Modern Methods to Measure Air Pollutants
(U.S.) Environmental Protection Agency
Research Triangle Park, NC
Oct 85
PB86-129798
U.S. Department of Commerce
National Technical brformatiofi Service
-------
EPA/600/D-85/260
October 1985
MODERN METHODS TO MEASURE AIR POLLUTANTS
Robert K. Stevens
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina, U.S.A.
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
TECHNICAL REPORT DATA
(Please read Instructions on the rcvenc before completing/
i. REPORT NO.
EPA/600/D-85/260
3. RECIPIENTS ACCEiSLON-NO. _
P23b 1-2 9" 9 8 /AS
4. TITLE AND SUBTITLE
MODERN METHODS TO MEASURE AIR POLLUTANTS
5. REPORT DATE
October 1985
6. PERFORMING ORGANIZATION CODE
7. AUTMOR(S)
Robert K. Stevens
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Atmospheric Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
10. PROGRAM ELEMENT NO.
iTCONTR ACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Atmospheric Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This paper discusses the requirements for the collection and analysis of
ambient particles to satisfy data requirements for source and receptor models as
applied to pollution control applications. The paper describes the following
analytical procedures as applied to receptor modeling: X-ray fluorescence (XRF)
?vonN0n ac.t1vation (NAA)« ion exchange chromatography (1C), X-ray diffraction
(XRD), optical microscopy (OM) and scanning electron microscopy (SEM) as well as
gas phase denuders to measure a wide range of pollutants
ammonia.
the recent application of
including nitric acid and
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS C. COSATI I leld. Group
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Rtporl/
UNCLASSIFIED
21. NO OF PAGES
61
20.
Tliii page i
22. PRICE
EPA P*rm 2220-1 (R*y. 4-77) PREVIOUS EDITION 11 OBSOLETE
-------
' Abstract
The atmosphere is an enormously complex mixing vessel. And yet, in spite
of that complexity the aerosol particles found there retain distinctive
distributions with respect to size and composition. The bimodal size
distribution for aerosols has been found in various regions by many
investigator?, and analysis of the particle size fractions has shown that the
two modes differ considerably in their elemental constituents and in their
chemistry. For example, the fine particle fraction is often acidic; and the
coarse particle fraction, is often basic. This interesting dichotomy is
consistent with the view that the fine particle fraction is constituted
largely of particles formed from atmospheric gases, and that among these gases
are the acidic oxides of sulfur and nitrogen. On the other hand, the coarse
particle fraction is mainly composed of crustal elements, among which are the
metals Ca and K, which form the basic oxides. In practice, our primary
interest in measuring the chemical properties of particles is to provide
information for pollution control. Two complementary orientations exist to
address pollution control issues. One approach, termed 'receptor modeling1,
begins with the measurements of elemental abundances in particles collected
downwind of the sources and works back through a model to determine the
contribution of those sources io the particulate mass loading. The second
approach, termed 'source' or 'dispersion modeling', is based on a continuity
(conservation) equation describing material in the atmosphere, and uses source
emission rates, meteorological parameters, rates of transport, and
transformation data to calculate the expected air quality. The test of the
validity of these model calculations is the agreement of the predicted values
ii
-------
with experimental observations of air quality. This paper discusses the
requirements for the collection and analysis of ambient particles to satisfy
data requirements for source and receptor models as applied to pollution
control applications. The paper describes the following analytical procedures
as applied to receptor modeling: X-ray fluorescence (XRF), neutron activation
(NAA), ion exchange chromatography (1C), X-ray diffraction (XRD), optical
microscopy (OM) and scanning electron microscopy (SEM) as well as the recent
appli:ation of gas phase denuders to measure a wide range of pollutants,
including nitric acid and ammonia.
-------
Introduction
The atmosphere is an enormously complex, more or less effective, mixing
vessel. And yet, in spite of that complexity (numerous sources, sinks,
transport processes, and transformations), the aerosol particles found there
retain distinctive distributions with respect to size and composition. The
bimodal size distribution for aerosols* has been found in various regions by
many investigators, and analysis of the particle size fractions has shown that
the two modes differ considerably in their elemental constituents and in their
chemistry. For example, the fine particle fraction is often acidic; and the
coarse particle fraction is often basic. This interesting dichotomy is
consistent with the view that the fine particle fraction is constituted
largely of particles formed and accreted from atmospheric gases, and that
among these gases are the acidic oxides of sulfur and nitrogen.** On the
other hand, the coarse particle fraction is mainly formed from the larger
particles breaking up, among which are certain metals (Ca, K), which form
the basic oxides. These properties of atmospheric particles are schematically
represented in Figure 1.
In general, we find that the elements sulfur and lead appear mostly in
the fine fraction, and calcium, aluminum, and silicon appear mostly in the
*Aerosol particles will be referred to as "aerosols1!, not to be confused with
gas and solids or liquids.
** The relative contribution of nitrogen oxides to particle formation is now a
controversial subject, and evidence exists that previous estimates were
considerably too high.
-------
CHEMICAL CONVERSION
OF OASES TO
1
LOW VOLATUTY
r VAPOR 1
1
HOMOGENEOUS CONDENSATION
NUCLEATION GROWTH
OF NUCLEI
j
V
PRIMARY _ fr, COAGULATION
EMISSIONS \ w
SIZE DISTRIBUTION
AND ORIGINS OF
AMBIENT PARTICLES
WIND BLOWN DUST
. - AND
• : EMSSIONS
-: b*£.«n-i 'fciwujti.*'. -.-. .. .
SBXMENTATON
r
0.01
I I
0.1 1
PARTICLE DIAMETER.
FINE PARTICULATES •»
.. . . .
10
MICROMETER
COARSE PARTICJLATES
I
100
Figure 1. Size distribution and origins of ambient particles.
-------
coarse fraction. These elements, which usually appear in quantities measured
in micrograms per cubic meter, are not thought of as trace elements.
In practice, EPA's primary interest in measuring materials in the
atmosphere is to provide information for pollution control. Two complementary
orientations to this problem exist. In "source apportionment models," one
begins with the measurements of elemental abundances in particles collected
downwind of the sources and works back through a model to determine the
contribution of those sources to the particulate mass loading. In "air
quality simulation models," one takes a continuity (conservation) equation
describing material in the atmosphere and various input and output data
(source emission rates, meteorological parameters, rates of transport and
transformation, etc.) and calculates the air quality. The test of the model
calculation is the agreement of its predicted values with experimental
observations of air quality. Experimental determinations of elements in
atmospheric particles are necessary, then, in both approaches to the
assignment of atmospheric sources: one approach works from the observations
back to the source, and the other work from the source to the observations.
These remarks are illustrated in Figure 2.
