioning potential of organic compounds is
highest in the Eastern US and driven by anthropogenic water. Atmospheric Chemistry
and Physics 13, 10203-10214.

Carlton, A. G., R. W. Pinder, P. V. Bhave and G. A. Pouliot, 2010. To what extent can biogenic
SOA be controlled? Environmental Science & Technology 44, 3376-3380.

Chen, X., D. Day, B. Schichtel, W. Malm, A. K. Matzoll, J. Mojica, C. S. McDade, E. D.

Hardison, D. L. Hardison, S. Walters, M. Van De Water, J. L. Collett Jr. 2014. Seasonal
ambient ammonia and ammonium concentrations in a pilot IMPROVE NHX monitoring
network in the western United States. Atmospheric Environment 91, 118-126.

Du, E., de Vries, W., Galloway, J. N., Hu, X., Fang, J., 2014. Changes in wet nitrogen deposition
in the United States between 1985 and 2012. Environmental Research Letters 9,

095004.

Ellis, R. A., D. J. Jacob, M. P. Sulprizio, L. Zhang, C. D. Holmes, B. A. Schichtel, T. Blett, E.
Porter, L. H. Pardo and J. A. Lynch, 2013. Present and future nitrogen deposition to
national parks in the United States: critical load exceedances. Atmospheric Chemistry
and Physics 13, 9083-9095.

Erisman, J.W., A. Bleeker, J. Galloway, M.S. Sutton. 2007. Reduced nitrogen in ecology and the
environment. Environmental Pollution 150, 1, 140-149.

Jickells T, A. R. Baker, J. N. Cape, S. E. Cornell, E. Nemitz. 2013. The cycling of organic

nitrogen through the atmosphere. Philosophical Transactions of the Royal Society of
London B: Biological Sciences, 368, 20130115.

Li, Y., Schichtel, B., Walker, J. T., Schwede, D. B., Chen, X. I., Lehmann, C., Puchalski, M.,
Gay, D., Collett, J. L. Jr., 2016. The Increasing Importance of Deposition of Reduced
Nitrogen in the United States. Proceedings of the National Academy of Sciences, 113,
5874-5879.

Martin, M. V., C. L. Heald, J. F. Lamarque, S. Tilmes, L. K. Emmons and B. A. Schichtel, 2015.
How emissions, climate, and land use change will impact mid-century air quality over the
United States: a focus on effects at national parks. Atmospheric Chemistry and Physics
15, 2805-2823.

34


-------
Pye, H. O. T., Nenes, A., Alexander, B., Ault, A. P., Barth, M. C., Clegg, S. L., Collett Jr., J. L.,
Fahey, K. M., Hennigan, C. J., Herrmann, H., Kanakidou, M., Kelly, J. T., Ku, l.-T.,
McNeill, V. F., Riemer, N., Schaefer, T., Shi, G., Tilgner, A., Walker, J. T., Wang, T.,
Weber, R., Xing, J., Zaveri, R. A., and Zuend, A., 2020. The acidity of atmospheric
particles and clouds, Atmospheric Chemistry and Physics, 20, 4809-4888.

Silvern, R. F., D. J. Jacob, P. S. Kim, E. A. Marais, J. R. Turner, P. Campuzano-Jost, J. L.

Jimenez. 2017. Inconsistency of ammonium-sulfate aerosol ratios with thermodynamic
models in the eastern US: a possible role of organic aerosol. Atmospheric Chemistry
and Physics, 17, 5107-5118.

Thermo/Dionex Corporation, 2009. ICS-1600 Ion Chromatography System Operator's Manual,
Rev. 01, Document No. 065290. March 2009

U.S. EPA, 1997. Compendium Method IO-4. 2: Determination of Reactive Acidic and Basic
Gases and Strong Acidity of Fine-Particles (<2. 5 |jm). EPA/625/R-96010a; U.S. EPA:
Cincinnati, OH.

U.S. EPA, 2016. Air Pollutant Emissions Trends Data, https://www.epa.gov/air-emissions-
inventories/air-pollutant-emissions-trends-data

U.S. EPA, 2018. Office of Air Quality Planning and Standards (OAQPS). Quality Assurance
Project Plan: Improving characterization of reduced nitrogen at IMPROVE and CSN
monitoring sites Version 1.0 Revision 2.

U.S. EPA, 2019. Clean Air Status and Trends Network (CASTNET) Quality Assurance Project
Plan (QAPP) Revision 9.2. Prepared by Wood Environment & Infrastructure, Inc. for U.S.
Environmental Protection Agency (EPA), Office of Air and Radiation, Clean Air Markets
Division, Washington, DC. Contract No. EP-W-16-015. Gainesville, FL.
https: //j ava. e pa. g ov/castn et/docu m e nts. do.

Yu, X., Lee, T., Ayresa, B., Kreidenweis, S., Malm, W., Collett, J., 2006. Loss of fine particle
ammonium from denuded nylon filters. Atmospheric Environment 40 4797-4807.

