EPA-600/2-78-067
April 1978
Environmental Protection Technology Series
SAMPLING AND ANALYTICAL METHODOLOGY
FOR ATMOSPHERIC PARTICULATE NITRATES
Final Report
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield. Virginia 22161.
-------
EPA-600/2-78-067
April 1978
SAMPLING AND ANALYTICAL METHODOLOGY
FOR ATMOSPHERIC PARTICULATE NITRATES
Final Report
by
Chester W. Spicer, Philip M. Schumacher,
John A. Kouyoumjian and Darrell W. Joseph
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
Contract No. 68-02-2213
Project Officers
James Mulik/Eva Wittgenstein
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences
Research Laboratory, U.S. Environmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
1i
-------
ABSTRACT
Environmental conditions that affect atmospheric particulate
nitrate sampling were identified, and improved sampling and analytical
procedures were developed. Evaluation of potential sources of error in high
volume nitrate sampling showed that artifact nitrate formation on commonly
used glass filter media was the most serious. Both laboratory and field
results demonstrated that high purity quartz filters provide a significant
improvement over glass filters and are easily substituted for glass filters
in traditional high volume sampling equipment. A sensitive, accurate and
rapid nitrate analytical procedure was developed using thermal decomposition
of nitrate and chemiluminescent detection of the decomposition products.
Ion chromatography was also investigated and found to be sensitive,
accurate, reproducible and rapid. Ion chromatography has the added
advantage of determining both nitrate and sulfate simultaneously.
This report was submitted in fulfillment of Contract No. 68-2-
2213 by Battelle's Columbus Laboratories under the sponsorship of the U.S.
Environmental Protection Agency. This report covers the period June 11,
1975 to December 10, 1977, and work was completed as of January 10, 1978.
iii
-------
CONTENTS
Abstract iii
Figures vi
Tables vii
Acknowledgment ix
1. Introduction 1
2. Phase I: Laboratory Studies of Artifact Nitrate Collection
on Filters ....!; 8
Gas-Filter Interactions 9
Breakthrough Experiments 21
Nitrogen Balance 23
Effect of Filter Saturation 25
Gas-Filtrate Interactions 31
Gas-Soot Interactions 31
Effect of Sampling Rate 33
Effect of Sampling Time 35
Phase I Summary 40
3. Phase II: Screening and Development of Nitrate Analysis
Methodology 41
Gas Sensing Electrode 41
Introduction 41
Experimental 42
Results and Disscussion 43
Thermal Decomposition/Chemiluminescence 44
Introduction 44
Experimental 45
Results and Discussion 46
Ion Chromatography 65
Background 67
-------
Determination of Nitrate in Atmospheric Samples
by Ion Chromatography 67
Experimental 68
Accuracy and Precision 68
Sensitivity 70
Phase II Summary 73
4. Phase III: Field Evaluation of Sampling Media for Nitrate
Collection 75
Experimental 75
Results and Discussion 79
Results of High Volume Collections - Nitrate ... 79
Results of High Volume Collection - Sulfate. ... 87
Low Volume Filter Sample Results - Nitrate .... 90
Low Volume Sample Results - Sulfate 95
Identity of the Nitrate Precursor(s) 97
Phase III Summary 102
5. Phase IV: Conclusions Regarding Sampling and Analysis of
Atmospheric Particulate Nitrate 104
References 106
Appendix
A. Experimental Conditions 109
VI
-------
FIGURES
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Phase I sampling and analysis apparatus
Close-up of sampling manifold and filter holders
Schematic of sampling systems for eight simultaneous
filters
Gas-filter interaction
Gas-filter interactions
Comparison of various filters exposed to N02
Comparison of various filters exposed to HNOg
Artifact nitrate as a function of NOg exposure
Artifact nitrate as a function of HNOo exposure
Calibration curve for the chemi luminescence technique. . . .
Thermal decomposition apparatus
Comparison of chemi luminescent and ion chromatographic
methods for ambient filter samples
Results of direct injection experiments
Nitrate sensitivity vs column age
Simultaneous hi^h volume nitrate collection results
Comparison of filter sulfate responses
Relative nitrate collection for low volume filters
Relative sulfate collection for low volume filters
Daily variation of measured variables
AID results for artifact nitrate .....
Page
10
10
11
16&17
19
20
22
28
29
51
57
64
66
74
82
91
94
96
99
101
-------
TABLES
Number Page
1 Gas Analysis Instrumentation 12
2 Physical/Chemical Properties of Phase I Filter Media 13
3 Methods of Analysis 14
4 Filter Adsorption Results for Nitric Acid 23
5 Pan-Filter Interaction Study Results 24
6 Results of Low Concentration Nitric Acid Experiments 25
(350 ppb HN03)
7 Comparison of Low Concentration Nitric Acid Experiments with
Different Sample Volumes 26
8 Presoiled Filter Analyses (mg/Filter) 32
9 Soot Interaction Study Results (mg/Filter) 33
10 Sampling Rate Study (yg/Filter) 34
11 Results of the Sampling Time Study 36
12 Compound Filter Results 38
13 Results of the Storage-Time Study 38
14 Chemiluminescence Response Curves for Selected Inorganic
Nitrates 52
15 Chemiluminesce.it Nitrate Method Reproducibility Study Using
Ambient Aerosol 54
16 Effect of Sample Volume on Nitrate Analysis 58
17 Comparison of Chemiluminescence With Ion Chromatographic Nitrate
Determinations 61
18 Replicate Nitrate Analyses 62
19 Hot Leach vs Ultrasonic Filter Extraction 69
20 Results of EPA Performance Audit For Nitrate 69
21 Comparative Nitrate Analyses 71
22 Measurements at Upland Station 77
23 Analytical Methods 78
24 Comparison of Nitrate Collected on Various Filter Types 80
vi i i
-------
25 Comparison of Nitrate Collected on Quartz and Glass Filter. ... 83
26 High Volume Filter Nitrogen Balances 86
27 High Volume Sulfate Results 88
28 Average Pairwise Sulfate Differences 92
IX
-------
ACKNOWLEDGMENT
We wish to thank the Environmental Sciences Research Laboratory,
U.S. Environmental Protection Agency for financial support of this program.
Helpful discussions with R. Coutant and D. Miller of Battelle-Columbus, and
E. Wittgenstein and J. Mulik of EPA are gratefully acknowledged.
-------
SECTION 1
INTRODUCTION
It has been known for many years that particulate nitrate is a
fundamental constituent of our atmosphere and that the nitrate* burden
increases considerably as one approaches major urban centers. Questions
as to the physiological impact of particulate nitrate were raised at the
time of the Chattanooga epidemiological study* 'and have been asked with
increasing frequency since that time. However, it has only been in recent
months that the correlations between particulate nitrates and certain
types of morbidity have become available. Certainly a strong indication
(2)
that nitrates cause detrimental health effects is the recent finding* '
that airborne nitrates are associated with aggravation of asthma, even in
areas where primary ambient air quality standards are not exceeded.
Particulate nitrate has been determined for many years through-
out the United States by standard high volume sampling techniques using
glass-fiber filters. Robinson and Robbins^ ' have estimated the global
background nitrate concentration to be on the order of 0.2 yg/m in the
lower atmosphere. Measurements of nonurban nitrate levels by the National
Air Surveillance Network* ' generally exhibit a range of annual averages
3 "3
between 0.1 and 1.0 yg/m , with an overall mean of approximately 0.5 yg/m .
The lowest nitrate values are found in such areas as Glacier National Park
in Montana and Black Hills National Forest in South Dakota, well away from
industrial and population centers. Nitrate as nitric acid, has been shown
to exist in the stratosphere at concentrations on the order of 0.003 ppm,
and is closely associated with the stratospheric ozone layer. There is
—
Throughout this report the term "nitrate" will be used interchangeably
with "particulate nitrate." Gaseous nitrates will be specified as such.
-------
some evidence^6) that this stratospheric nitric acid can be transported
across the tropopause and thus contribute to the background tropospheric
nitrate burden.
The level of particulate nitrate in and around major urban centers
can be considerably higher than background levels. Based on an 8-year
study^ ' of the mean concentrations of selected particulate contaminants
in the atmosphere of the United States, it appears that nitrate on the
average contributes somewhat less than 2 percent of the total suspended
***"W
particulate weight. The figures vary depending on location. Several
3 3
representative urban areas were Atlanta at 2.0 yg/m , Chicago at 2.5 yg/m ,
3
and Pittsburgh showing 3.0 yg/m .
In certain areas, such as the California southcoast basin, the
o
nitrate levels are even higher, averaging nearly 5 yg/m in the vicinity
of downtown Los Angeles according to NASN results^ . It is interesting
to\
to note however, that Gordon and Bryanv ' have reported yearly average
nitrate levels in downtown Los Angeles ranging between 9.8 and 15.4 yg/m .
Short-term nitrate levels in the eastern basin have been reported as high
as 247 yg/m3*9).
The major source of atmospheric particulate nitrate is thought
to be oxidation of natural and anthropogenic NO and NOg- The major sinks
for particulate nitrate are precipitation scavenging and dry deposition,
with the precipitation mechanism estimated^ ' to be three times as important
as dry deposition on a global basis.
Between the emission of gaseous NO or N0« and the ultimate scaveng-
ing of the particulate nitrate, there is a highly complex series of reactions
which may result in a variety of reaction products prior to the ultimate
formation of particulate nitrate. In a recent smog-chamber study of nitrogen
oxides reactions conducted by Spicer and Miller, the major initial products
of nitrogen oxides reactions in simulated photochemical smog were peroxyacetyl
nitrate (PAN) and nitric acid. Excellent nitrogen mass balances were main-
tained throughout the experiments and the complex mechanisms leading to the
formation of organic and inorganic nitrates were investigated. There is
little doubt that in the actual atmosphere several other forms of nitrate
-------
exist. For example, we have observed^ low concentrations of alkyl
nitrates and peroxypropionyl nitrate in urban atmospheres. We have also
detected low levels of participate organic nitrate in atmospheric aerosol
samples' . Heuss and Glasson' ' have observed another organic nitrate,
peroxybenzoyl nitrate (PBzN), in smog simulations. There is also reason
to suspect important, albeit low, levels of NgOg, a nitrate precursor, in
urban atmospheres. It is almost certain that all of these forms of
gaseous nitrate ultimately end up as particulate nitrate.
Because of differences likely to be associated with the physio-
logical impact of gaseous and particulate nitrates, it is important to
distinguish between the two. The most common means of collecting particulate
nitrate is filtration. However, there are a number of known problems
involved in filtration sampling of particulate nitrate, and there are also
several potentially serious problems which are not so widely recognized.
As with any system designed to collect samples by filtration, a most
important variable is the filter medium itself. Pate and Tabor' ' have
described the characteristics of a wide variety of glass-fiber filters
which have been employed by NASN and others for nitrate collection. The
manufacture of such glass-fiber filters requires at least six separate
steps and usually four participating manufacturers. It is therefore not
surprising that substantial variations in the composition and character-
istics of these filters often occur. Such variations can affect the
accuracy or efficiency of particulate nitrate sampling by
(1) Changing the nitrate filter blank
(2) Affecting the efficiency of particulate
collection
(3) Affecting the degree to which particulate
nitrate may react with the filter and become
unavailable for leaching and analysis
(4) Affecting the extent of extraneous nitrate
formation on the filter by nitrate precursors.
-------
Pate and Tabor' 4' have reported that the nitrate blank can
typically comprise up to 10 percent of the nitrate collection in urban
areas, and a considerably higher fraction of nonurban filter samples.
However, as long as frequent blank determinations are carried out, this
should not be a major difficulty except when sampling extremely low levels
of particulate nitrate.
A potentially serious problem in the collection of particulate
nitrates was pointed out in 1974^ J5.16) This pr05iein relates to the
collection of artifact nitrate on filters due to the interactions of
gaseous nitrogen compounds with certain filter materials. Two different
studies at Battelle-Columbus have revealed the problem of artifact nitrate
collection on certain filter materials. In one study^'7' investigating
nitrate in auto exhaust, it was found that glass-fiber filters collected
almost twice the quantity of nitrate in exhaust as did quartz-fiber filters.
Nitrate also appeared on backup filters for both quartz and glass, providing
an additional indication of artifact nitrate formation.
In a separate investigation at Battelle, the interaction between
nitric acid and a variety of filter materials was studied. The impetus
for this study came from the discovery of discrepancies in our atmospheric
nitrate data collected on quartz as opposed to glass filters. The re-
sults of the study and the implications in terms of atmospheric chemistry
and past particulate nitrate data have been reported^ ' and presented' ''
elsewhere. Briefly we find that both the absolute concentrations and also
the assumed size distributions of ambient particulate nitrate from many past
studies may be in error, due to gaseous nitric acid interference.
One possible explanation for the different collection efficiencies
of gaseous nitric acid by quartz and glass filters involves the filter pH.
Studies of filter characteristics at Battelle-Columbus^18^ have shown that
quartz filters are nearly neutral (100 ml filter extracts yields pH of 5-7)
while glass filters are often distinctly alkaline (100 ml filter extract
yields pH of 9.4). Thus, neutralization and trapping of nitric acid and
other acid gases may occur to a greater extent on glass filters than on
quartz.
-------
O'Brien, et al. ' have described a study of photochemical
aerosols in the Los Angeles basin in which high-volume samplers and a
cascade impactor were employed to determine concentrations and size dis-
tributions for NOg, NH4, and SO^, among other species. The high-volume
sampler and each stage of the cascade impactor employed glass-fiber filters
as the collection medium. The results of the study yielded very unusual
nitrate size distributions which appeared to be dependent on sampling site.
The investigators reported that the strange results could be attributed to
formation of artifact nitrate on the glass filters by some gaseous pre-
cursor such as nitric acid, and that the true nitrate size distribution
was masked in their study by the conversion of gaseous nitrate precursors
on the filter.
In addition to interactions between gaseous nitrate precursors
and filter media, several other potential interferences with particulate
nitrate collection should be considered. For example, certain nitrate pre-
cursors may be stable with respect to the filter medium but may be converted
by interaction with some component of the aerosol collected on the filter.
An example of this type of interaction might be the formation of NaNOj by
the reaction of N02 with Nad collected on the filter. In addition, there
may be certain conditions of relative humidity, temperature, atmospheric
composition, etc., under which species such as NHU, N02» PAN, ^Oc. or N«0
could adsorb and/or react with filters or collected aerosol on the filters.
Some precursors may be held on the filter by adsorption long enough to be
oxidized to artifact nitrate by ozone or some other oxidizing agent.
Another type of interference might involve release of particulate
nitrate collected on a filter by conversion to some volatile form. An
example might be the reaction of sulfuric acid aerosol with particulate
nitrate already collected on the filter to form nitric acid, which could
then be lost by volatilization. In the same manner NH^NO^, which has a
significant vapor pressure, could be lost from particulate collection
filters.
Two recent studies which touch on the kinds of interferences
just discussed have been reported by Lovelock and Penkett' ' and Chang
(21)
and Novakov* '. The former investigators found that PAN and PPN do not
-------
exist in the clean air over the Atlantic ocean but that clean air has the
potential for forming PAN and PPN when exposed to glass surfaces. The
potential for PAN formation was greatest on days of high solar intensity,
with maximum production during the afternoon hours. An important aspect
of this study is the possibility that there are precursors even over the
oceans which will form gaseous nitrates given the proper reaction surface.
Since the proper reaction surface was glass, the same material used for
filters in most high-volume samplers, the possible importance of this
reaction mechanisms for particulate nitrate formation must be considered.
(21}
Chang and Novakovv ' have studied the formation of particulate
nitrogen species due to gas/solid interactions using ESCA. They report
the formation of several types of particulate nitrogen due to the reaction
of gaseous NO and NH3 with soot (carbon) particles. Since soot, along with
other forms of carbon, is collected on high-volume filter samples, the
possibility of forming artifact nitrate from NO and NH3 reactions exists.
Because of the potentially serious effect of the interferences
described above on the accuracy of particulate nitrate data, the U.S.
Environmental Protection Agency has initiated a 2-year study at Battelle-
Columbus to investigate the impact of these factors on nitrate sampling
procedures.
Methods of particulate nitrate analysis were also investigated
during the study since most nitrate methods currently in use have important
deficiencies. For this reason there is no universally accepted nitrate
method and different labs employ many different procedures. What is needed
is a fast, reliable, sensitive, specific, accurate and reproducible
instrumental method. Our goal was to screen several potential techniques
and develop the most promising for ambient nitrate analysis.
The objectives of this program are threefold:
(1) To investigate the effects of environmental
variables on the sampling and analysis of
particulate atmospheric nitrate
(2) To develop an improved method for the
analysis of particulate atmospheric nitrate
-------
(3) To conceive, develop, validate, and optimize
a sampling and analysis methodology for
atmospheric nitrate.
The experimental aspects of the program have been broken into
four phases. A description of each phase is shown below:
Phase I A laboratory investigation of the factors
affecting atmospheric particulate nitrate
sampling
Phase II Development of analytical methodology for
atmospheric particulate nitrate
Phase III Development and evaluation of a sampling
procedure for atmospheric particulate
nitrate
Phase IV Optimization, simplification, and delivery
to EPA of a validated sampling and analysis
methodology.
The remainder of the report will be devoted to a discussion of the
experimental results. Each of the above phases of our work will be discuss-
ed in turn.
-------
SECTION 2
PHASE I: LABORATORY STUDIES OF ARTIFACT
NITRATE COLLECTION ON FILTERS
For purposes of this study we have defined particulate nitrate
as any form of nitrate which exists as a filterable particle or adsorbed on
a filterable particle under actual atmospheric conditions. This definition
excludes all gaseous nitrates and any other nitrogen compounds which might
be collected on filters as nitrate due to reaction with the filter material
or the collected particulate on the filter.
The goal of the Phase I effort has been to investigate a number
of factors which might affect particulate nitrate sampling, and to screen
filter materials to determine whether suitable nitrate collection media
exist- While novel methods of nitrate collection were not excluded from
consideration during the study, we concentrated initially on filtration
methods due to the vast apparatus for high volume filter sampling already
existing nationwide. If a realistic filtration method for nitrate could be
developed, it would be unproductive to study novel techniques which would
probably never be employed on a wide scale. Therefore, emphasis has been
placed on uncovering factors which affect collection of nitrate on filters.
The major effects and variables that were investigated in the
Phase I laboratory study include:
• Gas/filter interactions
• Gas/filtrate interactions
• Gas/soot interactions
• Sampling time
t Sampling rate
0 Storage time
Each of these factors is discussed below.
8
-------
GAS-FILTER INTERACTIONS
The principal goal of this task was to determine the effect of
several nitrogen-containing gases on common filter media and to identify
types of filters which might be suitable for particulate nitrate sampling
in the atmosphere. The experimental procedure involved exposing selected
filter materials to ppm concentrations of nitrogen-containing gases in
clean air. Nitrogen-containing gases were first diluted with clean
cylinder air in a 500-cu-ft Teflon ^chamber. The chamber was then evacu-
ated through 47-mm filters of the various materials chosen for study. The
pressure and concentration of the nitrogen gas were monitored above and
below each filter during the experiment; after exposure the filters were
analyzed for NO.,", NOp", NH. , and total N. The experimental apparatus is
shown in Figures 1 and 2. A schematic of the apparatus is shown in Figure
3. Each filter was typically exposed to more than 1 cubic meter of the
dilute nitrogen-containing gas. The concentration of the nitrogen gas was
adjusted (low ppm) so that the filter was exposed to approximately the
same mass of nitrogen compound as a standard high-volume filter collected
in an urban area. The face velocity also was quite similar to a standard
Hi Vol.
