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

-------