35


-------
10.0 List

of Acronyms

and Abbreviations

ADFPS	annular denuder filter pack system

AESMD	Atmospheric and Environmental Systems Modeling Division

AQAD	Air Quality Assessment Division

ARD	Air Resources Division

BLK	blank

BS	blank spike

°C	degrees Celsius

CAMD	Clean Air Markets Division

CASTNET	Clean Air Status and Trends Network

CCV	continuing calibration verification samples

CEMM	Center for Environmental Measurement and Modeling

CMAQ	Community Multiscale Air Quality Model

CPHEA	Center for Public Health and Environmental Assessment

CSN	Chemical Speciation Network

Dl	deionized

DQI	Data Quality Indicator

DQO	Data Quality Objective

EPA	U.S. Environmental Protection Agency

H3PO3	phosphorous acid

HNO3	nitric acid

IC	ion chromatography

IMPROVE	Interagency Monitoring of Protected Visual Environments

K+	potassium ion

K2CO3	potassium carbonate

km	kilometer

L	liter

LIMS	laboratory information management system

Lpm	liters per minute

m	meter

MAD	mean absolute difference

MARPD	mean absolute relative percent difference

MFC	mass flow controller

|jg	microgram

|jg/m3	micrograms per cubic meter

mg	milligram

MgO	magnesium oxide

ml_	milliliter

N	nitrogen

Na+	sodium ion

36


-------
Na2CC>3

sodium carbonate

NADP

National Atmospheric Deposition Program

nh3

ammonia

nh;

ammonium

NIST

National Institute of Standards and Technology

N0X

nitrogen oxides

no3

particulate nitrate

NPS

National Park Service

NTN

National Trends Network

OAP

Office of Atmospheric Programs

OAQPS

Office of Air Quality Planning and Standards

OAR

Office of Air and Radiation

ORD

Office of Research and Development

PHESD

Public Health and Environmental Systems Division

PM2.5

mass of particles with a mean diameter of less than 2.5 |jm

ppb

parts per billion

ppm

parts per million

QA

quality assurance

QAPP

Quality Assurance Project Plan

QC

quality control

RL

reporting limit

RPD

relative percent difference

sox

oxides of sulfur

S02

sulfur dioxide

SO'"

particulate sulfate

SOP

standard operating procedures

SQL

structured query language

SRM

standard reference material

SVOC

semi-volatile organic compounds

VOC

volatile organic compounds

Wood

Wood Environment & Infrastructure Solutions, Inc.

WRF

Weather Research and Forecasting Model

37


-------
11.0	Appendix A: Preparation and Extraction of Sampling Media

11.1	ADFPS Phosphorous Acid Coated Denuder

Denuders are coated with 2% phosphorous acid (H3P03), extracted with 10 mL deionized water,

and extracts are analyzed for NH+4 by ion chromatography. ADFPS H3P03 denuders are

prepared and extracted as described below.

Preparation of H^PO; (PA) Solution (2%):

1.	Add 10 grams of H3P03 to a 500 volumetric flask and use 50 mL of Dl to rinse weigh
boat into the flask.

2.	Dilute to volume with methanol.

3.	Wear a face mask and work quickly to avoid adsorption of ammonia from the air.

Washing:

1.	This procedure is performed in the washroom where all required components are
available.

2.	Rinse each cap with deionized water (Dl), filling, swishing, and dumping three times.

3.	Using a properly labeled squeeze bottle, rinse the denuder with methanol (MeOH) over
the labeled denuder waste container.

4.	Rinse the denuder, running Dl through the channels for -10 seconds on each side.

Rinse the outside threads of both ends.

5.	Put the denuder into a large wash tub and fill with Dl.

6.	Soak for -30 minutes.

7.	Repeat steps 4 through 6 two times, for a total of three soaking periods.

8.	Shake water out of the denuder and caps and allow drying in an aluminum foil lined heat
resistant tray covered with aluminum foil in the drying oven about two hours at 60° C.

Coating and Drying:

1.	This procedure is performed in the washroom where all required components are
available. Wear a face mask and gloves.

2.	Place a Pyrex dish containing about % inch citric acid next to where you will be working
in the hood.

3.	Pipette 10ml_ of 2% phosphorous acid coating solution into the denuder with bottom cap
attached. The bottom of the denuder is the end that the inner quartz tube is recessed
about % inch from the end of the aluminum denuder tube.

4.	Secure top cap of the denuder and invert 20 times. The top of the denuder is the end
that the inner quartz tube is recessed about one inch from the end of the aluminum
denuder tube.

5.	Remove the top cap and pour excess coating solution into the labeled denuder waste
container.

38


-------
6.	Screw the denuder securely into the URG drying manifold so the direction of drying flow
is downward; this will allow gravity to remove excess solution from the denuder.

7.	Dry for approximately 20 minutes with 5 liters per minute nitrogen purge. The
'honeycomb' quartz inside the URG denuder will cloud when dry.

8.	Remove the denuders from the drying manifold and install both caps. A blank red label
should be attached near the top end to signify H3PO3 coating. Write the logbook/page ID
on the label and place the denuder in a resealable plastic bag. Refrigerate until needed.