The nitrogen-containing gases examined include NO, NOp. HN03,
PAN, NH-, and N^O. The analytical techniques used for these gases are
listed in Table 1. N^O was not determined directly but was prepared by
known dilutions of the pure gas. Both dry and humidified conditions have
been employed. We have investigated the effect of these nitrogen-containing
gases on artifact nitrate formation on a number of filter types. Most of
our experiments were conducted with the following filter types:
Glass Fiber - Gelman A
Glass Fiber - Gelman E
Glass Fiber - Gelman AE
Teflon - Millipore Mitex
Polycarbonate - Nuclepore
-------
FIGURE 1. PHASE I SAMPLING AND ANALYSIS APPARATUS
FIGURE 2. CLOSE-UP OF SAMPLING MANIFOLD AND
FILTER HOLDERS
10
-------
To instruments
and manometer
To filters
H U
Sample port
u \
To filters
SOOcuft
Teflon chamber
Filter holder
To instruments
and manometer
Critical
orifice
To vacuum pump
Figure 3. Schematic of sampling systems for eight simultaneous filters
11
-------
Nylon - Millipore Dura!on
Cellulose Acetate - Millipore Celotate
Quartz Fiber - ADL (prototype filter prepared for EPA by Arthur
D. Little, Inc.)
TABLE 1. GAS ANALYSIS INSTRUMENTATION
Gas Analysis Method
NO, NOo Chemiluminescence (low
temperature carbon con-
verter for N02)
HNOo Microcoulometry
PAN Electron Capture Gas
Chromatography
NI-U Chemi luminescence (dual
temperature catalytic con-
verter
Toward the end of the Phase I effort, several experiments were
conducted with some additional filters. These experiments involved
Quartz Fiber - Gelman Microquartz
Quartz Fiber - Pal Iflex QAST
Glass Fiber - Gelman AA (EPA Type)
Glass Fiber - Gelman Spectrograde
Cellulose-Backed Quartz - Pall flex E 70-2075 W.
V
A brief description of each of these filter types is shown in Table 2. The
procedures employed for analysis of the filter samples are summarized in
Table 3.
12
-------
TABLE 2. PHYSICAL/CHEMICAL PROPERTIES OF PHASE I FILTER MEDIA
CO
Filter Type Pore Size^ ' pHP '
Cellulose Acetate
Polycarbonate
Teflon-Mitex
Glass-AA
Glass-A
Glass-AE
Glass-E
Glass-Spectro
Nylon
Quartz-ADL
Quartz- QAST
Quartz- E 70-2075 W
Quartz-Mi croquartz
1.0
0.8
5.0
NA
NA
NA
NA
NA
1.0
NA
NA
NA
NA
6.65
6.0
7.0
8.9
8.3
9.4
8.5
7.2
5.3
8.1
8.1
6.2
-
Alkalinity^
(1.8 x 10-3)
(9 x lO'4)
0.0
3.24 x 10-2
4.2 x lO-3
4 x 10-2
3.8 x 10-3
1 x 10-4
(3.6 x 10-3)
1 x 10-4
3.8 x 10"3
(7.6 x lO-3)
-
NOj Bl.nkW>
<0.005
<0.006
<0.005
<0.005
<0,005
0.005
<0.005
<0.005
S0.007
<0.005
<0.005
<0.005
0.008
Supplier
Mi 1 1 i pore
Nuclepore
Millipore
EPA/Gel man
Gelman
Gelman
Gelman
Gelman
Mi 1 1 i pore
EPA/ADL
Pall flex
Pal If lex
Gelman
(a) Pore size in micrometers where applicable.
(b) ASTM-D-202; pH of 100 ml H20 extract.
(c) Milliequivalents of acid or base required to titrate to neutral point per gram of
filter. Parentheses indicate acidic filter.
(d) mg/47 mm filter.
-------
TABLE 3. METHODS OF ANALYSIS
Specie Method of Analysis
NH* Gas sensing electrode
NOg Diazotization-colorimetric
NOg Brucine sulfate colorlmetric/
Ion chromatographic
Total Nitrogen Modified Kjeldahl digestion
Following the initial experiments all of the filters were analyzed
for nitrate and total nitrogen and many were also analyzed for ammonium
and nitrite, depending on the gas being studied. However, none of the
filters showed any significant increase in the nitrite concentration under
any circumstance, and only the nylon filters have shown substantial increases
in particulate ammonium levels when exposed to gaseous NH~. Therefore,
nitrite and ammonium analyses were discontinued.
The results of many of our Phase I experiments are shown in
Figures 4 and 5. These figures show in bar-graph format the quantity of
artifact nitrate found on the filter after exposure to ppm concentrations
of the gases shown on the left side of the figure. The experiment number
shown at the left of the figure is keyed to a complete tabulation of the
experimental conditions contained in the Appendix. In most cases, two or
more different concentrations of the nitrogen-containing gas were employed,
and frequently both dry and humidified conditions were examined
For purposes of this discussion, we will arbitrarily set the
level of significant nitrate interference as >100 yg. Since we are dealing
with a surface effect and a standard 8" x 10" Hi-Vol filter has 40 times
the effective collecting area of the 47-mm filters used in this study, a
standard Hi-Vol filter might be expected to collect about 40 times as much
artifact nitrate as our filters. If our filters collect 100 yg of bogus
nitrate, a standard Hi-Vol might collect 4000 pg of artifact. Assuming
q
a 24-hour sample volume of 2000 m for a standard Hi-Vol filter, the
14
-------
level of nitrate interference would be 2 vg/m3. This concentration approaches
the average NOo concentration measured in many urban areas and thus con-
stitutes a major interference.
Figure 4a shows the results of our experiments with cellulose
acetate filters. Two different experiments with nitric acid exhibit sub-
stantial interferences on this filter type, thus precluding its use as a
particulate nitrate sampling medium.
The results of our studies of polycarbonate filters are shown
in Figure 4b. It is clear from this figure that none of the gases studied
produced significant artifact nitrate on polycarbonate filters.
The results with Teflon filters, pictured in Figure 4c, are
similar to the polycarbonate results in that no important interferences
are apparent from any of the gaseous species studied.
The results of our investigation of nylon filters are shown in
Figure 4d. Nylon was chosen for study in the hopes that it might provide
quantitative collection of gaseous nitrates and thus serve as a gaseous
nitrate sampling technique. It is clear from the figure that large
quantitites of nitric acid are collected by the nylon medium. Judging from
the final experiment, high levels of NC^ (30 ppm) at high humidity also
lead to very high levels of artifact nitrate. Experiment No. 11 indicates
formation of considerable artifact N0~3 during NH3 exposure while the second
and third NH- runs show no such effect. We suspect that some NHLNO- may
have formed in our Telfon chamber during this experiment from trace quanti-
ties of HNO., remaining in the chamber from the previous experiment. Thus
the results of the first NH- experiment are probably in error.
The results of our investigations of two types of glass-fiber
filters (Gelman A and E) are shown in Figures 4e and 4f. Both filter
materials show substantial interferences from nitric acid and also from
high concentrations of N02 at high relative humidity. The interference
by N02 is extremely important due to the high concentrations of N02 which
frequently occur in urban areas. The interferences with particulate nitrate
collection on these alkaline-surfaced glass filters makes them rather poor
choices for nitrate sampling in the atmosphere.
15
-------
Experiment
Cellulose Acetate
1
2
3
«
s
6
7
8
9
10
11
12
13
14
15
16
17
Clcon Air
Clccn Air
NO,
NO,
NO, (R H )
HNOj
HNOj
HNO,
HNO,
HNOj (R H )
NH,
Blonk
NH,
PAN
N,0
N,0 (RH)
NH,IRH)
NO;(f 500 600 7OO BCC
Micrograms Nitrate on Filter
4a.
Polycarbonate
i
2
3
4
5
6
7
8
9
10
11
12
13
14
15
15
17
Clean Air
Clean Air
NO;
NO,
NO, (RH)
HNOj
HNO,
HNO,
HNO,
HNOj(RH)
NH,
Blo-ik
NH,
PAN
N,0
N,0 (RH)
NK,(R.H)
N02(R.H)
2OO 5OC 4OO '.CO 6OO 7CO 6OO
Micrograms Nitrate on Filter
4b.
Experiment
No.
' Ciron t'i
1 C.eon Air
» NO,
« NO,
5 1,0, IRr!)
6 HNO,
7 HNO,
8 UNO,
9 HNOs
10 UNO, (RH)
11 NH,
Blank
12 KM,
13 PAN
14 N,0
IS N,0(RH)
16 NHjlRH)
17 NO,(KH)
Teflon &'%;«"t Nylon
i
1 Clean Air
e Clean Air
3 N02
4 NO;
5 NO, (RH)
6 HNO,
7 HNO,
8 HNOj
9 HNOs
10 HNO, (RH)
11 NH,
Blonk
12 NH,
13 PAN
14 N20
IS N,0(RH)
16 NH,(RH)
17 N02(RH)
in
2500/i.s j
S'.OO/i? |
9700/ig \
BBOO^g \
}
2?00/,, \
1
FJ
a
«)O ?OO 30O 4OO !>UO fOO TOO BOO
Hicrograms Nitrate on Filter
4c.
o no ?oo 500 oou ?OD
Micrograms Nitrate on Filter
4d.
Figure 4. Gas-filter interactions.
16
-------
Experiment ., Experiment
No. Glass Fiber - Gelman A NO. Glass Fiber - Gelman E
i
i
3
4
&
6
7
6
9
10
11
12
13
14
IS
16
17
ClfCM Air
Clcon Air
NO,
NO;
NO; (RH)
HNOj
HNOj
HNO>
HNO,
HNOj oo f.oo TOO
Micrograms Nitrate on Filter
Figure 4. Gas-filter interactions
17
-------
The results of our investigations using quartz-fiber filters are
pictured in Figure 4g. In terms of artifact nitrate collection, the quartz
filters look quite good. Only one experiment, at high nitric acid and
high relative humidity, shows any significant interference. However, the
interference suggested by this experiment is questionable since the total
nitrogen analysis indicated no artifact nitrogen collection by the filter.
The results of experiments with four additional filter materials
compared with the ADL quartz filters are shown in Figure 5. The Pall flex
QAST quartz filters show no evidence of serious interference. However, the
three remaining materials were all influenced by nitric acid, and the Gelman
AA filter by nitrogen dioxide. The interference with particulate nitrate
collection on Gelman AA filters is particularly significant since the
material is widely employed in the NASN program. We strongly suspect that
the special surface coating of the Spectrograde filter is attacked by nitric
acid, the interaction with the exposed surface resulting in very high levels
of bogus nitrate.
Toward the end of our program the ADL microquartz filter became
commercially available through Gelman. The new filter is called Gelman
Microquartz. Since the original ADL filter appeared so promising for
nitrate sampling, we initiated an investigation to compare the new Gelman
Microquartz filter with the original ADL filter, Pall flex QAST quartz,
Spectrograde, Gelman AE, MSA glass and EPA type AA filters. All experiments
were conducted at 30 +_ 10 percent relative humidity, Each filter was
exposed to 1-2 cubic meters of air containing N02 or HIWL. The results
of nitrate analyses of filters exposed to two different concentrations
of N02 are pictured in Figure 6. As we found in our earlier work, ADL
quartz and Pal Iflex quartz exhibit the lowest artifact collection at both
of the N02 concentrations. Gelman AE and AA glass filters show the
greatest artifact collection in these N02 experiments. In both experiments,
the Gelman variation of the Microquartz filter collected more artifact
nitrate than either the original ADL Microquartz or the Pal Iflex quartz
filters. The only filter which approached our interference criterion of
100 ug was Gelman AE in the high concentration N02 experiment.
18
-------
Experiment
No.
23
24
25
23
24
25
23
24
25
23
24
25
23
24
25
NO,
HMOs
N02 (R.H.)
N02
HN03
N02 (R.H.)
N02
HN03
N02(R.H.)
N02
HN03
N02(R.H.)
N02
HN03
N02 (R.H.)
Quartz -Microquartz
p
Quartz- Tissuequartz
^
Quartz - Fiber - Backed
1
J
Glass Fiber- Gelman Spectro
1
Glass Fiber- Gelman AA
1
""~|
i 1 1
100 200
Micrograms Nitrate on Filter
Figure 5. Gas-filter interactions
300
-------
IOO
80
UJ
<
o
£
01
O
r
o
3
V
e
o
i
c
o
IT
O
w
O
i
o>
•o
o
6>
o
w
*-
o
0>
K
w
o<
O)
^ 60
c
o
S
o
UJ
c
o
5
o .
^
t)
2
o
"Si
o
9
o
o
en
-------
The results of nitric acid exposures at 0.27 ppm and 3 ppm
are shown in Figure 7. In these experiments, all three of the glass-fiber
filters exceeded the interference criterion, and the Gelman microquartz
exceeded it in one case and equalled it in the other. It seems likely that
the commercial Microquartz is inferior to the original prototype filter for
nitrate sampling. Only ADL quartz and Pall flex quartz filters showed
minimal interference in these experiments. In agreement with our previous
results (e.g., Figure 5 and Table 4), less artifact is collected on ADL
than on Pall flex. The results of these additional N0« and HNO-, experiments
confirm our earlier studies that ADL Microquartz and Pallflex QAST quartz
are clearly superior to glass fiber filters for nitrate sampling.
Breakthrough Experiments
The interaction of the gaseous nitrates PAN and nitric acid with
filter materials is of particular concern in this study because: (1) We
have observed these gases at relatively high concentrations (50 ppb) in
urban atmospheres and (2) They could easily form artifact nitrate by simple
adsorption on filters. In addition to the experiments already reported
with these two gases, we have carried out several experiments using a
different technique. In these experiments either PAN or nitric acid was
generated at low (sub-ppm) concentrations in a 200 liter glass smog chamber.
The mixture was then pulled through the filters under study. The concentra-
tion of PAN or nitric acid in the chamber was determined at the start of
the experiment, and then the concentration was monitored down stream of
the filters. After an initial induction period during which the gaseous
nitrates were removed by the filter, breakthrough occurred and the concen-
tration downstream of the filter slowly increased to the level in the chamber.
The breakthrough response curve was integrated and used to calculate the
mass of gaseous nitrate removed by the filter. A system blank (same
apparatus but filter holders empty) was run for each gas and subtracted
from the filter results.
21
-------
5OO
Ul
4
o
E
at
O
o
o
2
E
«S
in
IA
O
en
S
V)
4
O
X
a>
s
u
Z
o
E
3
Ul
4
400
o
.y
5
O
4
O
o>
E
s
d S
4 2
o
300
ZOO
100
e
o
a
E
•I
O
Figure 7. Comparison of various filters exposed to HN(L.
(1-2 m3 exposures)
22
-------
The results of the filter adsorption experiments for nitric acid
are shown in Table 4. These data are in qualitative agreement with the
nitric acid results presented earlier, i.e., nitrate interference due to
nitric acid adsorption on Teflon, quartz and polycarbonate filters is
negligible. Interference due to adsorption on the two types of glass
filters can be serious.
TABLE 4. FILTER ADSORPTION RESULTS FOR
NITRIC ACID
Nitrate Removed by 47-mm
Filter Material Diameter Filter, yg
Teflon 0.13
Quartz 0.11
Polycarbonate 2.8
Glass (Gelman E and AE) Very large (no breakthrough
after 8 hours of sampling)
PAN was generated at ambient concentrations for the breakthrough
experiments by the photolysis of ethyl nitrite in dry air. A gas chromato-
graph equipped with an electron capture detector was used to monitor the
PAN. The mass of PAN removed by the various filters is shown in Table 5,
along with a list of the potential artifact nitrate which might be expected
under actual ambient sampling conditions for each filter type. These data
indicate, as did the results from experiment 13, that nitrate interference
due to PAN adsorption is unimportant for the filters studied.
Nitrogen Balance
A check of the filter nitrogen balance for nylon, Gelman A and
Gelman E filters was carried out using the results of experiment numbers
6, 7, 8, 9, 10 and 17. These experiments were chosen because the quantity
of artifact nitate found on the filters was well above the filter blank.
23
-------
TABLE 5. PAN-FILTER INTERACTION STUDY RESULTS
Potential Hi-Vol.
PAN Adsorbed, Nitrate Interference,
Filter Material yg yg
Quartz - ADL
Quartz - Pall flex QAST
Quartz - Pal If lex E 70-2075 W
Glass - Gelman AA
Glass - Gelman A
Glass - Gelman E
Glass - Gelman AE
Glass - Gelman Spectrograde
Nylon - Duralon
Cellulose Acetate
Polycarbonate
Teflon
0.003
0.006
0.009
0.35
0.010
0.012
0.18
0.020
0.008
0.007
0.007
0.005
.065
.13
.20
7.6
.22
.26
3.9
.43
.17
1.3
.15
.11
24
-------
The percentage of the total filter nitrogen which could be accounted for
as nitrate-nitrogen has been calculated for these three filter types, taking
both the nitrate and total nitrogen blank into account. The average
accountability for the six experiments was
Nylon - 71 percent
Gelman A - 127 percent
Gelman E - 93 percent.
The value shown for nylon is probably not very accurate due to the large
total nitrogen blank correction. The high Gelman A percentage seems to
result from low total nitrogen analyses of Gelman A filters for two of the
experiments. The reason for the low analytical results is unknown.
Effect of Filter Saturation
All of the experiments described above have involved sampling
about 1 cubic meter of air containing ppm quantities of some nitrogeneous
gas through individual filters. While the results of these experiments will
be used to indicate types of filters suitable for particulate nitrate
sampling, still it would be useful to understand the relationships among
artifact nitrate formation, sample concentration and sample volume. The
results of a preliminary investigation of these relationships are listed
in Table 6. The experiment (No. 22) shown in this table was designed to
TABLE 6. RESULTS OF LOW CONCENTRATION NITRIC
ACID EXPERIMENT (350 ppb HN03)
Fil
Glass
Quartz
Quartz
ter Material
(Glaman AE)
(ADL Microquartz)
(Pal If lex QAST)
N03~,
mg
0.23
<0.005
0.086
mg
<0.
<0.
<0.
003
003
003
Total N,
mg
0.04
0.01
0.03
investigate the effect on filters of sampling low nitric acid concentrations
3
over longer exposure times. Approximately 4 m of the sample mixture (350
ppb nitric acid in air) passed through the filters. Based on the 100 yg
25
-------
interference criterion put forward earlier, the glass filter shows
significant artifact nitrate. The ADL microquartz was unaffected by
the dilute gaseous nitric acid, while the Pallflex tissuequartz
collected artifact nitrate at levels approaching the 100 yg inter-
ference criterion. These data can be compared with the results of a
second nitric acid experiment at a similar concentration. A side-by-
side comparison of these two experiments is presented in Table 7.