The following sections of the paper discuss the requirements for the
collection and analysis of ambient particles to determine the mass and
elemental composition and to determine the physical properties and chemical
species present in the sample. The choice of sampling equipment and
characterization procedures for ambient particles for environmental study data
collection should be dictated by the objectives of the study. In the case of
determining the effects of aerosols on surfaces, the identification and the
physical and chemical properties of as many of the components that constitute
the aerosol sample should be an integral part of the study.
-------
SOURCE APPORTIONMENT:
Chemical element balance.
Factor analysis
EXPERIMENTAL DATA
Elemental abundances
POLL'JTIOM
SOURCES
AIR QUALITY SIMULATION MODELS:
Climatological dispersion model
Air quality display model.
Figure 2. Model paths connecting experimental measurements of elements in
atmospheric particles and the source of those particles.
-------
The discussion that follows also describes the procedures and the
application of the following instrumental methods for particle characteriza-
tion: X-ray fluorescence (XRF), neutron activation analysis (NAA), ion
exchange chromatography (1C), X-ray diffraction (XRD), optical microscopy (OM)
and scanning electron microscopy (SEM). Also, the procedure for and
application of gas phase denuders to measure HN03 and nitrate and a combustion
method to measure the carbon content of particles will be described.
From this array of analytical procedures, the chemical and physical
properties of ambient particles are determined and their potential impact on
man and materials can be inferred. In addition, the sources of the ambient
particles can frequently be identified from these same chemical properties,
e.g., elemental composition is used as input to receptor models (1).
SAMPLERS AND FILTER MEDIA
Of the many aerosol sampling methodsin use today, no one method can
provide the range of measurements necessary for the comprehensive chemical
analysis required in many air quality environmental impact studies. In this
discussion of samplers and filter media, we assumo that particle collection
will be made over 24 hour periods and that particles <2.5 urn in diameter (fine
frartion) and the particles > 2.5 urn in diameter (coarse fraction) are of
interest and will be collected and analyzed. In addition, we will assume the
need to measure chemical and physical properties of each of these fractions
and for both inorganic and organic content.
Dichotomous Sampler: One partif* collection system that has been used
frequently by the U.S. Environmental Protection Agency (EPA) to provide
-------
samples compatible with the above discussed requirements is the dichotomous
sampler. This sampler is designed to use either Teflon or quartz filters that
satisfy another important requirement in aerosol characterization studies: the
sample collection will be nearly free from artifact formation. Table 1
contains a list of commonly used filter types and their respective properties.
Note the alkalinity of the glass fiber filters, responsible for sulfate
artifacts during sampling.
Table 1. Properties of filters used to collect ambient particulate
samples.
SUMMARY OF USEFUL FILTER PROPERTIES
FILTER AND FILTER
COMPOSITION
TEFLON (MEMBRANE)
(CF2)n(2/.m PORE SIZE)
CELLULOSE (WHATMAN 41)
(CeHlpO5)n
GLASS FIBER (WHATMAN GF/C)
"QUAh iZ" GELMAN MICRO-
QUARTZ
POLYCARBONATE (NUCLEPORE)
C15H14 CO3 (0 3j;mPORE SIZE)
CELLULOSE ACETATE/NITRATE
MILLIPORE (1 .2/ym PORE SIZE)
(CgH1307)n
DENSITY
mg/cm^
0.5
8.7
5.16
6.51
0.8
5.0
pH FILTER
EFFICIENCY %*
NEUTRAL
NEUTRAL
(P«ACTS WITHHNOjl
BASIC pH- 9
pH 7
NEUTRAL
NEUTRAL
IBFAI IS W"TH HNOjl
•MINIMUM EFFICIENCY FOR PARTICULATE DIAMETER >0.035//m AT V
99.95
58% AT
0.3 pm
99.0
98.5
93.9
99.6
= 10cm /sec
-------
Sampling artifacts are defined as the chemical reactions of acidic gases
with certain types of 'liters tc form nonvolatile species, e.g., S02 + M •»
S£>4M or HN03 + M •* MN03; where M is a metal impurity (e.g., Na, Ca) in the
collection substrate. Thus, the true sulfate and nitrate concentration is
perturbed by this artifact. Glass fiber filters and cellulose membrane
filters are notorious for producing sulfate and nitrate artifacts (2).
The dichotomous s.impler is a virtual impactor that aerodynamicslly
separates particles into size fractions corresponding to the fine and coarse
modes of the normal size distribution of ambient particles and deposits the
particles uniformly on the filters. A schematic diagram of the dichotomous
sampler is shown in Figure 3. More detailed diagram of the inertial impactor
separation stage is shown in Figure 4. Ambient particles should be collected
on 1 or 2 urn pore size Teflon filter (3) if the elemental composition is to be
determined accurately by XRF or ionic content (4). Teflon filters are the
filter medium of choice because they are inert, chemically pure, and have a
high collection efficiency, as noted in Table 1. The density of 2 urn pore
size membrane Teflon filter is approximately 500 ug/cm2. This low density is
ideal for XRF and mass measurements while having > 99% collection efficiency
for particles > 0.05 urn in aerodynamic diameters. However, quartz filters are
used for condensable carbon measurements in field studies because total
condensable carbon content is typically measured by a thermal combustion
procedure described by Huntzicker et al. (5). This combustion procedure will
be discussed in detail in the measurement section.of this report.
Cyclone Samplers: In field studies where only the fine particles (<2.5 urn)
are to be collected, cyclone inlets can be used to remove the coarse
particles. Figure 5 is an example of a typical sampling system that us>cS a
-------
INLET
1671pm
CONSTANT
PRESSURE DIFFERENTIAL PRESSURE REGULATOR
FLOW REGULATOR (OPTIONAL)
^
FINE
151pm
-HXH-
FINE
FLOW VALVE
COARSE 171pm
ROTAMETER
FILTER
COARSE
FLOW VALVE
DUAL HEAD
OIAPHRAMPUMP
Figure 3. Diagram of Dichotomous Sampler System
-------
VIRTUAL IMPACTOR
INIET
171pm
TO FINE
PARTICLE
FILTER
1531pm
TO COARSE
PARTICLE
FILTER
1.7 Ipm
Figure 4. Diagram of Virtual Impactor
-------
CYCLONE
SHOULD BE
VERTICAL
FLOW
CONTROLLERIADJUST
FOR DESIRED FLOW)
28.3l'min 2.1 ...n CUT POINT
PUMP
DIFFERENTIAL PRESSURE
GAUGE CONN. (OPTIONAL)
Figure 5. Schematic of fine particle sampler with cyclone inlet and pneumatic
flow controller.
10
-------
cyclone inlet, pneumatic flow controller, pump, and filter and particle
collection assembly. This aerosol collection arrangement samples the air and
deposits the fine particles onto a 37 mm diameter filter. One of the
advantages of using cyclone inlets when collecting particles is that a variety
of sample flow rates and cutpoints can be used by selecting the appropriate
cyclone design. Also, Lippmann and Chase (7) noted that cyclone samplers can
be designed to match respiratory deposition curves. John and Reischl (6) have
recently described the properties of several cyclones that would be applicable
to air pollution monitoring studies.