9.	Two lab blanks should be pulled for extraction each coating session and/or each batch
of coating solution made.

Extraction:

1.	This procedure is performed in the washroom where all required components are
available. Wear a face mask and gloves.

2.	Place a Pyrex dish containing % inch of citric acid in the hood where you are working.

3.	Ensure the bottom cap of the denuder is attached firmly and remove the top cap.

4.	Pipette 10 ml_ Dl water into the denuder and cap tightly. Invert the denuder 20 times.

5.	Tap the top cap a couple times. Unscrew the top cap and set aside.

6.	Decant the extract carefully from top end of the denuder into the sample extract bottle.

7.	Remove the red Lab ID label from the denuder and attach to the sample extract bottle.

8.	Refrigerate until analysis.

11.2 ADFPS Sodium Carbonate Coated Denuder

Denuders are coated with 1% sodium carbonate (Na2CC>3). In this study, the Na2CC>3 denuder

serves only to remove HN03 from the sample stream. For that reason, the Na2CC>3 denuder is

not quantitatively extracted and analyzed.

Na?C03 Coating Solution (1%):

1.	Dissolve 5 g of Na2CC>3 and 5 g of glycerol in 250 ml_ of Dl in a 500ml_ volumetric flask.

2.	Dilute to volume with MeOH.

Washing the Denuder:

1.	This procedure is performed in the washroom where all required components are
available.

2.	Rinse each cap using deionized water (Dl), filling, swishing, and dumping three times.

3.	Using a squeeze bottle rinse the denuder with methanol (MeOH) over a labeled denuder
waste container.

4.	Rinse the denuder, running Dl through the channels for -10 seconds on each end.

Rinse the outside threads as well.

5.	Put the denuder into a large wash tub and fill with Dl.

6.	Allow to soak for -30 minutes.

39


-------
7.	Repeat steps 4-6 two times, for a total of three soaking periods.

8.	Shake water out of the denuder and caps and allow drying in an aluminum foil lined heat
resistant tray, covered with aluminum foil in the drying oven, for about two hours at

60° C.

Coating and Drying URG Na?C03 Denuder:

1.	This procedure is performed in the washroom where all required components are
available.

2.	Pipette 10ml_ of 1% Na2CC>3 coating solution into the denuder with bottom cap attached.
The bottom of the denuder is the end that the inner quartz tube is recessed about % inch
from the end of the aluminum denuder tube.

3.	Secure top cap of the denuder and invert 20 times. The top of the denuder is the end
that the inner quartz tube is recessed about one inch from the end of the aluminum
denuder tube.

4.	Remove the top cap and pour excess coating solution into a labeled denuder waste
container.

5.	Remove the bottom cap and place the denuder and caps in a desiccator to dry
overnight. The desiccant should be fresh.

6.	Remove the denuders from the drying manifold and install both caps. A blank green
label should be attached near the top end to signify Na2C03 coating. Write the
logbook/page ID on the label. Place the denuder in a clean resealable plastic bag and
refrigerate until needed.

11.3 Phosphorous Acid Impregnated Cellulose Filters

Improve (37 mm, SKC cellulose type 40, #225-18A) and CSN (47 mm, Whatman cellulose type

41, #1441-047) filters are coated with 3% phosphorous acid (H3P03), extracted with 20 ml_

deionized water, and extracts are analyzed for NH+4 by ion chromatography. H3P03 impregnated

filters are prepared and extracted as described below.

H3PO3 (PA) Solutions:

1. 3% PA - Dissolve 30 g of H3P03 in 100 ml_ of Dl in a 1L volumetric flask. Bring to
volume with methanol.

Impregnating Cellulose Filters with Phosphorous Acid:

1.	Gloves and a face mask must be worn whenever working with the filters and the filters
handled in a hood. In addition, a Pyrex dish containing about % inch of citric acid should
be placed near the area where you are working with the filters.

2.	Place filters into a wide-mouth polypropylene bottle and cover with 3% PA solution and
seal.

3.	Sonicate for 30 minutes. Drain the PA solution into an appropriate waste receptacle.

40


-------
4.	Add Dl to the bottle to cover the filters and sonicate again for 30 minutes. Drain the Dl.
Repeat this procedure two more time for a total of 3 Dl rinses.

5.	Cover the filters with 3% PA and sonicate again for 30 minutes. Drain the PA solution.

6.	Cover the hood work area with aluminum foil and wipe with a towel soaked with the 5%
PA solution.

7.	Tear another piece of foil and fold to fit inside the modified nitrogen desiccator. Wipe the
foil with a fresh towel soaked with the 5% PA solution. Also wipe the inside top of the
desiccator. Clean gloved fingertips, forceps and petri slide bottoms with the 5% PA
solution.

8.	Place each filter in a cleaned petri slide (do not cover) and place the slide in a
desiccator.

9.	When all of the filters are in the desiccator, attach the nitrogen gas source to the
modified desiccator with flow control. Turn on nitrogen gas flow to ~5 LPM. Close hood
sash completely.

10.	Check for dryness in 6 hours; continue the nitrogen purge overnight (if necessary) until
the filters are visibly dry.

11.	When the filters are dry, clean the tops of the petri slides with a paper towel soaked in
the 5% PA solution. Working quickly, place the tops on the slides. Arrange the slides on
the plastic tray and place in a marked sealable bag.

12.	Place a paper towel soaked in 5% PA solution in the bottom of the plastic storage
container. Put the bag containing the tray of impregnated filters on top of the soaked
paper towel and seal the container.