TABLE 7. COMPARISON OF LOW CONCENTRATION NITRIC
ACID EXPERIMENTS WITH DIFFERENT SAMPLE
VOLUMES
Filter Material 0.315 ppm HNO^3* 0.27 ppm
Gel man AE
ADL Microquartz
Pallflex QAST
N03", mg
0.23
<0.005
0.086
NO ~ mg
0.29
<0.005
0.020
(a) 4m3
(b) 1.3m3
The results of the two experiments are remarkably similar, considering that
they were conducted over a year apart. It is clear that the ADL Microquartz
was not affected by exposure, while the AE glass filter was seriously
influenced. The effect on Pallflex was variable, but in neither case was
the interference criterion exceeded. The amount of NO," found on the
Pallflex filter increased by a factor of four when the mass of HNO-j passed
through the filter was quadrupled. However, the Gelman AE filter appeared
to saturate at the lower concentration and showed no gain in NO-" with
quadrupling of the HN03 through-put. The saturation of the filter surface
with artifact nitrate is important to our understanding of artifact collec-
tion and will be explored further.
26
-------
Saturation of the surface active (basic) sites of filters by
artifact nitrate was suggested in our discussion of Table 6 and 7. If
saturation does occur, then the influence of artifact collection on
ambient nitrate measurements will go through a maximum, which will occur
at the time of saturation. Additional sampling through the saturated
filter will reduce the impact of artifact nitrate on the ultimate calculated
nitrate concentration. Therefore, short term collections during periods of
high precursor concentrations may result in the most serious sampling
errors.
In our laboratory studies, filter saturation could make the
artifact collection efficiency appear artifically low, thereby masking the
true extent of the interference. We have investigated this phenomenon with
N02 and HN03 using a glass and a quartz filter. The results of our studies
with NCL are shown in Figure 8. In these experiments, sub-ppm concentrations
of NCL (RH = 20%) were sampled through quartz and glass filters for
different times. The quantity of artifact nitrate collected on each filter
is plotted versus the mass of NCL which passed through the filter. As we
might have expected from our earlier results, the quartz filter saturates
earlier and at lower levels than the glass filter. From previous results we
would expect the ADL quartz and Pallflex quartz to saturate even more rapidly
than the Gel man Microquartz. Based on Figure 8, the Gelman Microquartz
appears to reach saturation at about 110 yg/47 mm filter. The AA glass is
not yet saturated even at 210 yg.
The results of a similar set of experiments with nitric acid are
depicted in Figure 9. In these experiments, the filters were saturated
even at the lowest nitric acid exposure, so that the left-hand portion of
the curves were extrapolated to zero. The filters were certainly saturated
by exposure to 2 mg HN03, but the amount required for saturation may have
been much less than this. Other data (from Figure 7) suggest that both the
AA and Gelman Microquartz filters become saturated with exposure to less
than 1 mg HNOg. Since AA glass seems to saturate at about 320 yg artifact
N03~, the filter is at least 30 percent efficient at scrubbing gaseous
nitric acid, up to the saturation point. Results of the breakthrough
experiments described earlier demonstrate that the glass filters are actually
100 percent efficient at collecting nitric acid up to the saturation point.
27
-------
240
ro
Co
200
o»
« 160
o
tt>
*-
U
= 120
o
U
0)
o
o
80
40
A Gelman Microquartz
X Gel man AA
45
Exposed to Filter.mg
8
Figure 8. Artifact nitrate as a function of N02 exposure.
-------
400
ro
V£>
A Gelmon Microquartz
X Gelman AA
I
I
I
I
I
5 6 7 8 9 10
HN03 Exposed to Filter,mg
II 12 13 14
16
Figure 9. Artifact nitrate as a function of HNCL exposure.
-------
The data in Figure 7 indicate that MSA and AE glass have similar high
efficiencies for collection of gaseous nitric acid. Thus these filters
should saturate at nitric acid exposures of 300-400 yg.
Gelman Microquartz seems to saturate at the same level of artifact
N03" for both N02 and HN03 exposures, although the efficiency of nitric
acid removal is much higher. The mass of artifact N03~ collected at
saturation probably varies somewhat from batch to batch and even from
filter to filter for the same filter type, but appears to fall in the range
of 100-150 ug for 47 mm Gelman Microquartz filters.
Gelman AA filters saturated at ^320 yg artifact N03~ in the
nitric acid experiments shown here and in Figure 7. The MX> exposures were
not sufficient to saturate this filter, but extrapolation of Figure 8 to
<320 yg artifact NOo" indicates that exposure to approximately 8 mg NOg
would be required for saturation under the conditions of our experiments.
The results of these experiments indicate that a 47 mm Gelamn AA filter
3
will saturate at ^320 yg artifact nitrate when exposed to 80 m of 0.05 ppm
NO- in air or 0.002 ppm HN03 in air. Obviously the filter will saturate more
rapidly when both N0« and HN03 are present. Based on Figures 8 and 9, even
smaller volumes will suffice to saturate the Gelman Microquartz filter.
Extrapolating these results to the surface area and flow rates
of typical high volume samplers, it seems highly likely that both filter
types would approach saturation over a 24 hour collection in a typical
urban area. Saturation of a Gelman AA Hi Vol would typically result in
3 3
^5 yg/m artifact nitrate. Gelman Microquartz could collect nearly 2 yg/m
of artifact. Although these values are based on extrapolations, the assump-
tions behind the extrapolations seem logical. Further discussion of these
data will be reserved for a later section of this report dealing with our
field measurements. The Phase I study of gas/filtrate interactions was not
designed to quantitate nitrate interference so much as to screen prospective
filter materials in terms of their suitability for particulate nitrate
sampling. A subsequent phase of this program has investigated the impact of
artifact nitrate formation under actual ambient conditions and should provide
a much more accurate estimate of the extent of nitrate interference. Indi-
caitons from our laboratory data suggest that we saturated the surface
sites of many of the filters early in our experiments by using ppm quantities
30
-------
of precursor gases. Under such conditions the apparent artifact collection
efficiency is lower than expected under ambient sampling conditions. In
other words, we believe that the percentage of artifact nitrate interference
reported here may only be the lower limit to that expected under ambient
sampling conditions.
GAS-FILTRATE INTERACTIONS
As mentioned in the introductory section of this report there are
several possible interactions between gases or aerosols in ambient air and
the particulate matter (filtrate) collected on high volume filters, which
could result in either positive or negative interference with particulate
nitrate determinations. The interaction between gases or aerosols passing
through a filter and the collected filtrate was investigated by exposing
actual ambient filter collections to exaggerated concentrations of several
potential interferences. The concentration of nitrate on the filters was
determined both before and after exposure, so that a simple comparison
should reveal any significant interferences. In terms of experimental pro-
cedure, ambient Columbus aerosol was collected by Hi Vol. sampling on several
142-mm filters. 47-mm circles were cut from these filters and exposed to
candidate gases in the same apparatus used for the gas-filter interaction
studies. The results of nitrate analyses of these filters are shown in
Table 8. The lack of consistent changes in N03~ levels upon exposure
indicates that interactions between candidate substances (NH-, HUSO., HNCL,
NOg) and the collected particulate are not significant. The one exception
seems to be cellulose acetate in experiment 20. Both N03~ and total N
increased significantly during this experiment.
GAS-SOOT INTERACTIONS
The potential role of soot collected on Hi Vol filters in convert-
ing gases such as NO, and NH, to nitrate has been discussed by Chang and
(21) j y
Novakovv '. The importance of gas-soot interactions was investigated in
this study by loading 142-mm Gelman A filters with 4 mg of finely dispersed
carbon-black and then passing approximately 2 m of air containing ppm levels
of NH3 or N02 through several 47-mm diameter circles cut from these filters.
31
-------
TABLE 8. PRESOILED FILTER ANALYSES (mg/FILTER)
Filter Material
3
NHj
Total N
Presoiled Filters (Before Exposure)
Glass-Gelman AE Or3'
Glass-Gelman AE (2)
Cellulose Acetate (3)
Cellulose Acetate (4)
Quartz-Mi croquartz (5)
Quartz-Tissuequartz (8)
Exposed Filters
Experiment No. 18 - 11.5 ppm NH3
Glass-Gelman AE (1)
Cellulose Acetate (3)
Quartz-Mi croquartz (5)
Experiment No. 19 - 97 yg/m3 H2S04
Glass-Gelman AE (1)
Cellulose Acetate (3)
Quartz- Mi croquartz (5)
Quartz-Tissuequartz (8) /. x
Quartz-Mi croquartz (unsoiled)* '
Experiment No. 20 - 3.4 ppm HN03
Glass-Gelman AE (1)
Cellulose Acetate (3)
Quartz-Mi croquartz (5)
Experiment No. 21 - 19.5 ppm N02
Glass-Gelman AE (1)
Cellulose Acetate (4)
Quartz-Mi croquartz (5)
Quartz-Tissuequartz (8)
0.36
0.26
0.037
0.047
0.038
0.12
0.35
0.033
0.025
0.28
0.030
0.055
0.12
<0.005
0.40
0.26
0.037
0.22
0.043
0.028
0.11
0.007
0.006
0.014
0.027
0.028
0.072
0.005
0.016
0.026
0.068
0.045
0.10
0.19
0.092
0.011
0.028
0.025
0.013
0.021
0.020
0.072
0.10
0.10
0.06
0.09
0.06
0.15
0.13
0.10
0.07
0.17
0.14
0.14
0.20
0.11
0.13
0.22
0.06
0.10
0.07
0.05
0.14
(a) Numbers in parentheses identify ports used during presoiling.
(b) S04"2 was 0.087 for this filter.
32
-------
Analysis of these filters gave the results shown in Table 9. The nitrate
level increased after exposure to NOp, however, the magnitude of the
increase is below the level of artifact nitrate formed during exposure
TABLE 9. SOOT INTERACTION STUDY RESULTS^
(mg/FILTER)
Exposure Conditions
Filter before exposure
11.5 ppm NH3
19.5 ppm N02
(a) Filter medium used
NOg NH*
0.008 <0.003
0.010 <0.005
0.017 <0.003
was Gelman A with
4-mg carbon-black loaded on a 142-mm
circle.
of clean Gelman A filters alone. Thus the gas-soot interaction does not
appear to contribute significant quantities of artifact nitrate under our
study conditions. Other types of soot were not examined, however so that
we cannot totally dismiss the possibility of an interaction under some
conditions.
EFFECT OF SAMPLING RATE
To examine the effect of sampling rate on nitrate collection
efficiency and artifact nitrate formation, ambient Columbus aerosol was
collected simultaneously over 24 hours on two groups of four Gelman AE
142-mm filters, each at two flowrates. The results of this study are
included in Table 10. The average nitrate collected on the four filters
run at 38.2 a/min flow rate was 2.65 ± 0.16, v9/m3 while the average for
the 99.3 £/min rate was 2.45 ± 0.20 wg/m3. Since this difference is not
statistically significant, nitrate collection is not affected by moderate
variation of sampling rate. However, since the variation in face velocity
was small (4-10 cm/sec) and even the highest, 10 cm/sec, is considerably
less than a typical Hi Vol (^50 cm/sec), extrapolation of these results to
higher flow rates may not be justified.
33
-------
TABLE 10. SAMPLING RATE STUDY (yg/FILTER)
Filter
Material
Gelman AE (1)
" (2)
Gelman AE (3)
" (4)
Gelman AE (5)
" (6)
Gelman AE (7)
11 (8)
N03-
138
380
146
360
153
315
146
345
NH/
29
82
29
89
42
97
29
124
Total N
70
160
70
170
70
200
70
200
m3
56
142
54
150
54
143
56
137
yg/N00"
m3
2.46
2.86
2.70
2.40
2.83
2.20
2.61
2.52
34
-------
EFFECT OF SAMPLING TIME
The effect of sampling time has been examined by simultaneous
collection of atmospheric aerosol samples on parallel samplers. During
the collections, the sample stream was split in half, with each half pass-
ing through an identical set of three filter types. One set of filters
continuously sampled the atmosphere for a 48-hour period, while the second
set of filters was changed after the first 24 hours of the sampling period.
Comparison of the amount of nitrate collected by the 48-hour filter with
the sum of nitrate collected by the two 24-hour filters will indicate
whether sampling time affects the collection of particulate nitrate. The
three filter types included quartz tissue (ADL), glass fiber (Gelman AE),
and a compound filter consisting of a quartz (ADL) and a nylon (Duralon)
filter inserted in the same filter holder. This dual filter was used to
investigate whether particulate and gas-phase nitrate could be separated
and determined simultaneously by a filtration technique. The quartz
filter has been shown to remove particulate nitrate but not gaseous nitrate,
while the nylon filter quantitatively removes gaseous nitric acid. Thus
the dual filter might make the simultaneous separation and determination of
the two nitrate types feasible.
The results of the total mass determinations and the aerosol
nitrogen analyses for these experiments are presented in Table 11. If
sampling time has no effect on the aerosol collection, then the sum of
the two 24-hour filter collections should approximate the 48-hour filter
sample. The data in Table 10 indicates that within the estimated experi-
mental uncertanity, the total mass, NH. , NO/, and total nitrogen values
are the same (no sampling time effect) for the 48-hour versus 24-hour com-
parisons, with two possible exceptions. The total mass collected by the
two 24-hour glass-fiber filters is considerably greater than the 48-hour
filter mass. We suspect a weighing error has caused this discrepancy. In
addition, the 48-hour glass filter collected more nitrate than the two
24-hour glass filters. This discrepancy cannot be readily explained but
is not significant enough to cause great concern.
35
-------
TABLE 11. RESULTS.OF THE SAMPLING TIME STUDY
Total Mass,
Filter Type mg
48- hour filter
Sum of 24- hour filters
48- hour filter
Sum of 24-hour filters
48- hour filter
Sum of 24-hour filters
Quartz
Glass Fiber
(Gelman AE)
Quartz and Nylon
23.53
26.44
27.92
35.03
30.63
30.03
NH^,
mg
1.11
1.21
0.74
0.93
1.91
2.11
NO^,
mg
0.56
0.64
1.92
1.44
1.24
1.28
Total N,
mg
0.97
1.30
0.85
0.94
0>94(a)
1.08
(a) Total nitrogen values for quartz filter only.
36
-------
Of major interest is the fact that the glass-fiber filters collected
much more nitrate than the quartz filters, almost four times as much for the
48-hour filters. This additional nitrate must be artificial and result
from collection of gaseous nitrogen compounds on the alkaline glass surface.
These actual atmospheric data tend to confirm our laboratory results and
show the potential impact of artifact nitrate formation on glass filters.
The glass filters collected less NH^ than either the quartz or
the quartz + nylon. This is understandable in the case of the dual filter,
since nylon was shown earlier to collect some NH3 as NH^. The lower levels
of NH^ collected on glass as opposed to quartz may be due to the alkaline
nature of the glass filter. Such filters may tend to reject alkaline sub-
stances such as NHU or NH, compounds.
Table 12 shows the analyses of the individual quartz and nylon
filters which made up our compound filter. Again, the sum of the 24-hour
filters compared to the 48-hour filters indicates no dramatic sampling
time effect. The quartz prefilter results from Table 11 compare quite well
with the single quartz filter results shown in Table 10, serving as a check
of our precision. The observation of NH. and N03~ on the nylon backup filter
indicates that gaseous ammonium and nitrate precursors are penetrating the
quartz filter and are adsorbed by the nylon backup. Thus, both the nylon and
the glass-fiber filters suggest that a gaseous nitrate precursor can strong-
ly influence the apparent particulate nitrate concentrations. This is com-
pletely consistent with our earlier Phase I experimental findings. It is
interesting to note that the sum of the artifact nitrate collected on the
nylon backup filters and the (assumed) actual particulate nitrate collected
by the quartz filters is 1.44 ng for both the 48 and 24 hour filters. This
is the same amount of nitrate collected by the 24-hour glass filters. While
one experiment is not definitive, it does appear that the glass filter
collected all the true particulate nitrate observed on the quartz filter and
all the artifact nitrate collected on the nylon backup filter.
37
-------
TABLE 12. COMPOUND FILTER RESULTS
48- hour filter
Sum of 24- hour filters
48- hour filter
Sum of 24- hour filters
Filter Type
Quartz (compound)
Nylon (Backup)
NhJ,
mg
1.11
1.13
0.80
0.98
N05,
mg
0.36
0.48
0.88
0.80
Total N,
mg
0.94
1.08
-
-
EFFECT OF FILTER STORAGE
Filters collected in the field for particulate nitrate determination
must frequently be stored for days, weeks, or even months before the actual
analyses are performed. The effect of this storage period on particulate
nitrate is uncertain. Therefore, a brief investigation of storage-time
effects was added to the Phase I effort.
The results of several storage-time experiments are shown in
Table 13. The filters were analyzed, stored for either 2 or 7 months in
TABLE 13. RESULTS OF THE STORAGE-TIME STUDY
Filter Materials
Gel man AE
Celotate
Quartz (Pal If lex QAST)
Duralon (Two 24-hr collections)
Duralon (One 48-hr collection)
NO]
Before
Storage
0.26
0.047
0.12
0.80
0.88
5
After
Storage
0.31
0.016
0.14
0.68
0.69
Storage
Time, Mos.
2
2
2
7
7
38
-------
glassine envelopes within sealed plastic bags, and then reanalyzed. There
does not appear to be any decay of nitrate on Gelman AE or Pal Iflex quartz
(QAST) during a 2-month storage period. Loss of nitrate from the Celotate
filter is indicated; however, the precision of the analysis at such low
levels is not good. This may account for part of the apparent decay. A
loss of nitrate was detected on the Duralon filters. This loss may be
related to the longer storage of the Duralon filter samples.
The results of this brief study suggest that storage of quartz
and glass (Gelman AE) filters for periods of at least 2 months prior to
analysis should have a minimal effect on particulate nitrate results.
39
-------
PHASE I SUMMARY
A great deal has been learned about particulate nitrate sampling
during this Phase I investigation. Our studies of the interaction between
gaseous nitrogen compounds and filter substrates indicate that nylon filters,
cellulose acetate filters and many glass fiber filters are subject to signif-
icant particulate nitrate interference due to the formation of artifact
nitrate on the filter. Teflon, polycarbonate and quartz fiber filters
showed only minimal interferences from gases studied. Considering other
factors such as cost, handling characteristics, pressure drop, efficiency
for submicron particle collection and mass loading considerations, the
quartz filters appear most appropriate for particulate nitrate sampling,
especially for large sampling networks.
Studies of the possible interaction between gases or aerosols
being pulled through a filter and the particles already collected on the
filter indicated no major interferences, either positive or negative, with
particulate nitrate determination. The interaction between NOg or NH3 and
soot collected on filters was also shown to result in negligible artifact
nitrate formation under the conditions studied.
Sampling time and rate were investigated in this study and were
found to have no effect on particulate nitrate collection over the limited
range of rates and times studied. The effect on nitrate determinations of
storing filter samples up to 2 months prior to analysis was found to be
negligible for glass (Gelman AE) and quartz filters.
The results of this laboratory investigation will be compared with
field data on many of the same filter materials in a later section of this
report.