ANALYSIS
Mass: The mass of the aerosol collected can be measured by gravimetric
procedures. Typically, the mass of aerosol collected on a 37 or 47 mm
diameter Teflon filter can be determined to within ± 25 M9 per filter.
Therefore, if 24 m3 of ambient air is pulled through a 'liter and the average
particle content is 50 ug/m3, the mass of particulates collected on the filter
can be measured to a precision of ± 2%. For some studies performed by the
EPA, mass measurements are made by p-ray attenuation using a method
characterized by Jaklevic et-al. (8) and Courtney et al. (9). The p-ray
attenuation and gravimetric procedures can both determine the mass of aerosol
collected on Teflon filters to ± 25 ug. Measurements of mass collected on
quartz filters are not as accurate because the quartz is extremely fragile and
the weight is difficult to obtain in a micro-balance.
Elemental Composition: There are a number of analytical procedures that can
be used to measure the elemental content of aerosols. Some of these methods
require that the sample be extracted and then analyzed by an emission
11
-------
spectrographic procedure. For example, Scott et al. (10) reported
multi-element analysis of about one thousand Hi-vol filters by.low temperature
oxygen ashing and followed by acid extraction of the filter and analysis of
the extract by optical emission spectroscopy (OES). Their paper elso provides
considerable information on detection limits and confidence levels determined
by extensive replicate and standard sample analysis. The absolute sensitivity
of OES is not as high as that of some other techniques. The Hi-vol sampler,
howjver, collected so much sample that OES was sufficiently sensitive to
provide quantitative determinations of Al, Be, Ca, Cr, Cu, Fe, Mg, Mn, Mo, Ni,
Pb, Ti, and V in a set of 898 samples from 248 sites in the National Air
Surveillance Network (NASN). Determinations were also made of As, Bi, Co, Bs,
Sn, Te, and Tl, but the levels were above the detection limit in fewer than 1%
of the samples. Table 2 shows the frequency of detection, which is defined as
the percentage of samples in which the element is detected at a level above
the detection limit. Scott et al. (10) defined the detection limit as the
concentration corresponding to a signal three times the standard deviation
above the average signal due to the extraction solution. The extraction
solution contained internal standards for normalization of measured values.
It is clear from Table 2 that OES analysis of the NASN samples has
provided sufficient data to follow trends in abundances of Cu, Fe, Mn, Pb, and
Ti, but certainly not As, Bi, Co, Sb, Sn, Te, or Tl, and probably not Mo.
Whether sufficient data exist for Al, Be, Ca, Cd, Cr, Mg, Ni, and V depends on
more detailed considerations, e.g., comparisons of element levels with
detection limits, interelement interferences, and the requirements of the
statistical treatment of the data. Obviously, these results refer only to the
1970 NASN samples; samples from different sites will probably have different
frequencies of detection.
12
-------
Table 2
Frequency of detection by OES of elements in NASN samples.
Number of Samples: 898
Sampling Period: 1970
Sampler and Flow Rate: Hi-vol, 1.7 mVmin
Element Frequency of Detection
Cu, Fa, Mn, Pb, Ti 80-100%
AT, Be, Ca, Cd, Cr, Mg, Ni, V 10-80%
Mo < 10%
As, Bi, Co, Sb, Sn, Te, Tl < 1%
For many recent air pollution studies, energy dispersive x-ray
fluorescence analysis (EDXRF) has been widely used to analyze nondestructive^
the elemental composition of ambient particles. Stevens et al. (4) used
calibration procedures described below to measure the elemental content of
ambient particles. The x-ray machine used in these studies -^s fabricated by
Lawrence Berkeley Laboratory and uses a pulsed x-ray tube to excite a
secondary target which, in turn, excites the sample with nearly non-energetic
x-rays (11).
To obtain high sensitivity for a wide range of elements, each sample is
excited by means of three different secondary targets. For the K x-rays of
elements with atomic numbers in the ranges 13-20, 21-38, and 39-56, the
secondary targets consist of titanium, molybdenum, and samarium, respectively.
The molybdenum target also excites the L x-rays of lead and other heavy
elements. The fluorescent x-rays from the sample are detected using a lithium
drifted silicon detector, that uses electronic collimation to minimize the
background. Because of a compact geometrical arrangement between components
of the spectrometer, an x-ray tube power of only 100 W is sufficient to
provide the maximum usable count rate for filter samples.
13
-------
After a sample has been irradiated in the EDXRF spectrometer, the
resulting spectrum is analyzed to determine the number of counts that each
element contributes. To accomplish this, a least-squares procedure is used to
find a linear combination of single-element spectra that best describes the
unknown spectrum. A library of single-element spectra is obtained for thin
standards of each element and is stored in computer memory. The library must
contain all elements that could contribute to an unknown sample (4, 11, 12,
13).
The least-squares method is slightly modified for the K x-rays from
sulfur to deal with the overlapping M x-rays from lead. Because the
concentration of lead can be accurately determined from the emission of L
x-rays, it is possible to 'strip' out the interfering lead M peak in the
vicinity of the sulfur peak. Such stripping is done prior to performing the
least-squares analysis for sulfur. The uncertainty in the result for sulfur
due to the presence of lead is estimated to be about 5% of the lead
concentration. For example, if the lead concentration were 2 ug m-3, the
resulting sulfur uncertainty would be 0.1 ug m-3. The complete description of
the calibration of the EDXRF system is described by Stevens et al. (4).
For most air pollution applications in which only the major elements are
needed to determine elemental content of a participate sample, EDXRF is
adequate. However, EDXRF cannot be used to accurately determine elements
lighter than Al, or routinely observe the various minute trace elements such
as Lu, Hf, Ta, W, Th, Sm, Dy, Yb, Se, As, and La.
A particularly useful combination is EDXRF followed by instrumental
neutron activation analysis (INAA). The latter technique, described by
Kowalczyk et al. (14) is totally instrumental and, depending on mass loadings
and filter blanks, can observe up to 40 elements. Gordon et al. (15)
14
-------
demonstrated the value of applying INAA to samples from the Regional Air
Pollution Study that had previously been subjected to XRF. These studies show
that the combination of XRF and INAA yielded data for 28 elements in the fine
fraction and 32 in the coarse fraction for virtually every sampling period.
An additional 18 elements in the fine fraction and 14 in the coarse fraction
were observable in about half of the samples. Furthermore, 11 elements (Al,
Ca, K, Cl, Ti, V, Mn, Fe, Zn, Br, and Ba) can usually be measured by both
methods for a quality assurance check. Thorough INAA requires access to a
reactor with a flux of at least 1013 n cm-'-'-s-1 and a high-resolution Fe (Li)
or intrinsic Ge y-ray spectrometry system. Irradiations of two different
lengths are needed for species of various half-lives (see below).