13.	Refrigerate the container.

14.	Two impregnated filters should be acceptance tested with each impregnating procedure.

Loading, Unloading and Extracting Exposed Filters:

1.	Gloves and a face mask must be worn whenever working with the filters. In addition, a

Pyrex dish or large weigh boat containing about % inch of citric acid should be placed

near the area where you are working with the filters.

2.	Loading PA filters:

a.	Cover the lab bench with aluminum foil and wipe with a fresh towel soaked with a
5% PA solution. Clean gloved fingertips, forceps and the inside of the sample
bottles and tops with the 5% PA solution.

b.	Use clean forceps to remove the filter from the petri slide and place in the filter
holder or filter pack. Attach corresponding Lab ID sample labels.

c.	A pair of lab blanks should be pulled weekly for extraction.

3.	Unloading exposed filters:

a. Cover the lab bench with aluminum foil for cleanliness and wipe with a fresh
towel soaked with a 5% PA solution. Clean gloved fingertips, forceps and the
inside of the sample bottles and tops with the 5% PA solution.

41


-------
b.	Use clean forceps to remove the filter from the filter holder or filter pack by the
edges; fold using only the forceps so that the sampled side is inward and place in
a 30ml_ Nalgene bottle. Cap tightly. This will serve as the extraction bottle. Attach
corresponding Lab ID sample extract labels.

c.	Place the sample bottles in sealable bag and store in a freezer until extraction.
4. Extracting exposed filters:

a.	Remove the sample bottles from the freezer and allow to equilibrate to room
temperature.

b.	Using the Teflon extraction solution, pipette 20 ml_ into each filter extraction
bottle. Cap tightly.

c.	Sonicate for 45 minutes.

d.	Prepare a method blank and cation blank spike with each extraction.

11.4 Nylon Filters

ADFPS and CSN (Channel 2) samplers employ 47 mm nylon filters (Pall Nylasorb #66509,
1 |jm) for collection of NH+4 aerosol. Filters are extracted with 20 ml_ deionized water and
extracts are analyzed for NH+4 by ion chromatography. Filters are prepared and extracted as
described below.

Loading, Unloading and Extracting Exposed Filters:

Before assembling modules with clean filters, examine filters for tears, holes, etc. If any are
damaged, discard the filter. Wear gloves when handling filters and modules. Use forceps when
handling the filters.

1.	Loading nylon filters:

a.	Use clean forceps to place in the sampling module. Attach corresponding Lab ID
sample labels.

b.	A pair of lab blanks should be pulled weekly for extraction.

2.	Unloading exposed filters:

a.	Use clean forceps to remove the filter from the sampling module and place in a
30mL Nalgene bottle. Cap tightly. This will serve as the extraction bottle. Attach
corresponding Lab ID sample extract labels.

b.	Store the sample bottles in the cold room until extraction.

3.	Extracting exposed filters:

a.	Remove the sample bottles from the freezer and allow to equilibrate to room
temperature.

b.	Using the nylon extraction solution, pipette 20 mL into each filter extraction bottle.
Cap tightly.

c.	Sonicate for 60 minutes at 23-27°C. Monitor the temperature and add ice to the
sonicator to keep the temperature from exceeding 27°C and then overnight on a
shaker table at 1 Hz and 4°C.

d.	Prepare a method blank and cation blank spike with each extraction.

42


-------
12.0 Appendix B: Chemicals for IC Analysis

Reagent water: deionized (Dl) water of resistivity of 15 mega ohms (MQ) or greater derived from
mixed bed ion exchangers, activated carbon filters, and polishing exchangers. Water should contain
particles no larger than 0. 20 |jm.

Methanesulfonic acid (MSA), >99%, ACS reagent grade or better.

Cation concentrated eluentforCS16 (1. 0 N MSA): Dissolve 48. 05 g of 99% MSA in a final volume
of 500 ml. with Dl.

Cation working eluent solution (30mM MSA): Dilute 120 ml. of 1.0 N MSA concentrate to 4.0 L with
deionized water.

Blank spike solutions (BS). Cation blank spike solution purchased from High Purity Standards
(HPS). A Certificate of Analysis and an expiration date is provided with each lot. The BS is used to
verify the accuracy of the extraction.

Individual Cation Calibration stock solutions (1000 |jg/ml_) are purchased as NIST traceable
solutions. A Certificate of Analysis and an expiration date will be provided with each stock.

Lithium stock solution (100 jjg/mL) is purchased as NIST-traceable solution. A Certificate of Analysis
and an expiration date will be provided with each stock.

10 |jg/ml_ intermediate cation calibration solution. 5 ml_s of each 1000 |jg/ml_ stock standard are
added to a 500 ml. volumetric flask and diluted to volume with Dl. The intermediate solution will have
an expiration date 6 months from preparation (or the expiration date of an individual stock if sooner).

100 |jg/ml_ intermediate cation calibration solution. 20 ml_s of each 1000 |jg/ml_ stock standard are
added to a 200 ml. volumetric flask and diluted to volume with Dl. The intermediate solution will have
an expiration date 6 months from preparation (or the expiration date of an individual stock if sooner).