40
-------
SECTION 3
PHASE II; SCREENING AND DEVELOPMENT OF
NITRATE ANALYSIS METHODOLOGY
GAS SENSING ELECTRODE
Introduction
The first technique which we investigated for nttrate determination
was a new gas-sensing electrode. This electrode responds to nitrite in
aqueous solution. Therefore the analytical procedure which we envisioned for
nitrate determination on ambient filter extracts would involve measurement
of nitrite in the extract, reduction of nitrate to nitrite, followed by a
second determination of total nitrite. The difference between the two
measurements represents the nitrate concentration. An advantage of this
method is that both ML" and N03" are determined on the filter extract.
The nitrate reduction is the critical step in the procedure. Many
methods have been used for nitrate reduction over the years, with variable
success. However, we chose to investigate the nitrate reduction/gas-sensing
electrode procedure because of reports in the literature describing new highly
efficient nitrate reduction techniques. It seemed that coupling the new
nitrate reduction methods with the novel gas-sensing electrode procedure
might yield a highly sensitive and specific analytical technique for ambient
particulate nitrate.
The electrode which we employed for this investigation was a
nitrogen oxide gas-sensing electrode which responds to N02" in solution. The
NO gas sensing electrode, unlike nitrate specific ion electrodes, is almost
/\
interference free. Anions, cations, common gases, sample color, turbidity
and suspended solids do not interfere with the measurement. The gas-sensing
electrode is also considerably more sensitive than the nitrate ion electrode.
41
-------
some volatile weak acids such as formic or acetic could interfere if
present at high concentrations, but this is extremely unlikely in
atmospheric aerosol samples.
The electrode utilizes a hydrophobic highly permeable membrane
which allows dissolved HNCL from the sample (formed by NO^" in acid
solution) to diffuse into the internal filling solution until equilibrium
is established. Hydrogen ions formed in the internal filling solution by
dissociation of HNCL are measured by the internal sensing element. The
electrode potential is claimed to be Nernstian with respect to HNOp con-
centration.
The procedures used over the years to reduce nitrate to nitrite
generally require extremely careful control of the reaction conditions to
produce reliable results, and even then the reduction efficiency is often
variable. Some of the procedures are also subject to interference by
chlorides and other anions. Both zinc and copperized cadmium have been used
for nitrate reduction. The use of zinc dust for the reduction presents the
problem that the reduction efficiency is strongly dependent on the quality
of the zinc, and in addition, some fraction of the nitrate may be reduced to
ammonia. For these reasons the method is considered unreliable.
Major improvements in the reduction of nitrate to nitrite have
recently been reported. Our investigation of these methods coupled with the
gas-sensing electrode is described below.
Experimental
The gas-sensing electrode employed for these experiments was an
Orion Research Nitrogen Oxide Electrode - Model 95-46. The electrode output
was monitored with a Keithley Electrometer - Model 600 A. For some experi-
ments the output was also read with an Orion Specific Ion Electrode Meter:
the results were identical. The electrode was assembled and checked out
according to the instructions provided with the electrode. An immediate
problem developed when the electrode displayed non-Nernstian behavior. A
factor of ten variation is solution N02" concentration resulted in a 50 ± 3
mv change rather than the 58-60 mv variation expected. The electrode membrane
was replaced, the internal solution was changed, and new standards prepared,
42
-------
all without effect. Conversations with Orion personnel resulted 1n several
additional changes and tests, all of which failed to bring the electrode
response up to the proper value. Since all of the other characteristics of
the electrode were 1n order, calibration curves were prepared and the electrode
was used without further efforts to adjust the response behavior.
One type of reduction catalyst was prepared according to the
method Lambert and Du Bois* ' from copper (II) sulfate, powdered cadmium,
ammonium chloride and dibasic sodium phosphate. A second catalytic procedure
required amalgamated cadmium in alkaline solution. Reagents employed for
this procedure were mercuric chloride, cadmium powder, sodium citrate,
sodium hydroxide, barium hydroxide and perchloric acid. This second catalyst
and reduction procedure are described in the Instruction manual for the
nitrogen oxide electrode - Model 95-46, Orion Research, Cambridge, Mass.
Samples and standards employed during this Investigation were prepared with
deionized and distilled water. Prior to each measurement with the gas-
sensing electrode the sample pH was adjusted to approximately 1.2 using an
acid buffer. This 1s required to convert nitrite ions into dissolved gaseous
nitrous acid.
Results and Discussion
The objective of this phase of our program was to screen methods
for nitrate analysis and, based on the results of the screening process,
develop a nitrate analysis procedure which Is fast, reliable, sensitive and
specific. At the start of our study, the gas-sensing electrode coupled with
a nitrate reduction procedure seemed to meet these criteria. During our
subsequent investigation we found that the gas-sensing electrode has all of
the required characteristics. It was fast (2-3 minutes per determination),
reliable and sensitive to less than 0.1 ppm nitrite. The electrode is
relatively free from interferences except at the lowest nitrite levels.
Under conditions where highest sensitivity is required, it is recommended
that the sample be pretreated to remove dissolved atmospheric CO-.
While the gas-sensing electrode met the criteria we had establish-
ed the overall analytical procedure did not. The problems with the method
occurred 1n the nitrate reduction step. The reduction catalysts which we
i
43
-------
employed were difficult to prepare uniformly and tedious to use. The
greatest problem was the slowness of the reduction step. Our initial
attempts at using a reduction column showed very erratic reduction effi-
ciencies. The efficiency was increased by using catalyst preparations of
finer mesh, but at an acceptable efficiency the flow rate through the
reduction column was intolerably slow. Experiments with different catalyst
particle sizes and different column geometries failed to provide a reduction
which was both fast and efficient.
Since the objective of this phase of our program was to screen
potential nitrate analysis procedures, we also investigated a chemiluminescent
and an ion chromatographic procedure simultaneously with the gas-sensing
electrode. Due to the difficulties encountered with the nitrate reduction
step and the positive results we were obtaining with chemiluminescence and
ion chromatographic techniques, work on the gas-sensing electrode procedure
was terminated in order to devote our full attentions to the latter two
methods. These methods are described in the next sections of this report.
THERMAL DECOMPOSITION/CHEMILUMINESCENCE
Introduction
A method for particulate nitrate determination which theoretically
seemed to possess the required criteria of speed, sensitivity and selectivity
is thermal decomposition of nitrate followed by chemiluminescent detection.
We envisioned a technique wherein nitrate in ambient aerosol, either on a
filter substrate or dissolved in water, would be decomposed to NO by rapid
A
heating. A carrier gas would then transport the NO to a commercial chemi-
A
luminescent nitrogen oxides monitor. The integrated output signal was
expected to be proportional to the original nitrate concentration. Brief
(23 24)
investigation of this procedure by other researchersv ' ' confirmed that
the method seemed promising. Chemiluminescent NO monitors measure the
A
light emitted (600-900 nm) when NO in the sample reacts with excess 03. The
sensitivity of commercial instruments is generally 5-10 ppb by volume of NO.
Various catalytic converters are employed to reduce N02 (and other gaseous
nitrogen compounds) to NO for determination; thus the instrument is said to
44
-------
monitor total NO . For determining NO the chemiluminescence instruments are
A
fast, sensitive and free from interference. Our task in developing and
validating a thermal decomposition/chemiluminescent method for atmospheric
particulate nitrate determination involved primarily the thermal decomposi-
tion step and introduction of the gaseous decomposition products into the
chemiluminescence monitor. Our results demonstrate that the procedure
discussed in this report is a rapid, sensitive, selective, and relatively
simple method for nitrate determination.
Experimental
Two commercial chemiluminescence instruments were employed in this
investigation. The majority of the work was conducted using a Bendix Model
8101-B chemiluminescence NO/NO monitor. This instrument was employed in a
j\
continuous mode (as opposed to its normal cyclic operation). For most
experiments the instrument's heated carbon catalytic converter was employed
to reduce any NO,, in the sample stream to NO, even though several experiments
demonstrated that NO accounted for >90 percent of the nitrate decomposition
products. Use of the converter ensured that even the traces of N02 formed
in the decomposition process were measured. For some experiments a Thermo
Electron Corp. Model 14-D chemiluminescence NO/NO monitor was used. The
A
Model 14-D is a dual channel instrument which simultaneously monitors NO and
NO . It employs a heated molybdenum catalytic converter for reduction of N09
A C
to NO. Both instruments were calibrated with low concentrations (0.5-5 ppm)
of NO in N£. The calibration standards were referenced to a National Bureau
of Standards "NO in N2" primary standard. A Hewlett-Packard Model 3370 A
integrator was used to integrate and digitize the instrument output.
One series of experiments employed an 18 cm by 2 cm quartz tube
surrounded by a resistance heated furnace for sample decomposition. A
second series used a resistance heated 18 cm by 0.30 cm stainless steel tube
for sample decomposition. Two electrodes connected the loop to a stepdown
transformer. The temperature of both the quartz tube and the stainless
steel loop was determined with a chromel-alumel thermocouple. Teflon tubing
was employed for the gas flow system. Tedlar (polyvinyl fluoride) bags were
45
-------
used for a number of experiments to collect and integrate the gaseous
decomposition products. The stability of NO in these bags was excellent
^\
over the short times required.
Ambient filter samples as well as aqueous standards were frequently
analyzed simultaneously by the chemiluminescence method and ion chromatography.
Discussion of the ion chromatographic procedures and results will be reserved
for a later section of this report.
Results and Discussion
Our investigation of the feasibility of a thermal decomposition/
chemiluminescence procedure for atmospheric particulate nitrate determination
suggested two different experimental configurations. The first design
utilized direct injection of the gaseous decomposition products into the
chemiluminescence monitor, with digital integration of the chemiluminescence
signal. This method is rapid, extremely sensitive, but not highly precise.
The second experimental set-up integrates the gaseous decomposition
products in a small Tedlar bag. The concentration of the products in the
bag is then determined by the chemiluminescence monitor. This method is
cheaper and simpler to set-up and operate but not as sensitive as the first
procedure.
Many of our initial experiments to characterize the thermal
decomposition process utilized the first, direct injection, method. The
following section will describe one variation of the direct injection
method and the results of our temperature and interference studies. Subse-
quent sections will discuss the bag integration technique and variations
in the direct injection method.
Direct Injection—
The apparatus for the initial series of experiments consisted
of a Vycor tube placed in a resistance furnace with a 450°C maximum
temperature. The inlet of the Bendix chemiluminescence monitor was con-
nected directly to the Vycor tube by Teflon tubing, so that room air was
drawn through the heated tube and directly into the monitor. The sampling
46
-------
pump of the Bench'x monitor pulled air through the system at approximately
120 cc/minute. The monitor was operated in the continuous NO mode and
A
the signal routed to the digital integrator.
In the initial experiments small quantities of nitrate compounds
were introduced into the hot zone of the Vycor tube using either platinum
or Vycor boats. For most experiments 1-50 yl of aqueous sample were placed
in the boat, dried by evaporation at the entrance to the Vycor furnace tube,
and finally inserted into the hot zone of the furnace with Vycor rod.
Using the platinum boat, this procedure had a detection limit of
less than 1 yg N03~; however the Pt boat blank was equivalent to about 0.2
yg N03~. Vycor boats were employed with the same result. These blank values
need to be reduced substantially in order to obtain the required sensitivity.
Different methods of washing, baking and flaming the boats were tried in
order to reduce the blanks. These methods were successful if the boats were
used immediately after cleaning. However, a few minutes exposure to room air
resulted in high blanks again. It appears that NO from room air is adsorbed
A
at the platinum or Vycor surface. This NO then desorbs and interferes with
A
the N03~ determination when the boat is injected into the furnace. This
same phenomenon was observed when a Vycor rod was used to insert and withdraw
the boats from the furnace hot zone. The problem with the rod was easily
overcome by leaving the rod in the furnace continuously.
Before concentrating on the blank problem, we investigated the
other characteristics of the apparatus to ascertain the feasibility of the
thermal desorption/chemiluminescence method. Using the Vycor boat,
quantities of NaN03 standards equivalent to 0.1, 0.2, and 0.5 yg of N03"
were flashed in the furnace and the resulting NO determined by chemilumi-
A
nescence. To extend the calibration curve to even lower values, injections
of dilute NO gas equivalent to 4, 8, 21, and 41 ng NO-" were performed with
a gas-tight syringe. A log-log plot of peak area vs yg N03~ yielded a
straight line and suggested that a detection limit of less than 10 ng might
be possible if interference and blank problems could be overcome. The
results obtained with this crude furnace apparatus confirmed that the method
was promising. Subsequent experiments were designed to optimize, simplify
and validate the method.
47
-------
Reduction of Blank Values—
Since the results obtained with the crude experimental apparatus
appeared promising, we investigated different ways of lowering the high
blank values associated with the Vycor and platinum boats. As a first step
toward improving the procedure we replaced the Vycor tube and furnace with
a heated gas chromatographic injector port. G.C. injector systems are
employed routinely to flash microliter volumes of sample into a carrier gas
stream. The small volume of the injector should deliver the decomposition
products to the chemiluminescent analyzer in a very sharp pulse, which would
lead to an improvement in minimum detectability limits. A nitrogen carrier
gas was used in our system to purge the injector port and carry the decom-
position products to the chemiluminescent monitor. Nitrogen carrier gas
was used to minimize the potential interference of NH. on the N03~
determination, since the removal of 0? from the system should inhibit the
4. ^
conversion of NH. to NO in the hot zone. Liquid syringes were used to
H J\
inject the sample through the septum and into the hot zone, thus avoiding
the use of sample; boats. A special convection cooled extension was fitted
to the injector to maintain the septum temperature at <200°C while the hot
zone exceeded 425°C.
Results using the G.C. injector port for thermal decomposition of
aqueous nitrate samples were highly scattered. It appeared that the high
temperatures required for decomposition of nitrate salts prevented repro-
ducible injections. The water apparently evaporated from the syringe needle
before the injection was complete, leaving nitrate salt deposites to decom-
pose in the needle rather than in the injector port. A number of modifica-
tions to the procedure were attempted, but none improved the reproducibility
to an acceptable level.
Because of the negative results with the injector port, the apparatus
was reassembled in its original Vycor furnace tube configuration, and alter-
natives to the Vycor and platinum boats were sought. After considerable
investigation the sample vessel which showed the lowest blank was found to
be small (^4 mm dicimeter) circles of high purity quartz fiber filter material
used in our ambient filter sampling experiments. The filter vehicles are
prepared by punching out a portion of a quartz filter with a No. 2 cork borer.
48
-------
2
This provides a filter circle 4.3 mm in diameter with 0.15 cm area. For
ambient filter samples, such small pieces may be analyzed directly by
flashing in the furnace, or a known volume of aqueous filter extract can
be syringed onto a clean filter pad prior to flashing. The lower
detectable limit for NO.," using this procedure is below 10 ng due to the
low and reproducible nitrate blank of the filter pads.
A series of experiments was undertaken to document the
reproducibility of the filter pad injection technique. Microliter volumes
of aqueous nitrate standards were syringed onto filter pads and subsequently
flashed and analyzed. The peak area was then used to determine the relative
standard deviation at each nitrate concentration. For these initial tests
a concentration range from 100-2000 ng NO.," was employed. The relative
standard deviation of multiple determinations (4-6) was 10-15 percent.
More detailed reproducibility studies were postponed pending further
refinement and characterization of the method. The next step in characteriz-
ing the procedure involved an investigation of interferences.
Interferences--
The likely interferences with this procedure are nitrite, ammonium,
and possibly organic nitrogen compounds which can decompose to NO .
X
Experiments with NaN03 and NH.NO-, solutions demonstrated that
ammonium interferes with the determination of N03~. The ammonium pre-
sumably decomposes in the presence of oxygen in the 425°C furnace to yield
N0x- To eliminate the NH. interference the Vycor furnace tube was con-
nected via a ball joint and Teflon tubing to a gas cylinder of oxygen-free
nitrogen. The Vycor tube was connected to the N« source via a tee which
was extended and tapered to permit access to the furnace for the Vycor rod.
A continuous flow of excess N2 passed out of the open end of the tee,
thereby excluding room air from the furnace.
Experiments were then conducted with both NH.C1 and NH.WL
solutions to determine the extent of NH. interference in the absence of Op-
The results demonstrate that the NH^* interference is almost totally
eliminated by decomposing the sample in a nitrogen atmosphere. For example,
the response of 100 ng NH^ in NH4C1 produced an equivalent NO," response
49
-------
of less than 4 ng. Amines and other organic nitrogen compounds, which may
be present in atmospheric filter samples at very low concentrations, were
investigated in several experiments; no significant interference was
observed as long as Op was excluded from the decomposition tube.
Interference due to decomposition of nitrite salts 1s essentially
quantitative, as expected. We do not believe that nitrite interference is
a serious problem in the analysis of ambient filter samples however, because
of the low levels of nitrites present in ambient aerosol samples. For
example, our studies in several U.S. cities have shown* ' that particulate
nitrite averages 1-2 percent of the particulate nitrate concentration. Thus
nitrite interference should be minimal for ambient particulate analysis.
Calibration--
In theory, calibration of the chemiluminescent instrument with
known concentrations of NO or N0o» as is done when the instrument is used
for continuous monitoring, should also be sufficient for determining nitrate
if the decomposition and transfer to the monitor is quantitative. To
determine whether this is the case, and also to check the linearity of the
system, a standard calibration curve was prepared using aqueous solutions of
NaNO-. Our initial attempt to prepare such a curve demonstrated that the
response time of the instrument was insufficient to efficiently monitor the
tall sharp peaks resulting from high concentrations of nitrate. This
problem was overcome to a great extent by inserting a 200 ml pyrex ballast
vessel between the Vycor furnace and the chemiluminescence monitor. This
vessel served to lower and broaden the peak, thereby improving the peak
area measurement. Subsequent sections of this report will describe other
means of overcoming the response time problem.
Using the system incorporating the ballast, a calibration curve
was prepared with NaN03 standards. Twenty-six points were used to derive
the curve shown in Figure 10. The best fit equation of the line shown in
the figure is
[N03~] = 5.98(Peak Area)1'05
with [NO,"] in ng and peak area in volt-seconds. The theoretical response
50
-------
10,000
1,000
£
o
too
10
[NO'] = 5.98 (Peok Areo)
i.os
10 IOO
Peak Area, volt-sec
IOOO
Figure 10. Calibration curve for the chemiluminescence technique.
51
-------
curve equation at 25°C and 750 mm Hg 1s
,1.00
[N03~] * 5.93(Peak Area)
The agreement between the theoretical and measured responses over the con-
centration range we studied indicates that the N0~~ decomposition and the
transfer of the decomposition products to the chemiluminescence monitor must
be nearly quantitative
Responses of Various Nitrate Salts--
The nitrate salts expected to dominate ambient filter collections
are NH.NCk, NaN03 and, to a lesser extent, KNO, and other alkali or alkaline-
earth salts. A series of experiments was conducted to determine the response
of the thermal decomposition/chemiluminescence method to various nitrate
salts. All experiments were run under nitrogen to minimize the oxidation of
ammonium compounds to NO , as discussed earlier.
J\
Ammonium, sodium and potassium nitrates were studied since they
are likely to be found in ambient filter samples. Calcium nitrate was
selected because it has a high decomposition temperature; we presume that if
calcium nitrate decomposes and is measured under our experimental conditions,
then the vast majority of nitrates in ambient samples will also be measured.