As presently performed, compared with XRF, INAA requires considerably
more sample handling, interpretation of results, longer analytical time (2-3
weeks after irradiation for certain long-lived isotopes), for the resolution
of certain elements (S + Pb), and, consequently, is considerably more costly
per sample. Tnus, XRF is often the preferred screening technique. All
samples could be subjected to XRF and the results used to identify only the
most interesting samples for further analysis by INAA.
Another vay to reduce INAA cost is an approach taken by Watson (16) and
Core et al. (17) at the Oregon Graduate Center. Complete INAA requires at
least two irradiations of samples: the fi-st for a few minutes to observe
species with half-lives of <15 h, and the second for several hours to observe
species with half-lives of uj, to several years. The former method is less
costly to perform, as the -y-ray spectra are simple enough to be resolved
primarily by computer. Furthermore, in the former, results are usually
available within a day after the irradiation, compared with about three weeks
after irradiation for long-livd isotopes. The Oregon Graduate Center group
15
-------
has employed INAA cost effectively by performing only the short irradiation
following XRF analyses. Depending on the conditions, most of the following
elements are usually observed: Na, Mg, Al, S, Cl, K, Ca, Ti, V, Hn, Cu, Br,
La, Sin, Dy, and W. However, some important elements that are normally
observed in long irradiations (Cr, Co, As, Se, and Bs) are usually sacrificed
in this approach.
Many other techniques are used to observe elements in atmospheric
particulate matter, including atomic absorption spectrophotometry, inductively
coupled plasma atomic emission spectrometry, spark-source mass spectrometry,
various electrochemical methods (such as anodic stripping voltammetry), and
chemical separations followed by colorimetry. Most require considerable labor
per sample to obtain accurate results or (unlike EDXRF and INAA) require
dissolution of samples prior to analysis. Moreover, any chemical manipulation
greatly increases the chance of contamination by trace elements and loss of
volatile or insoluble species.
Ionic Species: The major ionic species present in ambient particles are
sulfate (SOT), nitrates (NOZ), hydrogen (H ), chlorides (Cl~), bromide (Br")
and ammonium (NH.). After EDXRF analysis, samples may then be extracted and
the anion and cation concentration measured. The extraction process used when
analysis of particles collected on Teflon filters is required consists of
removing the filter from its holder, loading it into an extraction vessel,
filling the vessel with extraction solution, and then extracting by the use of
an ultrasonic bath.
The filters are carefully removed from the filter holders and as quickly
as possible placed into the extraction solution in order to prevent
contamination. For acidity measurements, care must be taken not to breathe on
16
-------
the filters in order to minimize exposure to exhaled ammonia. The filters are
placed in the extraction vessel with the back of each filter facing the bottom
in order to prevent any material from being inadvertently removed as the
filter is inserted into the vessel.
The extraction vessel recommended is a 30-ml Nalgene polypropylene bottle
(Nalge, Inc., Rochester, NY) that has been conditioned by soaking in
extraction solution for at least 8 h before use. To keep the filter submerged
and open during extraction, a fluted Teflon pipe is placed in the vessel so
that the fluted end rests on the outer, unloaded edge of the filter.
The volume of extraction solution used depends on the type and number of
analyses performed and ranges from 8 to 20 ml. The extraction solution,
perchloric acid, is diluted to 5 x 10-5 N with distilled water. The
extraction solution is delivered with 0.05% precision to the vessel by en
Oxford Laboratories macro set pipet. The vessel is then capped and placed in
an ultrasonic bath (Model 8845-60, Cole-Palmer Instrument Co., Chicago,
Illinois) for 20 min. The ultrasonic bath tends to produce standing waves in
the water so that regions exist where no agitation takes place. To overcome
this problem, the extraction vessel is continuously moved during the
extraction period.
The above extraction procedure is used for analysis of SOT, NO,, NH. and
H . For the fine and coarse particle fractions, Stevens et al. (3) found the
extraction efficiencies for sulfur to be 98 ± IX and 95 ± 2%, respectively.
For those samples on which a sulfite determination is to be made, the
extraction procedure is modified to minimize the conversion of sulfite to
sulfate. Stevens et al. (4) demonstrated that carrying out the extraction at
room temperature caused complete conversion of sulfite to sulfate. To prevent
such conversion, a cold extraction process was developed. In the modified
17
-------
procedure, the water bath was made up of a slurry of ice and water that
maintained the temperature of the sample at 0°C during the extraction. With
this procedure 90% of the sulfite was preserved.
Hydrogen ion analysis Titrimetry and a Gran's function plot are recommended
to determine strong acidity using the procedure of Brosset and Perm (18). The
instrumentation includes an Orion pH meter, a combination electrode, a Gran's
function generator, and a Radiometer autoburett ABU 12. The data can be
recorded on a 1-mV strip-chart recorder. The Gran's function generator is an
antilog amplifier (Model 755, Analog Devices, Norwood, MA).
The acid analysis procedure is as follows: 1) standardization of sodium
hydroxide titrant, 2) measurement of the concentration of the acid extraction
solution, 3) titration of extract using the 0.001 N NaOH, and 4) data
analysis. Steps 1 and 2 are done once for each set of filter extracts or at
least once each day. Nitrogen is bubbled through the solution being titrated
to remove an interference by CO-. The ionic strength of the volume being
titrated is maintained at 0.02 M by the addition of KC1.
The method has been characterized by performing 50 measurements on
sulfuric acid standards that had a range of concentrations. The linear
dynamic range was 10-6 to 10-4 M. The relative standard deviation was 2% and
the minimum detectable level was 1 nano-equivalent ml-1 (neq m-1).
Ion exchange chromatcgraphy: Ion exchange chromatography is a relatively new
technique for routine analysis of anions and cations in aqueous extracts of
ambient aerosols. Mulik et al. (19) has described the application of a
commercial liquid 1C (Dionex Model 14, Sunnyvale, CA) to the analysis of water
soluble anions. Before injecting a sampu i-.to the 1C, 5 ml of the aerosol
18
-------
extract is spiked with the- 1C eluent (typically 1:1 mixture of 0.003 M Na2C03
and 0.0024 M NaHC03) to adjust the molar CO^ concentration to the range of
10-3 to 10-5. This basic mixture is then added to neutralize the minute
concentrations of perchloric acid used in the extraction procedure so that the
injected solutions are approximately the molar concentration of the eluent
used in the chromatograph. Because the instrument calibration curve is not
linear, multi-point calibrations are performed over the range of
concentrations expected in the samples. The signal from the 1C conductivity
detector can be processed by a Hewlett Packard 3385A chromatographic control
system. Peak areas and retention times are recorded to quantify and idc-ntify
specific anions.