The working curve will be prepared by adding the volumes listed in Table 7 below to separate 500
ml. volumetric flasks. The working curve will have an expiration date one month after preparation.

43


-------
Table 7. Cation concentration curves (|jg/mL)

Cation
Standard

Volume of Intermediate Calibration
Solutions (ml_)

Working Curve Concentrations
(HQ/mL)

10 |jg/ml_

100 |jg/ml_

100 |jg/ml_ Li

Li+

Na+, NH+4, K+

STD 1





5. 0

1. 0

0. 0

STD 2

1. 0



5. 0

1. 0

0. 02

STD 3

2. 0



5. 0

1. 0

0. 04

STD 4

5. 0



5. 0

1. 0

0. 10

STD 5

25. 0



5. 0

1. 0

0. 50

STD 6



5. 0

5. 0

1. 0

1. 0

STD 7



25. 0

5. 0

1. 0

5. 0

A minimum of five points shall be used for each calibration curve. The concentrations of the
daily curve are listed in Table 7 in units of micrograms (|jg) per ml_.

Thermo Cation I Standard. A Certificate of Analysis and an expiration date is provided with each
lot.

Cation Control standards used as continuing calibration verification (CCV) solutions are
prepared by diluting 5 ml_ of the Thermo Cation I stock to a 500 ml_ final volume with Dl. The
working CCV will have an expiration of 6 months from preparation (or the expiration date of an
individual stock if sooner). A CCV is used to verify accuracy.

Cation Standard Reference Material (SRM) reference solution is purchased as a ready-to-use
NIST-traceable standard solution containing analytes of interest with a Certificate of Analysis
and an expiration date.

44


-------
13.0 Appendix C: Chain of Custody Forms

Annular Denuder System

I	ftPS Channel 1

Installed:

I>#Se: 	/	I	Time: 	 luiiuls:

Removed

Date: / ! Time: IniliaU:
TpW Flow Volum#' 	rn1

I	APS Channel 2

Instated:

Dste	_ Time: 	 Inliib'

Removed:

Dale: I in:.: lailia'.v
Total Rpuu Volume! 	 m3

Blank ate

label

Blank an



•n2x4

Blank

3rea for label

COMMENTS:

UB U« WLTi
laii icc»i nrrt ifT

nxm t i

	I

45


-------
IMPROVE

Dale.



Time:

Simp*?

Date
+•«*+*•»«»

5/35/2017

B.QH.'ZQl?

NfliOH

Operator fnili«h

INITIAL READINGS

Opt	Ori

2		

4		

b\u\ B

«• • » > A« • ¦ ¦ » »»» ¦ »in»« i

.2x4

Blank area for label

M "

Blank arto-for label

Dale: 		Forta:. 	

Opcrilix Initials

FINAL fi;EADIf4GS

Cyt	on	er

»*i «>•*•«* t*n«» + 4U

¦ » « «-aa»R^aa*«	**««¦¦»<

.2x4. -



Blank arfea tai label







- .. 2*4 ... -
Blank arta for label

COMMENTS:

LAB USt 
-------
CSN SypcrSASS

Installed:

I>«B.	( f HlllC.	lliJUJli.

Removed:

Oali:: 	i1	/	Hint!: 	 Initial*:

{.'hanDcl
Nil lilllci

Rim Time

Snmjtfv Volim*1

1



Ul*

*





J



ro'

4



m-

5





*

7 FB

0

m}

0 Ol'

sris

0

0 n'

2*4

Dlank ar6
-------
shipping 1 emper»!~jre Loa








-------
14.0	Appendix D: Characterization of NH3 Breakthrough on Phosphorous Acid
Coated Annular Denuders

14.1	Project description and objectives

This effort builds on the primary project described above to assess the performance of NHX
collection by IMPROVE and CSN samplers in humid environments by comparison to a
reference ADFPS sampler. During that study, the ADFPS system was discovered to have a
certain degree of NH3 breakthrough from its primary acid coated denuders for samples collected
at both sites. Given the objective and purpose of the previous project, only limited extra
sampling was conducted to address such suspected breakthrough issues with the ADFPS. In
order to fully evaluate and address the potential impaired NH3 collection efficiency by the
annular denuders under humid conditions, additional sampling is proposed to address such
issues. The objective of this study is to extend previous work to further evaluate the
performance and NH3 collection efficiency by ADFPS denuders under warmer and more humid
environments across seasons.

Scientific approach

The proposed work deviates from the parent QAPP in three ways:

•	Measurements will be conducted at the AIRS site adjacent to the EPA campus rather
than at Duke Forest, NC or Gainesville, FL.

•	An additional acid coated denuder will be added to the ADFPS sampling train to quantify
NH3 breakthrough on the primary acid coated denuder.

•	Samples will be analyzed at EPA rather than the AM EC laboratory.

Other than these details, sampling and analytical methods, SOPs, and QA/QC procedures
described in the parent QAPP will be followed.

14.2 Field site

The sampling will be conducted at the AIRS site located on the EPA campus in Research
Triangle Park, North Carolina (35.8897 latitude, -78.8747 longitude).