The results of experiments with these nitrates are shown in Table
14.
TABLE 14. CHEMILUMINESCENCE RESPONSE CURVES
FOR SELECTED INORGANIC NITRATES
[N03'] = A[Peak Area]3
Compound
NaN03
NH.N03
KN03
Ca(N03)2-4H20
Theoretical
No. Samples
Analyzed
7
7
7
6
Range, ng
20-2000
20-2000
20-2000
100-4000
A
1.46±.03
1.63±.03
1.42±.09
1.13±.05
1.84
B
l.lli.Ol
1,08±.01
1.12±.02
1.13±.01
1.00
R
0.99
0.99
0.99
0.99
52
-------
Included in the table are the coefficients, exponents and correlation
coefficients of logarithmic plots of nitrate concentration versus
chemiluminescent response (peak area). The theoretical exponent and co-
efficient values for our experimental conditions are also tabulated.
In all cases, the exponent is greater and the coefficient less than
theoretical. These deviations counteract one another to a great extent,
except for Ca(N03)2. In all other cases, the error involved in using one
of the exponent/coefficient combinations rather than another is small.
For most applications we suggest using NH^NOg for calibraiton since
this is the most likely nitrate salt in ambient samples. The curve
resulting from the NH4N03 data was also closest to the theoretical curve.
The calcium nitrate results suggest incomplete decomposition,
with ^70 percent recovery at the low end of the concentration range and
higher recoveries at higher concentrations. Since the nitrates expected
to be present in ambient samples are nearly completely recovered, and even
a salt with such a high decomposition temperature as Ca(NO,), is largely
-------
TABLE 15. CHEMILUMINESCENT NITRATE METHOD
REPRODUCIBILITY STUDY USING
AMBIENT AEROSOL
Filter Material
Pal If lex QAST
„
M
„
Gelman AE (Port #1)
"
„
„
Gelman AE (Port #2)
„
„
i. ••
Region
Edge
"
Center
«
Edge
"
Center
M
Edge
"
Center
11
yg N03~
551
569
853
942
3650
3530
>3930(a)
>4160(a)
2980
2930
4270
4160
(a) Integrator saturated. Minimum value
shown.
54
-------
filter sampler designs may not result in such serious concentration gradients,
still we have conducted the remainder of our experiments on filter extracts
in order to eliminate this potential source of error.
At this point we reviewed our progress on the thermal decomposi-
tion/chemiluminescence method, with the aim of simplifying and optimizing the
procedure. This review led to some rather significant modifications to the
procedure, described in the following section.
Simplification of the Thermal Decomposition/
Chemiluminescence Apparatus and Procedures—
Our review of the apparatus and procedures suggested several
modifications which would greatly simplify the analysis and substantially
reduce the cost of the apparatus. During the studies just described, we
had also noticed occasional reproducibility problems relating to the time
constant of the chemiluminescence monitor. During the thermal decomposi-
tion of the nitrate, a pulse of NO is sent to the chemiluminescence
X
monitor. It appeared that the monitor did not always respond rapidly
enough to this pulse, so that not all of the NO was integrated. Our
A
initial attempts to overcome this problem made use of a ballast to spread
out the NO peak before it entered the NO monitor. However, because of the
X X
extreme sensitivity of the technique, even small portions of ambient filter
samples tended to overload the system, as seen in Table 15 presented earlier.
We have solved these problems and also greatly simplified the technique by
collecting and diluting the pulse of NO in a Tedlar bag. The bag serves
X
to integrate the sample so that the costly and complex digital integrator
is eliminated from the system.
In practice a liquid sample is injected into a stainless steel
loop, An evacuated Tedlar bag is connected to the system and nitrogen flow
is started through the loop and into the bag. At this point the loop is
heated rapidly to 425°C via two electrodes connected to a stepdov/n trans-
former. The N0~~ decomposes rapidly and the NO is carried into the bag by
O X
the N2 flow. The N« flow is continued for 1-2 minutes to insure that all the
NO has been collected; total volume collected in the bag is 1000-1500 ml.
A
The bag is then connected to the chemiluminescence monitor and the NO con-
X
centration determined. This concentration along with the N« volume is used
to calculate the original NO ~ concentration. The entire analysis require
55
-------
7 minutes. By continuing to dilute the NO collected in the bag, the upper
.A
limit of the technique can be extended almost indefinitely. Since the decom-
position temperature is the same as in the original apparatus and ^ is
again used as carrier gas, the previous results on recovery efficiencies and
interferences should still be valid
The importance of the simplifications described here should be
emphasized. The chemiluminescence apparatus can now be assembled in a few
hours and operated easily by laboratory technicians. The cost of the
apparatus and the time per determination have been significantly reduced.
These advantages come at a cost in sensitivity. However, the modified tech-
nique still has sufficient sensitivity for ambient filter analysis. A
detailed description of the modified apparatus is given below.
Description of the Modified Apparatus--
A schematic of the modified thermal decomposition/chemiluminescence
apparatus is shown in Figure 11.
The apparatus is used to decompose NO-" to NO by heating a liquid
400°C inside of a loop of 1/8" stainless steel tubing. The heat
is generated by passing a 5 volt, high amperage current through the metal
loop. The temperature is regulated by connecting the step-down transformer
to a Variac. A thermocouple is positioned at the site of sample injection.
Nitrogen, used as the carrier and dilution gas, is passed through the con-
version loop and into a Tedlar collection bag. The flow rate is regulated
to obtain the desired dilution of the sample.
The conversion loop was made by wrapping 1/8" stainless steel
tubing twice around a 1/2" diameter rod. Electrodes from the transformer
were placed on either side of the coil. A swagelok tee fitting was installed
on one end of the loop with a septum for sample injection. A non-conductive
type ferrule must be used to insure that the conversion loop is electrically
isolated from ground. Stainless steel fittings and Teflon tubing are used
between the loop outlet and the collection bag to minimize NO adsorption.
A
Quick connect fittings are utilized to facilitate the rapid transfer of the
bag to the NO analyzer.
/\
56
-------
Regulator
en
Sample inlet
Flow control
orfice
Rotometer
Tedlar
collection
bag
Step-down
transformer
Variac
Figure 11. Thermal decomposition apparatus.
-------
Characteristics of the Modified Apparatus-
While the apparatus pictured in Figure 11 is conceptually simple,
several variables, including sample volume, flow rate, accuracy and pre-
cision must be optimized and/or assessed. These variables are discussed
in the following sections.
Sample volume—Table 16 shows the results of sample volume studies,
Different volumes of a 0.099 yg/yl Na NOg standard were injected into the
sample loop, flashed and analyzed as described above.
TABLE 16. EFFECT OF SAMPLE VOLUME ON NITRATE ANALYSIS
(0.099 yg/yl NaNOg standard)
Volume of Sample Injected, yl Measured [NO-"]
1
2
4
7
10
15
20
25
40
0.113
0.084
0.088
0.096
0.094
0.098
0.093
0.077
0.061
The 1 and 2yl injections are not very accurate since the background
NO in the N9 carrier gas makes up a large fraction of the total measured NO
X £ A
at such low sample volumes. Injection volumes which fall in the range 4-20
yl yield reproducible and reasonably accurate results. Above 20yl the
recovery of NO falls off. Additional tests not shown here demonstrate that
A
the upper limit for the sample volume can be extended above 20yl by increas-
ing the sample residence time in the heated sample loop. This is accomplish-
ed easily by reducing the carrier flow rate. However, for most purposes a
58
-------
10-15yl injection volume should be entirely sufficient. As an example of
the accuracy and precision of the method, seven 15pl injections of the
0.099pg/yl NaNCL solution were run. These samples yield an average nitrate
of 0.098±0.005yg/yl.
Flow rate—The flow rate of N2 through the injection loop and
into the Tedlar collection bag is constrained in two ways. First, the
flow must be kept low enough that it does not blow the liquid sample
through the loop. Secondly, the flow should be optimized so that the
total volume of sample in the Tedlar bag is sufficient for chemiluminescence
analysis but not so great as to excessively dilute the NO decomposition
A
products. In practice we found a flow rate of about 200 ml/min and a total
volume of 1.0-1.5 liters to be ideal for our system. Operationally, the
following procedure yielded the most reproducible results:
(1) Inject sample into the conversion loop
(2) Initiate N2 flow through loop and into collection
bag
(3) After 30 seconds apply heating current to loop
(4) Discontinue both flow and heat 5 minutes after
flow was started
(5) Disconnect collection bag from loop and connect to
chemiluminescence monitor. Read the NO concentra-
X
tion after the signal becomes steady.
(6) Allow loop to cool below 100°C before injecting
another sample (this can be done rapidly by
squirting water on the loop)
(7) Periodically determine the system blank by analyzing
the pure water used for filter extractions in the
same manner as above.
59
-------
Accuracy and precision with various nitrates--To further
characterize the refined thermal decomposition/chemiluminescence procedure,
aqueous solutions of three different nitrate salts were prepared and
analyzed by ion chromatography and chemiluminescence. The results of these
analyses are shown in Table 17. Discussion of ion chromatography is reserved
for a later section of this report. However, the agreement between chemilumi-
nescence method and ion chromatography is quite good. Both instruments were
calibrated independently. The chemiluminescence results are 3-5 percent
lower than the ion chromatographic data. This small discrepancy could have
resulted from small errors in preparing calibration standards for one or the
other instrument. Based on the data in Table 17, there are clearly no
significant differences in the extent of decomposition among these three
salts.
The relative standard deviation for the chemiluminescence results
varies from 2 percent for sodium and potassium samples to 4 percent for the
NH.NCL solution. These deviations are based on 3-5 replicates for each
solution.
Since the preliminary experiments with the refined thermal
decomposition/chemiluminescence method seemed promising, a more detailed
set of comparisons was carried out using NH.NOo, KN03» and an actual ambient
filter extract. Solutions of NH^NO., and KN03 were perpared at 5, 50 and
550 ppm (yg/ml) and analyzed in quadruplicate by both ion chromatography and
chemiluminescence. The aqueous extract from a high volume filter collected
on October 19, 1976, during our Phase III field study (to be discussed
later in this report) was also analyzed by both methods. The results of
these experiments are presented in Table 18.
Referring to the data in Table 18, the following points can be
made.
(1) The I.C. results are consistently 3-7 percent below
the prepared sample concentrations. The chemiluminescence
results are less consistent, but in four out of six cases
the chemiluminescent results are closer to the true con-
centration than I.C.
60
-------
TABLE 17 COMPARISON OF CHEMILUMINESCENCE WITH ION
CHROMATOGRAPHIC NITRATE DETERMINATIONS
Compound
NaN03
KN03
KN03
NH4N03
Ion Chromatograph
yg N03-
2.97
2.89
1.93
2.80
Chemi 1 uminescence*
pg N03-
2.82 ± 0.06
2.79 ± 0.06
1.86 ± 0.04
2.73 ± 0.11
* 3-5 replicates
61
-------
TABLE 18. REPLICATE NITRATE ANALYSES
Sample
(Concentration Ion
for N03~)
5 ppm NH4N03
50 ppm NH4N03
500 ppm NH4N03
5 ppm KNO,
J
50 ppm KNO,
J
500 ppm KNO,
3
Los Angeles
Filter -10/19/76
Chroma tographlc
Method
N03", ppm
4.6
4.5
4.6
4.6
4.6 ± 0.1
46.2
46.0
46.5
46.6
46.3 ± 0.3
464
464
464
468
465 ± 2
4.6
4.6
4.6
4.6
4.6 ± 0.0
48.5
48.8
48.8
48.8
48.7 ± 0.2
477
477
479
479
478 ± 1
131
134
132
132
132
Chemi luminescent
Method
N03", ppm
5
5
5
6
5 ± 1
54
50
49
50
51 ± 2
505
519
495
466
496 ±
5
3
4
4
4 ± 1
50
50
48
50
50 ±
466
485
452
476
470 ±
126
132
138
132
139
22
1
14
132 ±1 133 i 5
62
-------
(2) The reproducibility of the ion chromatograph is
considerably better than the chemiluminescence pro-
cedure, especially at low concentrations. This is
due primarily to the fact that milliliter amounts of
sample are injected into the I.C. while microliter
samples are employed in the chemiluminescence pro-
cedure. The relative standard deviation of the I.C.
data is better than 1 percent in all cases but one.
The relative standard deviation of the chemiluminescence
results is better than 4 percent for all cases except
the two low-concentration samples. A 20-25 percent
deviation is observed for those samples. Since those
two samples are based on injection of only 50 ng of
nitrate into the chemiluminescence system, the imprecision
is perhaps understandable. In the case of such low con-
centrations, the direct injection thermal decomposition/
chemiluminescence system will provide much greater
sensitivity and precision.
(3) The average nitrate value for the Los Angeles filter
sample as determined by chemiluminescence is quite
similar to the ion chromatographic result.
Comparison With Ambient Filter Samples--
As a final test of the refined thermal decomposition/chemilumi-
nescence method, filter extracts from 17 high volume filters were analyzed
simultaneously by ion chromatography and chemiluminescence. These filter
samples were collected for 8 hours (0900-1700 PDT) a day at two sites in the
Los Angeles basin. The results of these analyses are plotted in Figure 12.
The "equivalent response" line has also been drawn in the figure. It
seems clear that, over the range of nitrate concentrations represented by
o
these actual ambient filter collections (i.e., 0.3-52 vg/m ), the chemilumi-
nescence and ion chromatographic methods yield similar results. Since
ambient nitrate concentrations will rarely fall outside this concentration
range, these results confirm the utility of the thermal decomposition/
chemiluminescence method for ambient particulate nitrate determination.
63
-------
: I Correspondence Lin*
8 10 12 W 16 18 20 22
Ion Crtromotographic Nitrate, mg/filter
24 26 28 30
Figure 12. Comparison of chemiluminescent and ion
chromatographic methods for ambient
filter samples.
64
-------
Direct Injection with Modified Apparatus--
As a final step in our Investigation of thermal decomposition
coupled with chemiluminescence to determine nitrate in atmospheric parti -
culate samples, we have studied the direct injection of sample decomposi-
tion products into the chemiluminescence monitor. Our earlier studies of
direct injection showed some irreproducibility, especially at high
concentrations, due to a chemiluminescence response time which was too slow
to quantitatively detect the sharp pulse of NO which results from the
A
thermal decomposition. In reinvestigating direct injection we employed
a TECO 14D chemiluminescence monitor and the modified decomposition
apparatus described in the previous section. The TECO 14D has a much
higher flow rate (^1 1pm) and more rapid response than the instrument employ-
ed for the initial direct injection experiments.
Results of the direct injection experiments with the TECO 14D are
shown in Figure 13. A 0.5pl sample volume was used for these experiments.
Digital integration was employed for peak area determination. The data
points are scattered about the theoretical line, showing good linearity.
Linear response up to 1000 ppm N0~~ has been observed. The technique can
detect about 3 ng NO-" in a 15pl sample; approximately 30 samples can be
analyzed in an hour. With this sensitivity, even the lowest rural Hi Vol
filter nitrate concentrations can be determined.
ION CHROMATOGRAPHY
Shortly after our investigation of the thermal decomposition/
.chemiluminescence procedure was initiated, a new analytical technique was
introduced to the atmospheric analysis community. The technique was called
ion chromatography. To evaluate the utility of this new technique for
atmospheric particulate nitrate determination, our original contract was
extended by several months. During this time several reports and much data
have accumulated on the use of ion chromatography for atmospheric particu-
late analysis. In this section of the report we will briefly review these
studies and describe our investigations of this new technique for ambient
particulate nitrate determination.
65
-------
281 O 7-
20
E /
/
•o
3 -S *>- •- Thaortticol
10
O /
Q/
I __ I
10 20 30
Sample LNO j]«
Figure 13. Results of direct injection experiments
66
-------
Background
The technique now referred to as "ion chromatography" was first
(25)
described in detail by Small, Stevens and Baumanv ' in 1975. Its appli-
cation to the analysis of atmospheric particulate samples was subsequently
described by Rich^26^ and by Mulik^27^ at an EPA symposium in 1976. The
technique is based upon classical ion exchange chromatography combined with
a novel combination of resins and a universal conductivity detector. In
conventional ion exchange chromatography with continuous effluent monitoring,
the ionic species which are eluted from the chromatographic column are
usually monitored by spectrophotometric methods. However, a large number of
important anions and cations can not be determined in this manner due to
their lack of appropriate chromophores. An ideal means of monitoring these
ions might involve the conductance of the ions. Since ionic solutions must
be used as eluants, however, the conductance of the eluant would ordinarily
swamp the conductance of the sample ions. Small and coworkers overcome
this difficulty by employing a combination of ion exchange resins to remove
or neutralize the eluant ions, leaving the sample ions as conducting species
in a non-conducting background. Since conductance is a universal property
of ions in solution and is directly related to ion concentration, a simple
and almost universal technique for determination of ion concentrations
(25)
results. Details of the procedure may be found in Small, et al. '
Determination of Nitrate in Atmospheric
Samples by Ion Chromatography
(25)
The original report by Small, et al. ' demonstrated that nitrate
could be separated and determined by ion chromatography. Subsequent reports
by other investigators have characterized and documented the technique for
determining nitrate in atmospheric particulate samples.
(27)
Mulik, et al. ' described the successful use of ion chromatography
for analysis of water soluble sulfate and nitrate in ambient particulate
matter. They employed a D-ion-X Model 10 ion chromatograph with a 0.5yl
sample loop. A minimum detectable quantity of O.lvg/gl for sulfate and
nitrate was reported with the 0.5yl loop. Relative standard deviations of
3 percent sulfate and 1 percent nitrate were obtained for replicate samples
at concentrations of 5yg/ml.
67
-------
Otterson'2°) reported on the use of ion chromatography for
determination of nitrate and other anions from upper atmospheric filter
samples. The filters were collected as part of NASA's Global Air Sampling
Program. The author reported success at determining N0~~ as well as SO/",
F~, and Cl" at microgram levels and below. He described detailed procedures
for cleaning and purifying the experimental equipment in order to obtain
maximum sensitivity.
(29)
Lathousev ' has described Battelle's experience with ion
chromatography for N0~~ and SO*" determinations. Much of the N03~ data
she discussed was collected during this program. These data, along with
some more recent results, are presented below.
Experimental
A D-ion-X Model 10 ion chromatograph was employed in this investi-
gation. The eluant was 0.003M NaHC03 and 0.0024M Na2C03. A O.lml sample
loop was used. Blank filter and water samples were run between each sample
set. Filter samples were shredded and leached with deionized water on a
steam bath for 2 hours. The samples were then cooled and filtered through
0.22 millipore filters. For some ambient samples, both hot water leaching
and ultrasonic extraction techniques were employed. The results of these
extractions are shown in Table 19. The agreement between the extraction
techniques is excellent for sulfate and reasonably close for nitrate, with
the exception of sample number 4. These data, when combined with the round-
robin testing extraction and analysis results to be presented shortly, confirm
the validity of the extraction procedure for nitrate and sulfate.
Accuracy and Precision
One evaluation of the accuracy of the ion chromatograph and
extraction procedures is based on an Environmental Protection Agency inter-
laboratory study of nitrate and sulfate analyses. In this round-robin test,
strips of glass fiber filters were sent to 60 laboratories. These labora-
tories extracted and analyzed the filters according to their normal operating
procedures. Thus many extraction procedures and analytical methods were
employed in the study. The results of this study are shown in Table 20.