Typical relative standard deviations for mixtures of sodium nitrate and
ammonium aqueous sultate standards are shown in Table 4. Retention times and
relative standard deviations were determined daily. The minimum detectable
levels for sulfates and nitrates were 10-1 neq ml-1 and 5 x 10-2 neq ml-1,
respectively. A typical 1C chre-natogram of an extract of a fine particle
sample collected in Philadelphia is shown in Fig. 6. Note the low nitrate
concentration as compared to sulfate. For almost all aerosol samples
collected with the dichotomous sampler using Fluoropore filters, nitrate was
always a small fraction of the sulfate concentration.
Table 3. Relative standard deviations for 1C analysis of nitrate
and sulfate standards.
Molarity
1 x 10-«
5 x 10-5
1 x 10-5
Dionex
attenuation setting
10
10
3
Standard deviation*
so-
1%
0.3%
1.6%
N03
1%
1%
1*
* Based on four consecutive injections of 0.4-ml volumes of standard solutions.
19
-------
f.HROMATOGRAMOF tXTRACT OF FINE PARTICLES FROM A
DICHOTOMOUS SAMPLER
ill!
SAMPLE fXTRACT FROM 24 HOUR FINE PARTICLE
FRACTION AT PHILADELPHIA. PA. FEB. 25. 1977
ANALYZER OIONEXMOO. 14.
ANION EXCHANGE COLUMN 50 cm > 3 mm 1.0.
TIME, mm
Figure 6. Ion chromatogram of anion content of fine particle sample collected
in Philadelphia, ?A.
20
-------
Ammonium analysis: The concentration of ammonium ions may be determined with
an ion selective NH3 gas diffusion electrode (Orion Model 95-10). The
response is recorded from a Coming-Digital 112 pH meter. Analysis is
performed by mixing 7 ml of filter extract and 0.2 ml of 5 N sodium hydroxide
in a 20-ml glass beaker. The electrode is immersed into the solution with the
tip extending approximately 1 cm into the solution. Readings are made between
2 and 3 min. after immersing the electrode into the extract. When the sample
concentration is less than 1 x 10-5 M, equilibration required as long as
5 min. The electrode is calibrated using dilute NH.C1 standards (10-2 to
10-6 M). The minimum detectable level is 3 neq ml-1. At concentrations above
10 neq ml-1, the relative standard deviation is typically ± 5%. Recently
Abbas and Tanner (20) developed a technique which measures ambient levels of
ammonia. The method is based on measuring the intense fluorescence of the
reaction product of NH, and a thio-mecaptan in solution. The method measures
continuously NH, at sub-ppb concentrations.
Carbon Measurements and Radioisotopic Analysis of 14C/12C: Stevens et al. (4)
have documented the basis for using Teflon filter substrates to obtain mass
measurements and perform XRF analyses for elemental composition. However,
carbon measurements are not easily obtained from Teflon filters, because the
procedure used most frequently to measure carbon in aerosols is based on
high-temperature combustion methods. For this reason, a second fine particle
sample must be collected on quartz at the same time the aerosols are being
collected on Teflon. Quartz is used as the collection medium because it has a
low carbon content, high collection efficiency, and is chemically inert. An
aliquot of the quartz filter can be analyzed by a combustion procedure similar
to one described by Stevens et al. (20). This procedure measures volatile
21
-------
carbon and elemental carbon from the same sample. Recent comparisons of
methods by Cadle et al. (21) to meas.ure elemental, total and volatile carbon
content of aerosols indicate that procedures that depend on combustion or
pyrolysis to convert the carbon species into methane followed by flame
ionization measurement, may produce substantially different results, depending
on the conditions for combustion. For example, to convert completely all the
aerosol carbon on quartz filters requires a temperature of at least 850°C and
the presence of oxygen in the helium carrier gas. Without oxygen, a
substantial amount of elemental carbon will not be pyrolyzed and consequently
will not be measured.
At the present time, we measure the elemental and condensable carbon
present in ambient particulate samples using a modified Dohrmann DC-50 organic
analyzer (DC-50). Figure 8 is a diagram depicting the sample introduction,
pyrolysis, combustion and detector assemblies. The DC-50 measures the organic
carbon (condensable organics) by pyrolyzing the carbonaceous portion of the
sample to CO, and other pyrolysis products at 650°C in a helium atmosphere
reduces the products to methane over a bed of nickel catalyst in an atmosphere
of hydrogen, and measures the methane with a flame ionization detector. After
the organic analysis, elemental carbon is determined by combustion of the
carbonaceous materials remaining on the quartz filter at 850°C in a 2% oxygen,
98% helium atmosphere. Total carbon is considered to be the sum of the
organic and elemental analysis. Recently, we have modified the procedure to
collect particles on quartz filters in order to minimize the variations in the
blanks. The modification reduces the deposit area of the sample from 37 mm to
13 mm, as shown in Figure 7. This focusing of the sample to a 13 mm area
reduces the impact of the carbon blank and the area outside the deposit area
becomes the blank for that filter. The combination of increased sample per
22
-------
•••Vi
B
FINE PARTICLE SAMPLES COLLECTED ON QUARTZ FILTERS.
SAMPLE A. COLLECTED BY MODIFYING SAMPLE FLOW TO
FOCUS AEROSOLS INTO A 13 mm SPOT. THIS PROVIDES A
HIGH DENSITY OF SAMPLE PER UNIT AREA AND REDUCES
BLANK UNCERTAINTIES.
SAMPLE B. COLLECTED NORMALLY IN DENVER, CO, JAN.,
1982. THE HOLES IN THE FILTERS ARE WHERE PIECES
WERE REMOVED FOR CARBON ANALYSIS.
Figure 7. Fine particles samples collected on quartz filters.
-------
TRACE AND
DIGITAL RESULTS
Figure 8. Diagram of modified Dorhmann DC-50 elemental and organic carbon analyzer.
-------
unit area and a blank for each sample should improve the precision of the
combustion procedure to determine carbon content of aerosols.
Currie (22) determined the relative amounts of particulate carbon from
contemporary and fossil fuel carbon combustion sources by means of data on the
l4C/f*C ratio. The distinction is possible because contemporary carbon
contains trace amounts of MC, but fossil fuel carbon is essentially devoid of
14C. The 14C/1:*C measurements are relatively expensive and only two or three
laboratories in the U.S are capable of performing the analysis. In addition,
milligram amounts of material are needed to measure 14C/lidC in ambient
particulate samples. Currie (23) is currently working on a new procedure that
measures 14C with an accelerator mass spectrometer at the University of
Arizona. With this new system, as little as 100 ug carbon may be analyzed.