49


-------
Figure 20. EPA campus study site AIRS

14.3 Annular Denuder/Filter Pack System

We propose to evaluate the ADFPS NH3 denuder collection efficiency by using the URG ADFPS
samplers with two NH3 denuders in series. The configuration of the sampling train is as follow:

•	A 1% sodium carbonate (Na2C03)-coated annular denuder for scrubbing gaseous HN03

•	Two 1% hhPCh-coated annular denuders in series for collection of gaseous NH3

•	A 2-stage filter pack containing a nylon filter for collection of particulate NH!,

•	A backup 1% H3P03-coated annular denuder for collecting gaseous NH3 that may
volatilize from the nylon filter

Three weeks of 24hr (from 9am to 9am EST) samples will be collected each season covering
different meteorological conditions at the NC EPA site to provide insight into NH:, denuder
performance under warm and humid conditions. Approximately 60 or more samples will be
collected starting summer 2018 through winter covering three seasons, which include warmest
and humid summer periods as well as relatively dry and cold winter conditions.

50


-------
14.4	Chemical analysis

Denuder and filter extracts are analyzed for NH+4, N03 and S024 by ORD NRMRL lab using ion
chromatography (IC, Dionex model ICS-2100, Thermo Scientific, Waltham, MA). The IC is
equipped with guard (lonPac 2mm AG23) and analytical columns (AS23) for anions. The
samples are analyzed using an isocratic eluent mix carbonate/bicarbonate (4.5/0.8mM) at a flow
rate of 0.25ml_/min. Cations are analyzed by Dionex lonPac 2mm CG12 guard and CS12
analytical columns; separations are conducted using 20mM methanesulfonic acid (MSA) as
eluent at a flow rate of 0.25ml_/min. Multi-point (>5) calibration is conducted using a mixture
prepared from individual inorganic standards (Inorganic Ventures, Christiansburg, VA). A mid-
level accuracy check standard is prepared from certified standards mix (AccuStandard, New
Haven, CT) for quality assurance/quality control purposes.

14.5	Collection efficiency calculations

NH3 collection efficiency calculations are based on the assumption that any breakthrough or NH3
not captured by first primary denuder will be secured by second primary denuder in line; hence,
the collection efficiency is calculated by following equation:

JVi

x 100%

N1 + N:

Where r| is NH3 collection efficiency, Ni is the NH3 captured by first primary denuder and N2 is
the NH3 captured by second primary denuder.

51


-------
15.0 Appendix E: Validation Log for Study Samples

Sample ID

Site

Method

Sample
Date

Flag

Comment

Date
Applied

1722040-01

D

ADFPS

5/31/17

13

concentration determined as outlier
during data review

10/23/18

1725049-01

D

ADFPS

6/21/17

11

quick connect fitting not completely
sealed, no air flow

8/29/17

1725049-02

D

ADFPS

6/21/17

11

quick connect fitting not completely
sealed, no air flow

8/29/17

1725034-01

D

ADFPS

6/21/17

11

quick connect fitting not completely
sealed, no air flow

8/29/17

1733034-01

D

ADFPS

8/17/17

15

temperature thermocouple maxed
out, pushed flow rate down, reported
it was ~10 Ipm but was actually ~7
Ipm

9/20/17

1733035-01

D

ADFPS

8/17/17

15

temperature thermocouple maxed
out, pushed flow rate down, reported
it was ~10 Ipm but was actually ~7
Ipm

9/20/17

1733035-02

D

ADFPS

8/17/17

15

temperature thermocouple maxed
out, pushed flow rate down, reported
it was ~10 Ipm but was actually ~7
Ipm

9/20/17

1734066-01

D

ADFPS

8/26/17

15

MFC readout on pump box 8.9 Ipm,
previous week 10.1 Ipm, likely
blockage, pressure low, gas meter
high, actual flow likely normal

9/20/17

1734067-01

D

ADFPS

8/26/17

15

MFC readout on pump box 8.9 Ipm,
previous week 10.1 Ipm, likely
blockage, pressure low, gas meter
high, actual flow likely normal

9/20/17

1734067-02

D

ADFPS

8/26/17

15

MFC readout on pump box 8.9 Ipm,
previous week 10.1 Ipm, likely
blockage, pressure low, gas meter
high, actual flow likely normal