68
-------
TABLE 19, HOT LEACH VS ULTRASONIC FILTER EXTRACTION
Sample No.
1
2
3
4
5
Extraction Method
Hot Leach
Ultrasonic
Hot Leach
Ultrasonic
Hot Leach
Ultrasonic
Hot Leach
Ultrasonic
Hot Leach
Ultrasonic
Nitrate, ng
1.57
1.60
1.96
2.07
0.87
0.95
2.64
4.40
1.28
0.80
Sulfate, ng
2.32
2.50
3.05
3.16
2.35
2.73
12.8
12.3
16.1
16.5
TABLE 20. RESULTS OF EPA PERFORMANCE AUDIT FOR NITRATE
(concentrations in yg/m3)
Sample No.
1
2
3
4
5
6
Battelle
Ion Chroma tograph
11.780
11.100
5.250
5.420
2.250
7.650
Sample Range
10.374-11.466
10.374-11.466
4.912-5.428
4.912-5.428
1.843-2.037
7.011-7.749
Target Range
9.282-12.558
9.282-12.558
4.395-5.945
4.395-5.945
1.649-2.231
6.273-8.487
69
-------
Results which fall within the sample range indicate no detectable error.
Results within this range are the best that can be expected under the
conditions of the audit. The target range reflects the greatest variability
that is expected during the audit, with the extraction and analytical process
still operating properly. Results within this range indicate that the ex-
traction and analytical procedures are valid. All but one of the ion
chromatograph results fell within the target range and 4 of the 6 samples
fell within the sample range. These results demonstrate the validity of
both the extraction procedure and the ion chromatographic method.
A further evaluation of the accuracy of the ion chromatographic
procedure made use of the widely employed brucine method as a reference.
In these experiments, filter samples were extracted by hot leaching. The
results of the brucine and ion chromatographic nitrate determinations on
these extracts are presented in Table 21. The results show generally good
agreement between the two techniques. The ion chromatographic results
average about 3 percent lower than the brucine method. The relative
deviation between the two methods averages about 9 percent.
Additional information on the accuracy and precision of the
ion chromatographic procedure was obtained by analyzing prepared samples
of two different nitrate salts and the extract from a high volume filter
sample collected during the Phase III Los Angeles field study. Three
different concentrations of each nitrate solution and the filter extract
were run in quadruplicate by ion chromatography. The results of these
analyses were shown earlier in Table 18 in the section on chemiluminescence.
The ion chromatographic results were 3-7 percent below the prepared solution
concentrations. The precision of the ion chromatographic data is excellent.
The relative standard deviation is better than 1 percent in all cases and
considerably better for most of the samples.
Comparisons of ion chromatography with our thermal decomposition/
chemiluminescence method were shown earlier in Table 18 and Figure 12.
Sensitivity
The sensitivity of the ion chromatograph with a new column and
O.lml sample loop averages 0.09yg/chart division. This sensitivity
projects a minimum detection limit of approximately 250 ng.
70
-------
TABLE 21, COMPARATIVE NITRATE ANALYSES
(yg N03"/Filter)
Sample
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Brucine Method
3.6
<5
6
11.0
15.0
18.8
24.0
37.0
45
46
58.2
117
122.5
190
240
342.5
680
1,060
1,650
15,600
22,000
Ion Chromatography
2.5
5
5
10.0
20.0
20.0
25.0
38.5
47
45
65.0
140
145
211
245
750
750
1,060
1,650
15,250
20,870
71
-------
With the larger 0.5ml sample loop, the detection limit is less
than lOOng. As the column ages, the sensitivity changes considerably. As
the column degrades with use, the retention times become shorter and the
peaks taller. Thus the sensitivity improves at the cost of resolution.
This variation in sensitivity with column age is shown in Figure 14.
Although the sensitivity data in Figure 14 are based on peak height
measurements, comparison of peak areas shows a similar trend. Because of
this effect, standards must be run frequently and the column replaced as
the resolution degrades. We currently run approximately 800 samples before
column performance deteriorates to the point where a new column is required.
PHASE II SUMMARY
Our objective in this phase of the program has been to screen
several prospective new methods for nitrate determination in atmospheric
aerosol samples and to develop a method which is rapid, sensitive,
selective, accurate and reproducible. A secondary goal has been to more
fully characterize the new ion chromatographic method for nitrate
determination.
Our initial screening of potential new methods for particulate
nitrate determination included a gas-sensing electrode procedure and a
thermal decomposition/chemiluminescence method. The investigation of the
electrode procedure was terminated when it became apparent that the con-
version of nitrate to nitrite (required for the electrode measurement)
could not be carried out quickly enough on large numbers of samples to
meet our criterion for rapid measurement.
The thermal decomposition/chemiluminescence method was developed,
refined and extensively investigated. Two modes of operation were investi-
gated. For greatest sensitivity, the nitrate sample can be decomposed by
rapid heating to M25°C and the decomposition products drawn into a nitrogen
oxides chemiluminescence monitor. The chemiluminescence response is
integrated, the peak area being directly related to sample concentration.
For somewhat less sensitive but simpler and less expensive analyses, the
decomposition products can be integrated by collection in a Tedlar bag,
72
-------
after which the NO concentration in the bag is determined by chemilumines-
j\
cence. This concentration is also directly related to the amount of nitrate
in the sample. No electronic integration apparatus is required in this mode.
The detection limit of the former mode of operation is better than 10 ng N03"
in a 15 yl sample. The latter procedure can detect less than 50 ng NO./ in
a 15 pi sample. The thermal decomposition/cnemiluminescence method responds
to N0~~ as well as N0q~, although this is not a significant interference for
+
atmospheric particulate samples. Interferences due to NH^ and organic
nitrogen compounds are eliminated by decomposing the sample in a nitrogen
atmosphere. The apparatus can be assembled in 1-2 days and operated by a
laboratory technician. Analysis time (excluding sample preparation) is
approximately 7 minutes.
Our studies of ion chromatography indicate this technique is
rapid, sensitive, selective, accurate and highly reproducible. With a 0.5
ml sample loop, ion chromatography can detect less than 100 ng of nitrate in
a sample. The accuracy and reproducibility are excellent if the instrument
is standardized frequently. Nitrate can be determined in the sample in
less than 10 minutes. However, nitrate, sulfate, fluoride and chloride can
all be determined in the same sample in less than 15 minutes. For atmospheric
samples this is a major advantage over all other existing methods. Indeed,
this is such a significant advantage that ion chromatography will very likely
become the dominant method for atmospheric filter sample analysis. Our
investigation indicates that the method is worthy of this growth.
73
-------
456
Weeks of Use
Figure 14. Nitrate Sensitivity vs Column age.
74
-------
SECTION 4
PHASE III: FIELD EVALUATION OF SAMPLING
MEDIA FOR NITRATE COLLECTION
In keeping with the program objective to investigate the effects
of environmental variables on nitrate sampling, a laboratory study was con-
ducted in Phase I of the program to determine the effects of filter com-
position, gaseous pollutants, sampling rate, sampling time, humidity,
filtrate composition and storage time on particulate nitrate collection.
While nitrate collection methods other than high volume sampling were
considered, the Hi Vol method was given the greatest emphasis since it is
already in such widespread use throughout the atmospheric sampling community.
A number of different filter materials were screened in the lab-
oratory in an effort to uncover a material which would efficiently collect
particulate nitrate with minimal interference due to trapping of gaseous
nitrates or nitrate precursors. Subsequent to the laboratory screening
studies it was deemed advisable to evaluate selected filter materials
under actual field conditions and, at the same time, try to identify
the variable or combination of variables which leads to artifact nitrate
collection on certain filters. The ideal location for such a study is
Los Angeles, where the precursors to artifact nitrate should be at a
(4)
maximurtr . These conditions should provide the most severe test of filter
materials. Since we were just organizing a field program in the Los Angeles
area for another EPA project. "The Fate of Nitrogen Oxides in the Atmosphere
and the Transport of Oxidant Beyond Urban Areas (Contract No. 68-02-2439)" a
joint program was undertaken.
Experimental
The field program was conducted between October 15 and November
16, 1976. The overall study involved three ground stations and an Instru-
mented aircraft. However, all the data pertinent to the present study were
75
-------
collected at our base station at Cable Airport in Upland, California. The
base station consisted of the Battelle Mobile Air Quality Laboratory. The
variables monitored at this location are shown in Table 22. The techniques
used for these measurements are given in Table 23.
The base station was located approximately 37 miles east of down-
town Los Angeles and less than 5 miles south of the base of the San Gabriel
Mountains at an elevation of 446 meters above sea level. The location was
selected because it lies along the normal wind trajectory from downtown
Los Angeles. We expected to observe frequent episodes of well aged photo-
chemical smog at this site. The meteorology varied considerably during
the field program, from warm hazy weather to hot, dry, very clean desert
wind conditions.
Aerosol samples were collected from 0900-1700 PST each day by
three high volume samplers operating within the mobile lab. Samples were
collected from 10 meters above ground using three 15cm diameter alumlmum
stacks. The aerosol was collected on 152mm diameter filters backed by a
stainless steel fritted disk. One of these samplers was outfitted with a
cyclone to remove particles with diameters greater than 2.0 ym. High
purity quartz mat filters (Pallflex QAST) were always used with this
3
sampler at a flow rate of 0.57 m /min.
The other two samplers were operated simultaneously in an
3
identical manner at known flow rates of approximately 0.75 m /m1n.
Pressure drop measurements were made at the beginning and end of each
day's sampling to correct for day to day fluctuations in flow rate.
Initial calibration of the high volume blowers was performed with a
calibrated venturi. The filter material used in these samplers was
varied from day to day among two types of quartz and two types of glass
fiber filters. One or the other sampler (randomly varied) always operated
with Pall flex QAST high purity quartz, which was used as a comparison
standard. Thus one total aerosol sample and the small particle (<2.0 ym)
sample are always collected on QAST and are directly comparable. Besides
Pall flex QAST, the other three materials used in the high volume samplers
include "EPA type" Gelman AA glass fiber filter, Gelman A glass fiber
filter and a high purity quartz fiber filter developed by Arthur D. Little
under contract to EPA. Each filter was preweighed in the laboratory at
40 percent relative humidity and then stored in an individual glasslne
76
-------
TABLE 22. MEASUREMENTS AT UPLAND STATION
Meteorological Measurements Gas Measurements Aerosol Measurements
Temperature
Relative Humidity
Solar Intensity
Wind Speed
Wind Direction
°3
NO
N0x
NH3
HN03
PAN
Fluorocarbon-11
so2
CO
CH4
NMHC
C2H2
Mass Loading
Nitrate
Sulfate
Ammonium
Total Carbon
Total hydrogen
Total nitrogen
Nitrate £ 2.0 ym
Sulfate £ 2.0 ym
Ammonium <_ 2.0 y
Mass < 2.0 ym
77
-------
TABLE 23. ANALYTICAL METHODS
Determination
Instrument/Method
Calibration
U3
NO
HN03
PAN
THC
CH,
REM Chemlluminescence
Monitor
Bendix Chemiluminescence
Monitor
Bendix Chemiluminescence
Monitor
Battelle Micro-Coulometrlc
Instrument
Electron Capture Gas
Chromatograph
Beckman Model 6400
Chromatograph
IX Neutral Buffered KI
NBS Cyl. "NO in N2"
NBS Permeation Tube
Actual Samples of HCL and
HN03
Actual Samples Referenced to
I.R.
NBS Cyl. "Propane in A1r"
Actual Sample Referenced to
NBS Standard
CO
NH3
SO,
Fluorocarbon-11
NH4+
NO,"
Wind Speed
Wind Direction
Temperature
Relative Humidity
Solar Intensity
Dual Catalyst Chemllumi-
nescence
Flame Photometric
Electron Capture Gas
Chromatograph
High Volume Sampling/
D-lon-X Ion Chromatograph
MRI Model 1071 Weather
Station
Eppley 180° Pyrhellometer
NBS Cyl. "CO in A1r"
Matheson Calibration Cylinder
Permeation Tube
Permeation Tube
Actual Samples
78
-------
envelope enclosed in a sealed polyethylene bag. After sampling, the
filters were returned to their glassine envelopes and resealed in plastic
bags. On return to the Columbus laboratories the filters were equil-
ibrated at 40 percent RH and reweighed. The filters were then partitioned
for analysis.
On many days during the study up to five different 47 mm filters
were collected simultaneously from 0900-1700 PST. These filters were
held by five 47 mm filter holders which were arranged circularly around
and connected to one of the aluminum high volume stacks. All five filter
holders were connected to a common vacuum manifold which was evacuated by
a high capacity Gast pump. Pressure drop measurements were made several
times each day on each filter. Flow vs pressure drop calibration curves
were prepared for each filter type with a calibrated dry test meter and
used with the pressure drop measurements to calculate flow. Total
volume of air sampled varied between 3-11 m depending on filter type.
After collection, each filter was stored in a clean capped Petri dish.
All filters were extracted in deionized water over a steam bath.
As described earlier in this report, this extraction procedure is
quantitative for atmospheric particulate nitrates. The filter extracts
were analyzed for N03" and SO^" by ion chromatography. A Dionex Model 10
ion chromatograph was operated in the manner described in Section III of this
report.
RESULTS AND DISCUSSION
As mentioned earlier, three types of filter collections were made
during the Los Angeles field study. These will be designated High Volume
o
(152 mm diameter filters, 360 m ), Low Volume (47 mm diameter filters, 3-11
m ) and Small Particle U 2.0 ym)samples. Both nitrate and sulfate were
determined on these filters, and this section of the report will discuss both
anions. However, primary emphasis will be placed on nitrates.
Results of High Volume Collections - Nitrate
The results of the duplicate Los Angeles filter collections for
nitrate are shown in Table 24 where the nitrate concentrations from the QAST
filters and the test filters are compared on a daily basis. On some days two
79
-------
Date
Oct.
Oct.
Oct.
Oct.
Oct.
Oct.
Oct.
15
18
19
20
21
22
25
Filter
ADL
OAST
AA
QAST
AA
OAST
AA
OAST
AA
OAST
AA
OAST
A
OAST
TABLE
N03~'
wg/m3
1.6
1.6
14.4
0.39
17.0
1.2
28.7
2.3
18.8
0.82
11.2
0.49
38.4
0.78
24. COMPARISON OF NITRATE COLLECTED
VARIOUS FILTER TYPESl9) (°l
Date
Oct.
Oct.
Oct.
Oct.
Nov.
Nov.
Nov.
26
27
28
29
1
2
3
Filter
ADL
OAST
A
OAST
AA
QAST
ADL
QAST
QAST
QAST
A
OAST
ADL
QAST
N03~«
vg/m3
1.3
2.1
3.9
0.52
9.1
1.9
1.8
2.9
1.7
1.7
9.1
1.6
0.68
1.1
ON
Date
Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
Nov.
4
5
9
10
11
12
15
Filter
A
QAST
A
OAST
A
QAST
ADL
QAST
QAST
QAST
A
QAST
A
OAST
N03-,
wg/m3
3.0
1.1
6.1
0..98
8.4
3.1
1.9
2.3
2.0
1.9
6.0
1.3
14.3
3.0
(a) AA - "EPA Type" Gelman AA.
A - Gelman A
ADL - High Purity Quartz filter developed by Arthur D. Little under contract
to EPA.
QAST - Pall flex QAST.
(b) Nitrate blanks for these filters were always <0.05 mg.
80
-------
QAST filters were collected simultaneously on the two high volume samplers to
test the equality of the samplers. The symmetry and equality of the high
volume sampling systems are confirmed by inspection of the nitrate results
from November 1 and 11 when Identical quartz filters were employed in each
sampler. Two different types of quartz fiber filters were compared on October
15, 26, and 29 and November 3 and 10. The two types of quartz filter generally
collected nitrate concentrations of similar magnitude, with QAST tending to
collect somewhat more nitrate than ADL.
The differences in nitrate concentration collected on glass versus
quartz filters and the ratios of nitrate on glass and quartz are shown in
Table 25. Since the filtration efficiencies for quartz fiber and glass fiber
filters are the same, the differences 1n nitrate concentration In Column 1 re-
present artifact nitrate. For Gelman AA, artifact nitrate varied from 7.2 to
3 3
26.4 yg/m . Artifact collection on Gelman A varied from 1.9 to 15.3 yg/m .
On the average Gelman A showed less artifact nitrate than Gelman AA. This
may be attributed in part to the fact that less severe pollution conditions
prevailed during most of the Gelman A collection days.
Plots of nitrate collected on the two types of glass filters versus
the QAST filter are shown in Figure 15. Also shown is a plot of ADL versus
QAST for the days on which these two quartz filters were collected simul-
taneously. The slope of the ADL vs QAST Hne is about 0.7 and the intercept
approaches zero, suggesting some artifact collection on QAST. Our Phase I
laboratory studies also suggested that QAST 1s slightly more susceptible to
artifact collection than ADL. The intercept of the A vs QAST plot is also
zero but the slope exceeds 5. Thus, on the average the A filter collects
more than 5 times the nitrate collected by the QAST filter.
In constructing the AA vs QAST plot, one data point was ignored
since it fell well outside the pattern formed by the other points. It is
interesting to note that desert wind conditions prevailed on this day and
that the relative humidity was less than half that measured on the other
Gelman AA collection days. The effects of moisture and other environmental
variables will be explored more fully later in this section. The slope of
the AA vs QAST plot is nearly 8 and the intercept is 8.5. This curve Implies
considerably greater artifact collection for AA than A. Again, however, it
81
-------
30
28 -
26 -
24
22
20
18
I* 16
v
- 12
10
X Gelmon AA
O Gel man A
D ADL Quartz
0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
QAST [N03],/ig/m3
Figure 15. Simultaneous high volume nitrate collection results.
82
-------
TABLE 25. COMPARISON OF NITRATE COLLECTED
ON QUARTZ AND GLASS FILTERS*
Gelman AA-QAST
(yq/nr)
10-18-76 14.0
10-19-76 15.8
10-20-76 26.4
10-21-76 18.0
10-22-76 10.7
10-28-76 7.2
Gelman AA/OAST
36.9
14.2
12.5
22.9
22.9
4.8
Gelman A-QAST
(
Gelman A/QAST
10-27-76
11-02-76
11-04-76
11-05-76
11-09-76
11-12-76
11-15-76
3.4
7.5
1.9
5.1
15.3
4.7
11.3
7.5
5.7
2.7
6.2
5.9
4.6
4.8
The results from 10/25/76 have been excluded here
and in subsequent discussions due to a questionable
analysis.
83
-------
should be emphasized that the conditions thought to be favorable for artifact
collection (e.g., high precursor concentrations and high relative humidity)
occurred much more frequently during the Gelman AA collections. These
differences in pollution conditions may partially explain the apparently
large differences between the two glass filters.