The National Bureau of Standards (NBS) normally measures 14C/1V!C by
heating the quartz or glass fiber filters to distill the condensable organics
from the filters; the organics are then converted to CO. and the CiK is
condensed into a glass microcell. The cell is placed into a sealed chamber
and the radioactivity is measured with a sensitive proportional counter.
Recently, Stevens et al. (24) reported the presence of carbonates, acetate and
formate in aqueous extracts of ambient fine particles collected in the
Shenandoah Valley of Northern Virginia, U.S.A. Stevens et al. postulated
these species may interfere with 14C measurements, because the carbonate may
result from C02 adsorption on the quartz filter. Because almost all of the
ambient CO- is rich in 14C, the bulk 14C analysis described above may be
influenced by the C02 artifact formed with the quartz filter. For this reason
the improved 14C method described by Currie et al. (23) could be modified to
eliminate this problem through an additional step of organic extraction of the
ambient sample prior to i4C analysis. This extraction step would all but
25
-------
eliminate the interference due to adsorbed gases (002) on the quartz filters
because they would be lost during the evaporation of the solvent after the
filter extraction. In addition, measurement of the the 14C content of the
major classes of condensable organics is now possible and could enhance the
range of species used in receptor models. Another possibility for eliminating
the artifact influence would be fractional distillation of the organics from
the filter, followed by 14C analysis of the individual thermally relsused
organics. We would expect that only the fraction initially eluted from the
filter would contain most of the artifactual 14C.
Electron Microscopy For many aerosol characterization programs elemental
composition and/or ionic content is insufficient to determine the chemical and
physical properties of ambient aerosols. Either transmission electron
microscopy (TEM) or scanning electron microscopy (SEM) are in most cases the
only tools to study the properties of individual fine particlts. A
combination of SEM and TEM x-ray analysis (EDX) of the elemental content of
particles enhances the characterization. However, EDX cannot measure C, 0,
and N, and for certain types of particles, e.g., fly ash, element.il
composition of the particle provides insufficient information to identify the
particle, for this reason, size and shape coupled with elemental composition
provides the information to deduce the morphology of the particles.
More complex techniques are often needed to identify particles; most are
destructive techniques and include application of microchemical tests and
hot-stage electron microscopy for the characterization of micrometer and
sub-micrometer particles. Although these techniques have been used
successfully with OM, only a few microscopists have extended these techniques
to electron microscopes. Mamane (25), Mamane and de Pena (26), and Mamane and
26
-------
Puschel (27) have described the application of these techniques to the
analysis of a variety of ambient particle samples. The best examples of these
microparticle reaction tests are the SEM/EDX analysis of BaC^-treated
surfaces for sulfates (Figure 9) and nitron (C20Hi6N4; 4» 5-dihydro-l, 4
diphenyl-3, 5 phenylimino-1, 2, 4-triazol)-treated surfaces for nitrates
(Figure 10). The advantages of these techniques are as follows: a) the method
detects ions that cannot be seen by EDX (e.g., nitrates); b) they are
inexpensive, sensitive, specific and relatively rapid techniques to measure
sulfates and nitrates; c) provide information on mixed individual particles,
such as a mineral particle coated with a layer of nitrate or a fly ash
particle coated with sulfate (Figure 9), that cannot be obtained by other
conventional procedures..
With the hot-stage techniques, particles are exposed to a given
temperature while being observed by TEM. During the heating decomposition,
melting and sublimation take place. Although the information may not
positively identify the particle, these methods can be confirmatory,
eliminates certain possibilities and, with the use of other techniques,
assists in the identification of the particles.
Collection of particles for electron microscopy is completely different
from the collection for other bulk analytical techniques (as wet chemistry or
XRF). The following are important when collecting the particles: a) collect
into 3 to 4 size classes, b) particles should remain individual particles on
the collection substrate; (this normally requires that short sampling time be
used to collect the submicrometer particles and that long periods be used for
the larger particles). Mamane et al. (27) describe the use of the Casella
impactor to collect tne appropriate samples for SEM and TEM analysis.
Collection periods for submicrometer ambient particles is typically seconds
27
-------
COATED
H
SULFATE
Figure 9. Reaction spot of sulfate particles with the Bad- film, as viewed in
the transmission electron microscope. Both sulfate/ contain insoluable
nuclei, one of which is a fly ash particle. Sample was collected on Aug.
23, 1983, Deep Creek Lake, Maryland. Mixed sulfates were mostly
found in evening and night samples, almost none during mid-day.
28
-------
SULFATE
NITRATE «
Figure 10. Reaction spots of nitrate and sulfate particles with nitron
(2QH..,N4) film, as viewed in the transmission electron microscope.
Sample was collected on Aug. 20, 1983, Deep Creek Lake, Maryland.
Nitrates were found in particles larger than 1 urn and in very
limited concentration.
29
-------
and minutes for micrometer particles. Large particles (>10 urn) may need
sampling times of a few hours. These sampling constraints tend to make
comparisons between integrated measurements and St'M results difficult (29).
However, these problems can be reduced through careful selection of sampling
periods when bulk and particle-by-particle characterization are to be
compared.
X-ray Diffraction and Optical Microscopy Much interest in the composition and
source identification of atmospheric aerosols has developed over recent years.
In most studies prior to the last decade, much of the compositional
characteristics of particulates was inferred from wet chemical analyses,
bulk-sample XRF analyses, and optical scattering characteristics. Some work
with optical polarizing microscopy yielded result,, for the major components,
but only for those particulates at the coarse end of the size distribution.
During the past decada much more effort has been expended toward
development of techniques yielding direct compound identification. An example
of a quantitative study of this type is found in the report of Bradway and
Record (30) in which optical polarizing analysis of mineral particulates
trapped on glass fiber (Hi-vol) filters is given for 14 urban areas. In the
work of Stevens et al. (4), particulate compositions were deduced from ion
equivalents required by stoichiometry of the compounds with ion
chromatography, Thorin spectrophotometry, and x-iay fluorescence being
employed as the analytical methods. Although not a direct observational
technique, infrared absorption analysis does provide evidence of the presence
of certain compounds according to the observed bond frequencies. Work of
Adler and Kerr (31) and Moharram and Sowelim (32) provide infrared information
on the occurrence of sulfate ions in various sulfate minerals. Some
30
-------
participate compounds may be identified by morphological characteristics and
their composition deduced by single-particle microbeam scanning electron
techniques that will be discussed later. Perhaps the most powerful tool for
direct compound identification, both from a qualitative and a quantitative
standpoint, is XRD. Difficulty in application of this technique to aerosols
comes primarily from the low mass coverage of the particulates on the
collecting filter media. Nevertheless, good qualitative studies reporting
identification of atmospheric aerosols from filters have been completed by
Biggins and Harrison (33), Thompson et al. (34), O'Connor and Jaklevic (35),
and Brosset et al. (36). Several of these studies included a separation of
the particulate mass into size fractions to identify the dominant compounds in
each.