9/20/17

1737034-01

D

CSN

9/13/17

13

concentration determined as outlier
during data review

10/23/18

1738020-01

D

CSN

9/19/17

13

concentration determined as outlier
during data review

10/23/18

1739014-01

D

CSN

9/25/17

13

concentration determined as outlier
during data review

10/23/18

1742016-01 D IMPROVE 10/16/17 13 flow volume not nominal,	10/23/18

concentration determined as outlier
during data review

52


-------
Sample ID

Site

Method

Sample

Flag

Comment

Date







Date





Applied

1745059-01

D

CSN

11/9/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746078-01

D

CSN

11/12/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746080-01

D

CSN

11/15/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746035-01

D

IMPROVE

11/15/17

11

no flow

2/22/18

1746082-01

D

CSN

11/18/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746059-01

D

IMPROVE

11/18/17

11

no flow

2/22/18

1721041-01

G

ADFPS

5/25/17

11

pump filter old, clogged

8/29/17

1721042-01

G

ADFPS

5/25/17

11

pump filter old, clogged

8/29/17

1721042-02

G

ADFPS

5/25/17

11

pump filter old, clogged

8/29/17

1722030-01

G

ADFPS

5/28/17

11

pump filter old, clogged

8/29/17

1722031-01

G

ADFPS

5/28/17

11

pump filter old, clogged

8/29/17

1722031-02

G

ADFPS

5/28/17

11

pump filter old, clogged

8/29/17

1722049-01

G

IMPROVE

5/31/17

11

pump controller in standby

8/29/17

1722066-01

G

CSN

6/3/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1722067-01

G

CSN

6/3/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1723031-01

G

ADFPS

6/6/17

11

pump filter old, clogged

9/1/17

1723032-01

G

ADFPS

6/6/17

11

pump filter old, clogged

9/1/17

1723032-02

G

ADFPS

6/6/17

11

pump filter old, clogged

9/1/17

1723026-01

G

CSN

6/6/17

13

concentration determined as outlier
during data review

10/23/18

1723038-01

G

CSN

6/9/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1723039-01

G

CSN

6/9/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1724018-01

G

CSN

6/12/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

53


-------
Sample ID

Site

Method

Sample
Date

Flag

Comment

Date
Applied

1724019-01

G

CSN

6/12/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1724035-01

G

CSN

6/15/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1724036-01

G

CSN

6/15/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1724037-01

G

IMPROVE

6/15/17

14

possible switch with 6/18

8/29/17

1725024-01

G

CSN

6/18/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1725025-01

G

CSN

6/18/17

12

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
nominal

11/17/17

1725026-01

G

IMPROVE

6/18/17

SI

possible switch with 6/15

8/29/17

1725037-01

G

CSN

6/21/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1725053-01

G

CSN

6/21/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1725063-01

G

CSN

6/24/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1725070-01

G

CSN

6/24/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

11/17/17

1726018-01

G

CSN

6/27/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

9/1/17

1726029-01

G

CSN

6/27/17

13

ambient temperature sensor

9/1/17

inoperative, controlled flow rates to

54


-------
Sample ID

Site

Method

Sample

Flag

Comment

Date







Date





Applied

1726035-01 G CSN

1726040-01 G CSN

1727015-01 G CSN

1727020-01 G CSN

6/30/17 13

6/30/17 13

7/3/17 13

7/3/17 13

1727043-01 G ADFPS 7/6/17

1727044-01 G ADFPS 7/6/17

1727044-02 G ADFPS 7/6/17

1727040-01 G CSN

1727041-01 G CSN

1728024-01 G CSN

7/9/17

15

15

15

7/6/17 13

7/6/17 13

13

be extremely inaccurate, flow volume i
not nominal

ambient temperature sensor	11/17/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

ambient temperature sensor	11/17/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

ambient temperature sensor	11/17/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

ambient temperature sensor	11/17/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume |
not nominal

pressure at MFC much lower than	9/20/17

nominal indicating pump working

harder than expected, flow steady

but low throughout run, possible

obstruction

pressure at MFC much lower than	9/20/17

nominal indicating pump working

harder than expected, flow steady

but low throughout run, possible

obstruction

pressure at MFC much lower than	9/20/17

nominal indicating pump working

harder than expected, flow steady

but low throughout run, possible

obstruction

ambient temperature sensor	8/29/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

ambient temperature sensor	8/29/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume i
not nominal

ambient temperature sensor	8/29/17

inoperative, controlled flow rates to
be extremely inaccurate, flow volume :
not nominal

55


-------
Sample ID

Site

Method

Sample
Date

Flag

Comment

Date
Applied

1728025-01

G

CSN

7/9/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

8/29/17

1728043-01

G

CSN

7/12/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

8/29/17

1728044-01

G

CSN

7/12/17

13

ambient temperature sensor
inoperative, controlled flow rates to
be extremely inaccurate, flow volume
not nominal

8/29/17

1728045-01

G

IMPROVE

7/12/17

14

possible switch with 7/15

8/29/17

1728057-01

G

CSN

7/15/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1728058-01

G

CSN

7/15/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1728059-01

G

IMPROVE

7/15/17

SI

possible switch with 7/12

8/29/17

1729026-01

G

CSN

7/18/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1729027-01

G

CSN

7/18/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1729038-01

G

CSN

7/21/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1729039-01

G

CSN

7/21/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

8/29/17

1730019-01

G

CSN

7/24/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

9/20/17

1730020-01

G

CSN

7/24/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

9/20/17

1731027-01

G

CSN

7/30/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

9/20/17

1731028-01

G

CSN

7/30/17

14

outlier, suspect contamination
possibly from insect activity,
invalidated pending anion analysis