The question arises whether the differences in nitrate collected
on glass versus quartz filters could be attributed to differences in
particle collection efficiencies. In general, both glass and quartz
fiberous mat filters are claimed to have the same collection
efficiencies are silimar is demonstrated by our sulfate results on
the same filters. While the sulfate data will be discussed in more
detail shortly, we observed similar concentrations on the various glass
and quartz filters. Since more than 90 percent of the sulfate mass was
found in the small particle size range (<2.0 ym), sulfate collection should
represent a fairly severe test of filter efficiency. Because the glass
and quartz filters exhibited similar collection efficiencies for sulfate,
we can be assured that particle collection efficiency is not responsible
for the large differences in nitrate collected on glass versus quartz.
Now that artifact nitrate collection has been documented under
actual field conditions, it is enlightening to compare our Phase I lab-
oratory results with the field data. Based on the saturation experiments
and some other Phase I results, we concluded earlier that several of the 47
mm glass fiber filters saturate at 300-400 yg artifact nitrate. Due to
differences in collection efficiency, saturation required about 8 mg NOg or
300-400 yg HN03- Experiments also suggested that the NOp collection
efficiency increases with humidity, so that somewhat less than 8 mg N0«
should result in filter saturation at ambient humidities. Extrapolating
these values to the larger filters employed in our high volume field sampling,
we find that ~140 mg N02 or ~6 mg HNO- is required to saturate the larger
glass filters. Actually, less N02 would be required because of the higher
N00 collection efficiency at ambient humidities. During our field study, N09
3 3
averaged 100 yg/m and HNO, averaged about 7 yg/m . Therefore a typical glass
3
fiber filter which sampled 360 m of air was exposed to "36 mg N0? and -2.5 mg
84
-------
- The combination of these two gases should have resulted in about 70
percent filter saturation on the average. The filters may have been even closer
to saturation due to higher humidities. Again extrapolating from the 47 mm
filter laboratory results, 70 percent saturation of the larger filters should
result in collection of ~4,400 yg artifact nitrate or
4,400 yg N03" -_ 12 yg/m3
360 m3
of artifact nitrate. From Table 25, we actually observed an average of ~11
yg/m of artifact nitrate, in rather close agreement with the prediction based
on the laboratory study. While this crude exercise stretches the laboratory
results beyond their intended purpose, it does serve to tie the laboratory and
field studies together and demonstrates a fundamental correspondence in the
results of the two.
Note that a high volume sampler collecting over a full 24 hours
(rather than our 8 hour collections), would typically be completely saturated.
As discussed in Section 2, a saturated 8" x 10" glass fiber filter which samples
2400 m of air would suffer ~5 yg/m artifact nitrate collection. While such
filters may not normally saturate in many urban areas, still it is clear that
even partial saturation results in significant errors in particulate nitrate
measurement.
The nitrogen balance on the four high volume filter types is shown
in Table 26. Within the rather wide scatter of the data the total amount of
nitrogen on the filters (determined by an independent combustion/thermal
conductivity procedure) is accounted for by the nitrogen present on the
filter as nitrate and ammonium. The wide scatter in the data is caused by
high total N backgrounds for the filters relative to the total N collected
by the filters. Under these circumstances slight variations in the filter
background or small errors in the total N analysis can have a major effect
on the nitrogen balance calculation.
As seen in Table 26 the fraction of total nitrogen due to nitrate
in 3-4 times higher on glass than quartz filters. This is an expected result
of the collection of artifact nitrate by the glass filters.
85
-------
TABLE 26. HIGH VOLUME FILTER NITROGEN BALANCES*
Filter Type
ADL
QAST
Gel man AA
Gelman A
Number of
Filters
4
16
6
8
Amount of Total N + Amount of Total N
Accounted for by NO," & NH. , Accounted for by NO,",
Percent Percent
96 ± 40
84 ± 22
111 ± 18
87 ± 20
17 i 11
18 ± 14
63 ± 29
78 ± 24
*
Filters for which the total N values were less than twice the filter background have
been excluded.
86
-------
Based on comparisons of the small particle samples with the total
aerosol collections on QAST, about one half the nitrate mass was found in
the particle size fraction less than 2.0 vim. The filter nitrogen balance for
the small particle collections showed considerable scatter, but averaged 97
percent. If artifact nitrate due to a gaseous precursor were responsible for
a large fraction of the nitrate observed on the QAST high volume filters, then
the ratio of nitrate in the small particle size range to the total collected
nitrate would be close to unity (since the gaseous precursor would pass
through the cyclone). Instead, we observe a ratio of 0.5. This means that
at least half of the nitrate collected on the QAST total filters must be
true particulate nitrate of diameter * 2.0 ym. Of course, considerably more
than half of the nitrate collected on the QAST total aerosol filters must be
true particulate nitrate since there must be small particle nitrate in the
Los Angeles atmosphere. Indeed, most of the nitrate collected on the QAST
filters could be true particulate, but the size distribution results can not
confirm or deny this.
If it is assumed that the high concentrations of artifact nitrate
found on the glass filters are due to a gaseous precursor, then such high
concentrations would also be observed on glass filters backing up particle
sizing devices. This would lead to the conclusion that the great fraction
of ambient particulate nitrate mass is in the small particle size range, when
in fact the actual particulate nitrate might be distributed in an entirely
different manner. Since the health effects of nitrate particulates may well
depend on size distribution, future studies must make certain that artifact
collection not bias the size distribution results.
Results of High Volume Collections-Sulfate
Twenty one high volume sulfate sample pairs were
collected at the Cable Airport site. Results of these collections are
shown in Table 27. In these experiments, the primary reference filter was
Pallflex QAST. The simultaneous collections on QAST November 1 and November
11 again confirm the equality of the sampling systems. Pairwise comparisons
between QAST and ADL Microquartz, Gelman AA (EPA), and Gelman A are shown in
Table 27. Also shown in the table are calculated values of the artifact
sulfate, based on S02 and RH, as reported by Coutanv '. Calculated artifact
t 87
-------
TABLE 27. HIGH VOLUME SULFATE RESULTS
Date
10/15
10/18
10/19
10/20
10/21
10/22
10/25
10/26
10/27
10/28
10/29
11/1
11/2
11/3
F1lter
ADL
QAST
AA
QAST
QAST
AA
AA
QAST
AA
QAST
AA
QAST
QAST
A
ADL
QAST
A
QAST
AA
QAST
ADL
QAST
QAST
QAST
QAST
A
QAST
ADL
S04"jjg/m3
27.2
36.9
31.8
36.2
32.1
32.0
18.1
18.0
18.5
18.4
12.1
12.1
19.8
21.5
2.5
3.6
5.6
1.4
5.2
2.7
3.9
5.2
3.4
3.5
3.8
7.5
3.2
2.0
AS04"
measured
-9.7
-4.4
-0.1
0.1
0.1
0.0
1.7
-1.1
4.2
2.5
-1.3
0.1
2.7
-1.2
,^4 J Comment
calculated
0.0
1.1 No S02 data assume 50 ppb
0.95 No S02 data assume 50 ppb
1.0
1.1
1.1
1.0
0.0
0.84
0.84
0.0
0.0
0.81
0.0
88
-------
TABLE 27. (Continued)
Date
11/4
11/5
11/9
11/10
11/11
11/12
11/15
Filter(a)
A
QAST
A
QAST
A
QAST
ADL
QAST
QAST
QAST
A
QAST
A
QAST
MEAN
S04" v /m3
4.9
1.4
7.2
4.1
11.8
9.2
13.4
18.1
11.5
11.6
10.4
5.2
9.4
5.6
AS04C
measured
3.5
3.1
2.6
-4.7
0.1
5.2
3.8
0.85 ± 2.7
*S04 . . Coranent
calculated
0.78
0.83
0.81
0.0
0.0
1.5
0.93
0.65 ± 0.49
(a) QAST - Pallflex QAST Quartz
ADL ADL Microquartz
AA = Gelman AA (EPA Type) Glass
A = Gelman A Glass
89
-------
sulfate values are generally smaller 1n magnitude than those actually
measured, but seem to correlate qualitatively except in the comparison of QAST
and ADL filters. In this case, the ADL filter consistently yielded lower
total sulfate levels than the QAST filter. Coutant'32) has reported that the
collection of artifact sulfate by these two filter media was nearly equivalent.
The current results may be due to batchwise variations in the nature of the
QAST filter.
The sulfate results are shown in the form of average differences
in Table 28. From this table, it is clear that the Gelman A filter yielded
higher sulfate values than the QAST filter. The Gelman AA filter is
approximately equivalent to the QAST, although there is somewhat more scatter
in the data for this pair. Finally, the ADL filter response is consistently
less than that of the QAST.
(^
These differences in filter response for sulfate ate further
illustrated in Figure 16. This figure suggests that the responses of the
Gelman AA and ADL filters are simple multiples of those obtained for the QAST
filter. However, the data for the Gelman A filter do not extrapolate through
the origin, suggesting that an additional source of error exists with the
use of this filter. This behavior is similar to that observed with the AA
filter for nitrate.
Low Volume Filter Sample Results - Nitrate
On Many days during the Los Angeles field study up to five different
47 mm filter samples were collected simultaneously from 0900-1700 PST at the
mobile laboratory at Cable Airport. These samples were collected from one
of the aluminum high volume sampler stacks, which sampled air approximately
10 meters above ground.
These filters were collected at low flow rates, the rate depending
on the resistance to flow of each filter type. The low resistance filter
(quartz and glass) were collected at flows between 0.012 and 0.023 m /min.,
with occasional exceptions, and the high resistance filters (Duralon, Mitex,
o
Fluoropore) at flow rates around 0.008 m /min. At these flow rates, about
4 m3 of air passed through the high resistance and 6-11 m through the low
resistance filters. With nitrate averaging only 1.5 yg/m (as determined
from the high volume collections on quartz), these low volume filters were
generally collecting only low microgram quantities of nitrate. In many
90
-------
36
34
32
30
28
26
I*
J5 22
if
S is
£
u
V
O 14
12
10
8
6
4
2
b*
CO
O Gelmon A
X Gel man A A
D ADL Quartz
Slope = 0.9
I
I
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38
Collected on QAST,/ig/m3
Figure 16. Comparison of filter sulfate responses.
91
-------
TABLE 28. AVERAGE PAIRWISE SULFATE DIFFERENCES
Filter Pair Sulfate Difference, yg/m
QAST-QAST 0.1
ADL-QAST -2.1^
Gelman AA - QAST -0.3^
Gelman A - QAST 3.4^
(a) significant at 90% confidence level
(b) not significantly different from zero
(c) significant at 99.5% confidence level
92
-------
cases these low levels approached the limits of detection of the analytical
methods being used. The concentrations of nitrate on many of the filters are
not much greater than the filter nitrate blank, so any contamination or small
filter-to-filter variations in nitrate blank can have a major effect on the
calculated atmospheric particulate nitrate concentration. For these reasons
the low volume filter data are not highly accurate and should be viewed with
caution.
A bar graph showing the ratio of nitrate collected on several
different filter materials relative to nitrate on ADL quartz is presented
in Figure 17. The number of low volume filter pairs available for this com-
parison is shown for each filter type in parentheses. Mitex and fluoropore
filters were included in these filter experiments but only two comparisons
with Mitex and one with Fluoropore are available, and these occurred on days
when the nitrate concentration was very low. The resulting large uncertainties
have convinced us to forego quantitative comparisons with these two filters.
Qualitatively, both filters seemed to collect greater amounts of nitrate than
the quartz filters. This is surprising in light of our Phase I laboratory
results, but may be due to the relatively large uncertainties in the data and
the limited number of samples for comparison.
Figure 17 indicates that, for the environmental conditions, flow
rates and face velocities employed during these experiments, Pallflex quartz
collects slightly more nitrate than ADL quartz. The glass fiber filters all
collect between 1-1/2 and 2-1/2 times as much nitrate as ADL quartz. The
Duralon filter collected the most nitrate of all, as expected. We have
(33)
demonstrated previously that nylon filters quantitatively collect
gaseous nitric acid. Nylon filters were used during the study for this
purpose. Comparison of the nylon and ADL quartz filters suggests an average
nitric acid concentration of about 0.002 ppm during the three days when data
are available for comparison.
The days on which any given filter type was used for low volume
sampling were randomly distributed during the field study. Therefore, due
to daily variations in environmental conditions (precursor concentrations, RH,
etc.) some filter types may have been exposed to more severe conditions than
93
-------
o
o
<
o
u
O
o
o
o
x
Q>
O
O.
JO
u
a
£
a>
O
a>
O
in
<
o
"Z
o
O
O
fO
UJ
•o
o
u
w
a,
(A
2 3
t n
O
o> 2
S1
w
m
Filter Type
Figure 17. Relative nitrate collection for low volume filters.
94
-------
others. In order to increase the number of filter pairs available for compar-
ison, and thereby reduce potential distortions in the Figure 17 data due to
daily variations in environmental conditions, the ratios were recalculated
using either ADL or Pall flex quartz as the comparison standard. On days when
data were available for both quartz filters, the average was used. Use of
test filter/quartz ratios increased the number of days available for comparison,
but the relative results were almost identical to those shown in Figure 17.
A quantitative comparison of the low volume and high volume nitrate
results would be misleading and will not be undertaken due to the aforementioned
inaccuracies in the low volume data. The low vol sampling was designed to
screen a number of different filter materials for artifact collection under
actual field conditions, and it served that purpose. However, such low levels
of nitrate were collected during the 8-hour sampling period, that analytical
imprecision and variations in filter blanks resulted in considerable scatter
in the data. For this reason, we recommend that even the relative results of
Figure 17 be viewed with caution.
Low Volume Sample Results-Sulfate
The low-volume sulfate results are derived from the same 47 mm filters
just described, and the same caveats apply. The results of sulfate collections
on various filter materials is shown in Figure 18 in bar-graph form. The ratio
of the average sulfate on the test filters to the average sulfate on ADL quartz
is depicted in the figure. The number of sampling days available for averaging
is shown in parenthesis.
Pall flex quartz and Spectrograde glass both collect, on the average,
the same concentrations of sulfate as ADL quartz. This can be compared to the
Hi Vol. sulfate results where Pall flex quartz tended to collect somewhat more
sulfate than ADL. The Gelman A, AA and AE filters and the Duralon Filter
average 1.5-2 times the sulfate collected by ADL quartz. Gelman E collects
over three times as much sulfate as Spectrograde or either of the quartz
filters.
As discussed with the nitrate low volume sample results, the days
for which comparative samples are available were randomly distributed during
the study. Therefore some filters may have been exposed to more severe
95
-------
o
a
o
4
I '
o
£L
o>
o
o
o
o
<
o
O
M
W
O
Q.
S
o<
o
UJ
o
o
e
o
"5
o
i
S
o
O
o
"Z
o
m
to
UJ
I
o
Filter Type
Figure 18. Relative sulfate collection for low volume filters.
96
-------
conditions (higher SOp and RH) than others. For this reason, the relative
sulfate collection ratios shown in the figure may be somewhat distorted. To
minimize this distortion, we computed the sulfate ratios of the test filters
versus either ADL or Pall flex quartz. Since Figure 18 shows the two quartz
filters have equal collection efficiencies, the approach should be valid, and
it increases the number of sample days available for averaging, thereby
ameliorating the distortion caused by day to day variations in environmental
conditions. In fact, this procedure yields results almost identical to those
shown in Figure 18.
Identity of the Nitrate Precursor(s)
It is interesting to speculate on the precursors or environmental
conditions which lead to artifact nitrate collection on glass filter media.
Our Phase I laboratory studies suggested that a number of nitrogen containing
gases, including NO, N^O, NH3 and PAN were unlikely precursors to artifact
nitrate. The gases which did appear to be collected as nitrate by glass
filters included nitrogen dioxide and nitric acid. The collection of these
gases increased with humidity. Nitrogen dioxide had a much smaller effect on
Gelman Spectrograde than the other glass fiber filters, probably due to the
surface treatment of this filter. Nitric acid had a strong effect on
Spectrograde, even though the filter is almost pH neutral. We suggested
earlier that a different mechanism, involving acid attack of the surface
coating, may be occurring with the Spectrograde filter. Since Spectrograde
appears to differentiate between N0« and HN03> it may be a useful indicator
of the artifact nitrate precursor. This possibility will be discussed shortly.
As an initial attempt to identify the environmental variables which
foster artifact nitrate collection, the daily variations of a number of
97
-------
measured variables were plotted for the Los Angeles field study in Figure
19. Artifact nitrate (i.e., [N03~]Qin$s " ^^s'-'oAST^5 snown in the uPPer
portion of the figure. Variations in artifact nitrate can be compared to
the patterns of the other variables shown in the figure, including NOp,
relative humidity, average 07, average PAN, NO plus PAN, and particulate
•3 X
nitrate (on QAST). Nitric acid is not shown in the figure since its
concentration, as measured by microcoulometry, never exceeded the
detection limit of about 6 ppb. The nylon filter technique, which was
disucssed earlier, was also employed for nitric acid collection on six
study days (10/22, 10/26, 11/2, 11/4, 11/9 and 11/15). The daily average
(8 hour) concentration ranged from 1-5 ppb and averaged 2.6 ppb
3
(6.7 yg/m ). Therefore, the concentration of gaseous nitric acid was high
enough to significantly affect the glass fiber filters. We have found
in the past that gaseous nitric acid concentration is highly correlated
with PAN^11). Thus, the behavior of PAN in Figure 19 should also
qualitatively represent nitric acid.
It is clear from Figure 19 that several variables display patterns
which are similar in many resepcts to the artifact nitrate pattern. Nitrogen
dioxide seems to resemble the artifact nitrate behavior most closely. Ozone
and PAN (or HN03) show similarities but also significant differences. The
concentration of nitrate collected by QAST quartz, which we will assume
represents actual particulate nitrate, also shows some similarities to the
artifact nitrate pattern. It is unlikely that this signifies a cause/effect
relationship; most Hkely, particulate nitrate is serving as a stand-in for
other nitrate precursors. When the particulate nitrate concentration is high,
nitrate precursor (and artifact precursor) concentrations are also likely to
be high. As an example of this, the nitrate and N0~ profiles are similar.
Interestingly, the relative humidity appears to be negatively
correlated with artifact nitrate. This was not predicted by the laboratory
results, which indicated a positive relationship. Since some variables, such
as N0~ and relative humidity, are apparently Interacting, a more comprehensive
statistical investigation of these effects was undertaken.
To investigate potential Interactions among the variables, the data
were scanned using the Automatic Interaction Detector (AID) statistical program.
(34)
The details of this routine have been described by Sonquist and Morgan .
Briefly, the technique identifies which independent variables are the best
98
-------
40
90
20
n
.0.
I
O
z
o I0
o
60
40
20
8? 0
f 60
I 40
5 20
1 °
120
80
40
0
a
a I5
S I0
r—i
§ 5
^ 0
80
60
40
i I0
* 3
v> ^
^ '
& 0
i I
I i I
I I I I I I I
I I I I I
I I I
I I I I
I I I
ICVIB IO/2O KV22 IO/28 11/4 11/9 11/15
10/19 10/21 10/27 11/2 11/5 11/12
Dote
Figure 19. Daily variation of measured variables,
99
-------
predictors of a given dependent variable. The best predictor is defined
as the independent variable that maximizes an F-ratio. The F-ratio is the
ratio of the statistical variability of y that is accounted for by the
variability of x> to the variability of y that is not accounted for by the
variability of x- The AID program simply computes the F-ratios associated
with each independent variable and splits the data using the variable that
yields the maximum F-ratio. The results of the AID analysis are reported in
graphic-tree format.