During the past five years, the Institute of Atmospheric Sciences' (IAS)
Cloud Physics Laboratory at the South Dakota School of Mines, Rapid City, SD,
has worked toward the development of a quantitative technique for analysis of
particulate matter collected on various filter substrates. The results of
these studies have been reported by Davis et al. (37) and Davis and Johnson
(38 and 30). The basis for the technique is the reference intensity method
combined with direct beam x-ray transmission measurements. These combined
techniques allow the simultaneous determination of crystalline components and
the major classes of amorphous components present in thin aerosol layers
collected on filter substrates. Analyses may be completed using Teflon, glass
fiber, cellulose acetate, quartz, and polycarbonate filter media.
The XRD procedure developed by Davis et al. (40) to measure the quartz
content in ambient particles collected on Teflon filters is based on a mass
calibration procedure. The basis of this calibration procedure rests on the
provision that, for a monodisperse layer on a thin membrane filter, overlap is
31
-------
negligible and matrix absorption can be ignored. For 3 urn radius spheres of
2.65 density (quartz), such particles can accumulate to the extent of 900 ug
cm-2 before a monodisperse layer is formed on the filter. For these
conditions, therefore, a given laboratory instrument under proper quality
assurance procedures may maintain the target x-ray flux at a level
sufficiently invariant to complete direct mass calibration analyses over an
indefinite period of time. A Norelco diffractometer using a copper target
raised to 40 kV and 20 mA power was used for these analyses. Samples were
mounted on a Norelco high-angle goniometer fitted with a theta compensator,
graphite crystal monochronometer, and spinning sample mount. All samples were
scanned from 5° 26 to 40° 26 to make a semi-quantitative evaluation of other
components. The ratio of peak height to reference intensity constant for each
observed component provided a visual estimate of relative abundances for the
semi-quantitative evaluation. The quartz peak at 26.67° 26 (hkl = 101) was
used as a basis for the quantitative calibration and analysis.
In conjunction with the reference intensity analysis program, a number of
quartz samples were analyzed for scattering power (reference intensity
constant values): the Tuscarora quartz was selected from this group as
representative of quartz in the environment, the reference intensity constant
being close to the average of the set of eight samples analyzed. A linear
calibration curve was obtained between ug cm-2 and integrated intensity at the
26.67° quartz peak based on a total of 12 calibration analysis completed by
loading Teflon filters with a specified weight of fine particulate Tuscarora
quartz using an aerosol suspension technique described previously by Davis et
al. (40). In the initial data set, the regression on the experimental points
yielded a slope of 463 counts cm2 ug-1 with an intercept of 5.1 ug cm-2. The
limit of detection for quartz was found to be 0.6 ug cm-2 for particles
collected on Teflon filters.
32
-------
These studies have included a formal "variance error" analysis based on
uncertainties known to exist or anticipated in the various measurements and
physical parameters. In general, weight quantities are known to within 10%
for major components in the sample. For minor components (i.e., less than 10%
of the total mass) errors are quite variable, but may be as high as 100% or
greater in some cases.
Recently, Davis et al. (40) analyzed 104 pairs of Teflon filters
collected in 1980 from 22 U.S. cities as part of the EPA's Inhalable
Participate Network (41). The filters were analyzed for elemental composition
by EDXRF and for mineral content by an XRC procedure with the emphasis on the
quartz content. Table 4A shows the elemental composition and quartz content
of selected fine and coarse particles from four U.S. cities. Table 48 lists
minerals that were observed to be present in almost all particle samples
collected in the U.S. and their chemical composition.
From the analysis of these dichotomous filters from the 22 U.S. cities,
it was noted that the continental interior sites showed the highest average
ambient quartz concentration as well as the greatest variability. Coastal
regions and eastern interior sites, with few exceptions, showed the lowest
quartz concentrations in the ambient air. Perhaps the most notable exceptions
to this generalization are the sites of Inglenook and Tarrant, located in
industrialized north Birmingham, AL, and El Paso, TX, where a quartz level was
observed that was lower than expected for an arid inland site.
Comparison of these data with analyses made with polarizing OM by Bradway
and Record (30) shows some interesting relationships. The filters studied by
Bradway and Record were obtained from standard Hi-vol samplers of the National
Aerometric Sampling Network (NASN), which did not have a wind insensitive
sampling inlet. Therefore, the concentrations of quartz observed from these
33
-------
Table 4a. Elemental composition and quartz content of fine and coarse
particles from three U.S. cities.
TYPICAL MASS-ELEMENTAL AND QUARTZ CONCENTRATION
IN M9/rr>3 AT SELECTED SITES IN THE UNITED STATES
MASS
AL
SI
IS
Cl
K
Ca
V
Fe
Ni
Cu
Zn
Br
|Ph
QUARTZ
BOSTON
FINE
34.9
-
0.144
3.869|
-
.096
.070
.020
.1211
.012
.035
.046
.020
.285 |
0
, MASS
COARSE
105.9
13.458
I6.760
.502
.301
.533
1 .069.
.008
1.612
.022
.023
.054
.025
.177
I 8.00 |
KANSAS CITY. MO
FINE COARSE
25.7
.091
.434
pTsTel
-
.311
.519
-
.189
.002
.032
.034
.027
1. 180,1
0
44.7
2.053
4.542
.215
.
.349
3.852
.
.800
.003
.016
.026
.009
.057
4.70 I
RIVERSIDE. CA
FINE
35.2
0.036
.234
|l.653|
' .009
.120
.301
.003
.127
.007
.040
.029
.037
1 .376 1
0
COARSE
71.0
13.513
17.544
.720
.164
.961
4.781
-
1.888
.006
.021
.030
.028
.113
rs.oo j
Table 4b. Minerals commonly found in ambient particles.
MINERALS COMMONLY PRESENT IN AMBIENT PARTICLES THAT
CAN BE MEASURED BY X-RAY DIFFRACTION METHODOLOGY
MINERAL NAME COMPOSITION
BIOriTE
MUSCOVITE
GYPSUM
KALONITE
CALCITE
PLAGIOCLASE
DOLOMITE
HEMATITE
IVAGNETITE
ANGLESITE
MASCAGNITE
1HENARDI1E
'.lODANITER
CHLORITE
K2MgFe3|FeAI)AljSi3)02 (OH)3
KAI2(AISi3010)(GH)2
CAS04-2H20
(Fe.al)4Si401o(OH)8
CaC03
0.55 (NaAISijOs) + 0.45 [CaAI2Si2O8]
CaMg(C03)2
Fe203
FeO F203
PBSO4
[NH4]jSO4
NaN03
NaCI
34
-------
analyses should be considerably higher, in terms of weight percent of total
suspended particulate (TSP), than those values obtained from the dichotomous
sampler. Such is actually observed by comparison of the Bradway and Record
(30) data with those of this paper. Three cities that were common to both the
Bradway and Record (30) OM and the Davis et'al. (40) XRD, OM particle
classification studies were: North Birmingham, (Tarrand and Inglenook), St.