9/20/17

56


-------
Sample ID

Site

Method

Sample

Flag

Comment

Date







Date





Applied

1732033-01	j G j ADFPS j 8/8/17 j 15

1732034-01	G	ADFPS	8/8/17	15
1732034-02	G	ADFPS	8/8/17	15

1732040-01	G	CSN	8/11/17	14

1732041-01	G	CSN	8/11/17	14

1733022-01	G	ADFPS	8/14/17	SI

1733023-01	G	ADFPS	8/14/17	SI
1733023-02	G	ADFPS	8/14/17	SI
1733020-01	G	CSN	8/14/17	15

1736020-01	G	ADFPS	9/4/17	15

1736021-01	G	ADFPS	9/4/17	15
1736021-02	G	ADFPS	9/4/17	15

1736044-01	G	ADFPS	9/7/17	15

1736045-01	G	ADFPS	9/7/17	15
1736045-02	G	ADFPS	9/7/17	15
1736041-01	G	IMPROVE	9/7/17	S2

1737045-01	G	ADFPS	9/13/17	15

1737046-01	G	ADFPS	9/13/17	15

flow initially low and unsteady	I 9/20/17

through 1200, climbs to nominal
levels by 1800

flow initially low and unsteady	9/20/17

through 1200, climbs to nominal
levels by 1800

flow initially low and unsteady	9/20/17

through 1200, climbs to nominal
levels by 1800

outlier, suspect contamination	9/20/17

possibly from insect activity,

invalidated pending anion analysis

outlier, suspect contamination	9/20/17

possibly from insect activity,

invalidated pending anion analysis

flow volume slightly low,	11/17/17

concentration high

flow volume slightly low,	11/17/17

concentration high

flow volume slightly low,	11/17/17

concentration high

outlier, likely contamination,	9/22/17

suspected flow blockage possibly
from insect activity, nylon filter not
invalidated

low flow volume, suspected flow	9/22/17

blockage possibly from insect activity

low flow volume, suspected flow	9/22/17

blockage possibly from insect activity

low flow volume, suspected flow	9/22/17

blockage possibly from insect activity

low flow volume, suspected flow	11/17/17

blockage possibly from insect activity

low flow volume, suspected flow	11/17/17

blockage possibly from insect activity

low flow volume, suspected flow	11/17/17

blockage possibly from insect activity

concentration is outlier, other	11/17/17

methods also go up but do not match

low flow volume, suspected flow	11/17/17

blockage possibly from insect activity

low flow volume, suspected flow	11/17/17

blockage possibly from insect activity

57


-------
Sample ID

Site

Method

Sample
Date

Flag

Comment

Date
Applied

1737046-02

G

ADFPS

9/13/17

15

low flow volume, suspected flow
blockage possibly from insect activity

11/17/17

1738028-01

G

IMPROVE

9/19/17

SI

flow problem at sample removal but
15-minute flow looks OK

11/17/17

1738040-01

G

IMPROVE

9/22/17

SI

flow problem at sample removal but
15-minute flow looks OK

11/17/17

1739020-01

G

IMPROVE

9/25/17

SI

flow problem at sample removal but
15-minute flow looks OK

11/17/17

1740046-01

G

IMPROVE

10/4/17

SI

flow problem at sample removal but
15-minute flow looks OK

11/17/17

1740065-01

G

IMPROVE

10/7/17

SI

flow problem at sample removal but
15-minute flow looks OK

11/17/17

1743023-01

G

IMPROVE

10/22/17

11

pump failure

11/17/17

1743057-01

G

ADFPS

10/28/17

11

no flow, pump programmed
incorrectly

11/17/17

1743058-01

G

ADFPS

10/28/17

11

no flow, pump programmed
incorrectly

11/17/17

1743058-02

G

ADFPS

10/28/17

11

no flow, pump programmed
incorrectly

11/17/17

1743059-01

G

ADFPS

10/28/17

11

cap left on cyclone

11/17/17

1743060-01

G

ADFPS

10/28/17

11

cap left on cyclone

11/17/17

1743060-02

G

ADFPS

10/28/17

11

cap left on cyclone

11/17/17

1744027-01

G

IMPROVE

10/31/17

11

pump failure

11/17/17

1744039-01

G

IMPROVE

11/3/17

11

pump failure

11/17/17

1745021-01

G

ADFPS

11/6/17

11

cap left on cyclone

11/17/17

1745022-01

G

ADFPS

11/6/17

11

cap left on cyclone

11/17/17

1745022-02

G

ADFPS

11/6/17

11

cap left on cyclone

11/17/17

1745020-01

G

IMPROVE

11/6/17

11

pump failure

11/17/17

1745060-01

G

CSN

11/9/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1745042-01

G

IMPROVE

11/9/17

SI

auxiliary flow source used for first
half of sample runtime, then
replacement pump installed

11/17/17

1746079-01

G

CSN

11/12/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746081-01

G

CSN

11/15/17

Q1

duplicate CSN H3P03-impregnated
filter sampled without cyclone

1/15/18

1746067-01

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

1746068-01

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

58


-------
Sample ID

Site

Method

Sample
Date

Flag

Comment

Date
Applied

1746068-02

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

1746069-01

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

1746070-01

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

1746070-02

G

ADFPS

11/18/17

11

no flow, pump programmed
incorrectly

2/22/18

1746083-01

G

CSN

11/18/17

Q1

duplicate CSN H3P03-impregnated

1/15/18

filter sampled without cyclone

16.0 Disclaimer: Any opinions, findings, and conclusions or recommendations expressed in
this publication are those of the authors and do not necessarily reflect the views of the U.S.
EPA.

59


-------