The significance of the AID splits is related to the size of the data
base, the extent (graphically, the width) of the split, the number of times
splitting occurs on any one independent variable, and ultimately, the between
sum of squares to total sum of squares ratio. For our analysis of artifact
nitrate predictors, the list of independent variables included NOp, humidity,
03, N03~ (from QAST), N03~ * 2.0 ym, and PAN. PAN was employed primarily as a
stand-in for nitric acid for reasons described earlier. The results of the
AID analysis for artifact nitrate are reported in graphic-tree format in
Figure 20. With the limited data base, the results of this or any other
statistical analysis can not be highly significant. Consequently, we will
discuss the analysis only briefly.
The most important predictor of artifact nitrate is PAN, as shown
in the first split in Figure 20. Since the laboratory studies have demon-
strated that PAN is not collected as artifact nitrate, PAN as an artifact nitrate
predictor must represent some other variable highly correlated with PAN. For
the reasons discussed above, this is probably nitric acid, the second split
was a tie among NO^. 03 and PAN. Since Group 4 of the second split contains
only one value, however, the split may not be highly significant. The re-
mainder of the splits occur on PAN or 03 and are not very enlightening. It
is interesting to note that relative humidity does not appear as a significant
predictor of artifact nitrate. This does not agree with the Phase I laboratory
results, but the lack of a visible relative humidity effect may be due to the
quite limited data base.
The results of the AID analysis generally confirm our earlier
identification of nitric acid and N02 as the principal artifact nitrate precursors,
It is not possible from our data to determine the relative contributions of N02
and HN03 to the amount of artifact nitrate collected by the glass filters.
100
-------
AUTOMATIC INTERACTION DETECTOR
ARTIFACT NITRATE
BINARY T»K F STRUC TURF
tlI If M 04 U*tl
i ; i ? I (• »»
TABLF
ft. SO
13) (12
— TOTAL GROUP —
CRITERION - Artifact Nitrate
TO TAL G ROUP N = 13
FIEAN = 10.87
STD . DE V . = 6.67
PARENT 1 SPLITTING VARISBIE - PAN
PARENT 3 SPLITTING VARIABLE - NO,, 03, PAN
PA RFNT ? SPL i TTI NG v« R
piioictii t>iais -- t
FI«*L IKOU»
(ABLE - PAN
1
PARENT 7 SPLITTING VARIABLE - 03
...«- • M t 0 • 0< «. 1
1 I»>| ukj'jf
PARENT 6 SPLITTING VA R
MCOICTOI VOIUCS -• 10 11
PARENT 5 SPL 1 TT I NG VAN
rxoicioi »l»ls -- ' l' 1*
1 *
•f... « M < 0 , >
n «•-'.-'
ABLE - 0 3
^•(•ICTO* vtiuts -- ir tl
ABLE - PAN
MMICTOR V«IUIS -- <
1 1 > OB»
Figure 20. AID results for artifact nitrate.
-------
However, we can use the low volume filter results of Figure 17 to confirm
that both N02 and HN03 must be involved. We found in Phase I that Spectrograde
filters were not significantly affected by N0? but were very seriously
affected by gaseous nitric acid. Since Figure 17 shows that Spectrograde
filters collected more nitrate than quartz during the field study, nitric
acid must be responsible. Furthermore, since the other 4 types of glass
filters collected even more nitrate than Spectrograde, NOp must also be
contributing. While this is an admittedly crude test, it tends to confirm
that both N02 and HN03 were contributing factors in the collection of artifact
nitrate.
PHASE III SUMMARY
The objective of this phase of the program has been to evaluate,
under field conditions, filter materials which the Phase I laboratory
studies deemed potentially promising nitrate collection media. While both
quartz and Teflon filters scored high in the laboratory investigations, our
subsequent efforts emphasized quartz. The primary reason for this is that
quartz filters, if found suitable, could be readily adapted for use in the
extensive high volume sampling networks now in existence, whereas other
considerations, such as pressure drop and electrostatic properties, make
Teflon filters less desirable substitutes.
The field studies were conducted in the Los Angeles basin in
order to provide a stern test of nitrate collection. The results of the
field studies confirm the 1-iboratory predictions that glass fiber filters
collect considerably more nitrate than quartz filters due to collection of
gaseous nitrogen species on the glass surface. Both high volume and low
volume sampling demonstrated this effect. Artifact collection had the most
serious impact on the high volume samples. Since we consider the Hi Vol
data to be more reliable than the low volume results, we view artifact
collection as a rather serious problem.
Both high volume and low volume results demonstrate that ADL quartz
collects less nitrate than Pallflex QAST quartz. This agrees with the
laboratory studies, which showed that QAST was slightly more susceptible to
artifact collection than ADL.
Sulfate from both high volume and low volume sampling indicated
some glass/quartz differences, but the differences were much smaller for
sulfate than nitrate.
-------
In the course of the Phase III study an attempt was made to
identify the gaseous precursor(s) or environmental conditions which foster
artifact nitrate collection. It seems likely that N02, HN03> and humidity
all play a part in artifact formation, and that interactions among these
variables are also important.
The field data suggest that previous studies of nitrate size
distribution which employed glass fiber filters for collection may have yielded
misleading size distributions due to artifact collection. Future studies
should take great pains to eliminate this source of error, since the
physiological effects of nitrate particulate are likely a function of size
distribution.
Based on the Phase I laboratory studies (which demonstrated
minimal artifact collection by ADL quartz) and the field results on
artifact collection and size distribution, ADL quartz filters appear to
collect particulate nitrate with only minimal interference due to artifact
formation. The high and low volume field results also suggest minimal
artifact sulfate collection on ADL quartz in agreement with the results
(32)
of Coutantv '. For these reasons, ADL quartz (or equivalent) filters
would be highly desirable replacements for glass filters in high volume
sampling. Pallflex QAST quartz, a commercially available filter, is
susceptible to some artifact collection, although it also is very much
preferable to glass filters.
103
-------
SECTION 5
PHASE IV: CONCLUSIONS REGARDING SAMPLING AND
ANALYSIS OF ATMOSPHERIC PARTICULATE NITRATE
This report describes research on a number of different topics
pertaining to the sampling and analysis of atmospheric participate nitrate.
The goals of the program have been to identify environmental conditions which
affect nitrate sampling, to develop an improved sampling procedure and to
conceive and develop an improved method for nitrate analysis. A detailed
summary of each phase of the research may be found following each section
of the report. A more general statement of the results and conclusions of
the program follows.
This program investigated a number of potential sources of error
in high volume nitrate sampling, including gas-filter interactions, gas-
filtrate interactions, sampling rate, sampling time, storage time and gas-
soot interactions. The collection of artifact nitrate by certain filter
materials was found to be the most serious source of inaccuracy. Field
results and extrapolations of laboratory data indicate serious interferences
with particulate nitrate sampling, with both gaseous nitric acid and nitrogen
dioxide contributing to the interference. Both the laboratory and field studies
demonstrate that high purity quartz fiber filters represent a significant
improvement over commonly used glass filters for high volume nitrate sampling.
At least one type of quartz was shown to provide a relatively interference-free
nitrate and sulfate collection medium.
A method for the analysis of such ambient filter samples for nitrate
was developed during the program. The method is based on thermal decomposition
of the nitrate, with chemilumnescent detection of the decomposition products.
It is sensitive, accurate and very rapid. Ion chromatography was also investi-
gated and found to be very well suited to the analysis of atmospheric filter
samples. It is sensitive, accurate, reproducible and relatively rapid, and
provides for simultaneous nitrate and sulfate determination. This latter
104
-------
feature makes ion chromatography the method of choice for the majority of
ambient applications.
105
-------
REFERENCES
(1) Shy, C.M., et al., "The Chattanooga School Children Study: Effects of
Community Exposure to Nitrogen Dioxide," J. Air Poll. Control Assoc..
20 (9), 582 (1970).
(2) "Summary Report on Atmospheric Nitrates", U.S. Environmental Protection
Agency, Research Triangle Park, N.C., July 31, 1974.
(3) Robinson, E., and Robbins, R.C., J. Air Poll. Control Assoc., 2£ (5),
303 (1970).
(4) Air Quality Data for 1967 from the National Air Surveillance Networks
(Revised 1971) PB 203 546, National Technical Information Center.
(5) Murcray, D.G., Kyle, T.G., Murcray, F.H., and Williams, W.J., Nature,
218, 78 (1968).
(6) Lazrus, A.L. and Gandrud, B.W., Proceedings of Third Conference on
Climatic Impact Assessment Program, p. 161, Feb. 26, 1974.
(7) Public Health Service Publ. No. 978, U.S. Dept. of Health, Education,
and Welfare, Washington, D.C. (1962).
(8) Gordon, R.J. and Bryan, R.J., Environ. Sci. and Tech., 7 (7), 645
(1973). "
(9) Hidy, G.M., et al., "Characterization of Aerosols in California",
Final Report (Vo. IV), California Air Resources Board, Sept., 1974.
(10) Spicer, C.W., and Miller, D.F., "Nitrogen Balance in Smog Chamber
Studies", J. Air Poll. Control Assoc.. 26(1), 45 (1976).
(11) Spicer, C.W., "The Fate of Nitrogen Oxides in the Atmosphere", Battelle-
Columbus Laboratories Final Report to EPA/CRC, 1974.
(12) Miller, D.F., Schwartz, W.E., Jones, P.E., Joseph, D.W., Spicer, C.W.,
Figgle, C.J., and Levy, A., "Haze Formation: Its Nature and Origin --
1973", BatteHe-Columbus Laboratories report to EPA and CRC, June, 1973.
(13) Heuss, J.M. and Glasson, W.A., Environmental Sci. Tech., £, 1109 (1968).
(14) Pate, J.B. and Tabor, E.C., Ind. Hyg. Jour., March-April, 1962, p. 145.
106
-------
(15) Spicer, C.W., Gemma, J., Joseph, D., and Levy, A., "The Fate of Nitrogen
Oxides in Urban Atmospheres", presented at the American Chemical Society
Meeting, Atlantic City, New Jersey, Sept., 1974.
(16) Spicer, C.W., "Non-regulated Photochemical Pollutants Derived from
Nitrogen Oxides", presented at EPA Symposium on Automotive Pollutants,
Washington, D.C., Feb. 12, 1975.
(17) Pierson, W.R., Butler, J.W., and Trayser, D.A., Environ. Letters, ]_ (3),
267 (1974).
(18) Trayser, D.A., Blosser, E.R., Creswick, F.A., and Pierson, W.A., "Sulfuric
Acid and Nitrate Emissions from Oxidation Catalysts", paper presented at
SAE Congress and Exposition, Feb. 25, 1975.
(19) O'Brien, R.J., Holmes, J.R., Reynolds, R.J., Remoy, J.W., and Bockian,
A.H., Paper No. 74-155, presented at 67th APCA Meeting, Denver, Colorado,
(1974).
(20) Lovelock, J.E. and Penkett, S.A., Nature, 249, 434 (1974).
(21) Chang, S.G. and Novakov, T., "Formation of Pollution Particulate Nitrogen
Compounds by NO-Soot and NhL-Soot Gas-Particle Surface Reactions", Atm.
Environ., 9, 495 (1975). J
(22) Lambert, R.S. and DuBois, R.J., Anal. Chem., 43(7). 455 (1971).
(23) Coutant, R.W., Battelle-Columbus Laboratory, unpublished results (1975).
(24) Stevens, R., U.S. Environmental Protection Agency, personel communication
(1976).
(25) Small, H., Stevens, T.S., and Bauman, W.C., Anal. Chem., 47(11), 1801 (1975)
(26) Rich, W., "Ion Chromatography-A New Analytical Technique for Measuring
Ions in Solution", presented at Symp. on Recent Developments in the
Sampling and Analysis of Atmospheric Sulfate and Nitrate, Research
Triangle Park, N.C., March, 1976.
(27) Mulik, J., "Ion Chromatography of Sulfate and Nitrate in Ambient Aerosols",
presented at Symp. on Recent Developments in the Sampling and Analysis of
Atmospheric Sulfate and Nitrate, Research Triangle Park, N.C., March, 1976,
(28) Otterson, D.A., "Ion Chromatographic Determination of Anions Collected on
Filters at Attitudes Between 9.6 adn 13.7 kilometers", NASA Technical
Memorandum, NASA TM X-73642 (1977).
(29) Lathouse,J., "Practical Experience With Ion Chromatography", paper presented
at Symp. Ion Chromat. Anal Environ. Pollutants, Research Triangle Park, N.C.
April, 1977.
107
-------
(30) Gelman Instrument Co., product brochure on glass fiber filters, Ann
Arbor, Michigan.
(31) Pall flex Products Corp., product brochure on quartz fiber filters,
Putnam, Connecticut.
(32) Coutant, R.W., "Effect of Environmental Variables on Collection of
Atmospheric Sulfate," Environ. Sci. and Tech.. n_, 873 (1977).
(33) Spicer, C.W., Ward, G.F. and Gay, B.W., Jr., "A Further Evaluation of
Microcoulometry for Atmospheric Nitric Acid Monitoring", Anal. Letters,
11(1). (1978).
(34) Sonquist, J.A. and Morgan, J.N., "The Detection of Interaction Effects",
Monograph No. 35, University of Michigan (1964).
108
-------
APPENDIX
EXPERIMENTAL CONDITIONS
109
-------
TABLE A-l. EXPERIMENTAL CONDITIONS AND SAMPLE VOLUMES
(VOLUMES IN m3)
Experiment No. :
Experimental Conditions:
Filter Materials
Nylon (Ouralon)
Teflon (Mltex)
Cellulose Acetate (Celotate)
Glass (Gelman A)
Glass (Gelman E)
Polycarbonate (Nuclepore)
Quartz (ADL)
Experiment No.:
Experimental Conditions:
Filter Materials
Nylon (Duralon)
Teflon (Hltex)
Cellulose Acetate (Celotate)
Glass (Gelman A)
Glass (Gelman E)
Polycarbonate (Nuclepore)
Quartz (AOL)
1
Clean Air
1.08
0.98
0.95(0.95)
1.12
1.12
1.07
1.12
10
3.0 ppm HN03
301 RH
0.73
0.98
0.96
1.13
1.13
1.00
1.13
2
Clean Air
1.00
0.93(0.93)
0.89
1.05
1.03
1.00
1.04
11
80 ppm NH3
1.35 •
0.98
1.27
1.39
1.39
1.31
1.41
3
2.6 ppm N02
1.02(1.02)
0.92
0.89
1.02
1.03
1.00
1.05
12
5.5 ppm NH3
1.15
1.07(1.05)
.05
.18
.20
.12
.20
4
2.0 ppm NOj
0.99
0.91
0.89
1.02
1.03
0.97
1.01(1.03)
13
0.3 ppm PAN
0.93
0.82
0.85(0.82)
0.97
0.97
0.92
0.96
5
1.8 ppm NO?
401 RH
0.87
0.79
0.68
0.79
0.79(0.79)
0.75
0.78
14
15.6 ppm NzO
1.61
1.52
1.45
1.66
1.67(1.64)
1.61
1.64
6
1.4 ppm HN03
0.76
0.69
0.68
. 0.79
0.79(0.79)
0.75
0.78
15
18 ppm N?0
-vlOOI RH
1.33
1.25
1.17
1.41
1.39
1.31(1.33)
1.41
7
1.5 ppm HN03
0.99
0.89
0.85
1.00(1.01)
1.00
0.96
1.02
16
1.7 ppm NH3
401 RH
1.45
1.34(1.36)
.29
.47
.52
.40
.50
8
8.0 ppm HN03
.29
.22
.17
.35
.35
.31
.35
17
30 ppm NO?
701 RH
0.99(1.01)
0,84
0.91
1.04
1.03
0.89
1.03
9
3.0 ppm HNOj
.32
.22
.18
.34
.37
.28
1.35
18
11.5 ppm KH3
—
—
1.29
—
2.36
—
2.37
-------
TABLE A-2. EXPERIMENTAL CONDITIONS AND SAMPLE VOLUMES
Experiment
No.
19
20
21
22
23
24
25
Experimental
Conditions Filter Material
0.024 ppm H2S04 Glass (Gel man AE)
Quartz (ADL)
Cellulose Acetate
Quartz (Pall flex)
Quartz (ADL), clean
3.4 ppm HN03 Quartz (ADL)
Glass (Gelman AE)
Cellulose Acetate
19.5 ppm NOp Glass (Gelman AE)
Cellulose Acetate
Quartz (ADL)
Quartz (Pal If lex)
0.35 ppm HNOo Quartz (Pall flex)
J Quartz (ADL)
Glass (Gelman AE)
21 ppm N02 Quartz (pal If lex)
Quartz/Cellulose
Glass (Gelman AA)
Quartz (ADL)
Glass (Spectrograde)
17.5 ppm HNO, Quartz (ADL)
Glass (Gelman AA)
Quartz/Cellulose
Glass (Spectrograde)
Quartz (ADL)
16.5 ppm NOp Glass (Spectrograde)
17% RH Quartz (ADL)
Quartz (Pal If lex)
Quartz/Cellulose
Glass (Gelman AA)
Volume
Sampled, m3
2.24
2.29
1.22
2.30
2.25
3.68
3.71
2.24
2.14
1.87
2.10
2.10
4.15
4.18
4.09
0.95
0.98
0.97
0.98
0.96
1.25
1.28
1.28
1.28
1.25
1.26
1.29
1.29
1.30
1.27
111
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
i. m PORT NO. 2.
EPA-600/2-78-067
.1 I II I I ANI1SUUTITLE
SAMPLING AND ANALYTICAL METHODOLOGY FOR ATMOSPHERIC
PARTICULATE NITRATES
Final Report
3. RECIPIENT'S ACCESSION NO.
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
C. W. Spicer, P. M. Schumacher, J. A. Kouyoumjlan and
D. W. Joesph
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
10. PROGRAM ELEMENT NO.
1AD712BB-42 (FY-78)
11. CONTRACT/GRANT NO.
68-02-2213
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory- RTP.NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Environmental conditions that affect atmospheric particulate nitrate
sampling were identified, and Improved sampling and analytical procedures were
developed. Evaluation of potential sources of error 1n high volume nitrate sampling
showed that artifact nitrate formation on commonly used glass filter media was the
most serious. Both laboratory and field results demonstrated that high purity quartz
filters provide a significant Improvement over glass filters and are easily sub-
stituted for glass filters in traditional high volume sampling equipment. A
sensitive, accurate and rapid nitrate analytical procedure was developed using ther-
mal decomposition of nitrate and chemiluminescent detection of the decomposition
products. Ion chromatography was also investigated and found to be sensitive,
accurate, reproducible and rapid. Ion chromatography has the added advantage of
determining both nitrate and sulfate simultaneously.
17.
I.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
*Air pollution
*Particles
*Inorganic nitrates
*0rganic nitrates
*Sampling
*Filter materials
*Chemical analysis
*Chemi1umi nescence
*Chromatography
b.lDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
13B
07B
07C
14B
13M
07D
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21.
PAGES
WflSIiftfflT
(This page)
22. PRICE
EPA Form 2220-1 (9-73)
112
------- |