Louis, MO, and Cincinnati, OH. In the Birmingham study, the Bradway and
Record analyses showed 21 weight per cent quartz as an average of six filters,
with the aerodynamic diameter particle size ranging from 5 to 60 urn with a
median of about 15-um diameter. This high level may be contrasted to the 4-7%
observed in the dichotomous coarse fraction. Similarly, high values were
observed by Bradway and Record in St. Louis (26 weight per cent for the
26-filter average, 6 sites). The presence of 40% calcite in the optical study
agrees with our observation by XRO that calcite and dolomite constitute a
significant fraction of the St. Louis filter mass. A similar result was
obtained at Cincinnati where 19% quartz and 22% calcite was observed optically
for the 20-filter average.
Quartz is often a major source of silicon in the coarse particle fraction
of ambient aerosols. Feldspars, micas, and clays, however, contribute
significant amounts of silicon in certain localities.
Davis1 et al. (40) examination of the relationship between the silicon
and quartz concentration for Buffalo, NY indicates that quartz was the major
contributor of silicon, whereas less of a relative contribution from quartz
was found in the samples from Birmingham and only very little was found from
the Portland, OR site. Indeed, the examination of the XRD charts revealed
that feldspars were the primary crustal component of the aerosols at Portland,
OR, whereas at Buffalo only an occasional occurrence of feldspars and micas
35
-------
were observed; the other constituents were primarily carbonates and sulfates.
This feature is also consistent with the higher Al and K contents of the
Portland, OR sample.
DENUDER SAMPLERS
Nitrate and Nitric Acid Measurements: Evidence is accumulating that indicates
the importance of acidic nitrate in atmosphere deposition processes. For
example, Lewis and Grant (42) have recently reported measurements of
precipitation chemistry from a rural site in the Colorado Rockies and suggest
that a decrease of nearly one pH unit over a period of four years was due to
increasing nitric acid. The nature of nitrate deposition depends on the
distribution of nitrates between gas and solid phases. A number of
investigators (43) and (44) have recently made observations that indicate
that, in both rural and urban atmospheres, gaseous nitrate concentrations are
considerably larger than those of particulate nitrate. The contrasting
behavior of atmospheric nitric acid (HNO_), which appears in the gas phase,
and sulfuric acid (H-SO.), which appears in particulate form, is consistent
with their relative vapor pressures and heats of hydration.
The determination of background or rural values for nitrates is
difficult; even at 10,000 ft in the Rocky Mountains, Kelley et al. (45) have
observed changes in HNO, concentrations of a factor of 100 depending on
whether the wind was from the east or west. It is clear that techniques for
measuring atmospheric nitrates should combine high sensitivity with
unambiguous discrimination between gaseous and particulate nitrates.
As investigators have improved methods of sampling and analysis for
atmospheric nitrate, it has become evident that the distribution of
atmospheric nitrate between the gaseous and particulate phases has been masked
36
-------
by experimental artifacts. Many early participate nitrate data were based on
analyses of extracts from glass fiber aerosol filters used in Hi-vol samplers.
It is now known that these filters contain active sites that fix gaseous HN03
and make it appear as particulate nitrate (46, 47). Other filter materials
have also been shown to react with and collect gaseous HNO, and create a
positive particulate nitrate artifact (48). The use of an inert filter
material such as Teflon removes the "positive artifact" problem, except for
the possibility of reaction with the collected aerosol particles. It has been
shown, however, that collected aerosol nitrate particles (true particulate
nitrate) may be lost from filters due to reactions with other materials or to
evaporation. Loss of particulate nitrate is known as "negative artifact."
Reactive loss may occur if, for example, H?SO. aerosol comes in contact on the
filter surface with nitrate aerosol. Evaporative loss may occur due to
decreases in ambient gas concentrations causing the solid and gaseous nitrate
phases to no longer be in equilibrium. Thus we see that, on the one hand,
measurements of particulate nitrate using glass fiber filters are expected to
be systematically high. If the recent measurements mentioned above of the
distribution between HNO- and particulace nitrate are correct, the glass fiber
filter nitrate overestimates may be considerable. On the other hand,
measurements of particulate nitrate using inert Teflon filters are expected to
be systematically low, but the extent of loss due to reaction and evaporation
are difficult to predict.
Given the free energy change of vaporization for NH.NO^, it is
straightforward to calculate the vapor pressures of NH, and HNO^ in
equilibrium with NH.NO. in the solid phase (49). These calculations show
that at 10VC and 30VC the gas-phase equilibrium concentrations of both NH3 and
HNO, required to sustain the solid phase are s4 and 12 ppb, respectively.
37
-------
Tang (50) carried out considerably more detailed studies of the vapor
pressures of HNO, and NH, in equilibrium with solution droplets of ammonium
salts and has estimated the dependence of these vapor pressures on humidity
and droplet acidity. Tang's work provides quantitative estimates of the
increase of HNO, and decrease of NH, vapor pressi-res with increasing acidity
and of the decrease of HNO- and NH, vapor pressures with increasing humidity.
Given the results of these calculations, for estimating gas phase
concentrations of NH. and HNO- required to support solid or solution phase
nitrate and given the typical observed values of ambient NH, and HNO,
concentrations, one may wonder why nitrates are ever observed in particles
except in extraordinary circumstance?. The components of the atmosphere,
however, are generally not at thermodynamic equilibrium and the rate of
approach to equilibrium between the solid and gas phase may be decreased by
the intervention of other materials or phases. For example, the vapor
pressure of water above a saturated solution of NH.NO, is surh that the solid
phase is hygroscopic at relative humidites >60% over the temperature range
20-30VC. This means that at relative humidities exceeding 60%, particles of
NH.NO, will accumulate an aqueous layer. As long as the solid phase is
present, the equilibrium vapor pressures of NH, and HNO- will be unchanged,
but the rate of approach to equilibrium will be retardtd. Alternatively, a
particle may accumulate a layer of atmospheric organic material that would
retard the loss of material from tha solid to the gas phase. Because the
nonpolar organic molecules will reduce the dielectric constant jf the
solution, it is energetically favorable for them to remain on the surface of
the particle; hence, a skin may be formed. Chang and Hill (51) have shown
that aqueous drop evaporation is retarded by the accumulation of gaseous
mixtures of ozone and olefins and they provide references to earlier work on
38
-------
the reduction of evaporation by organic films. Apparently, losses of
atmospheric particles due to evaporation are strongly dependent on other
components in the atmosphere. It may be for these reasons that nitrate in
aerosols is stabilized with respect to evaporation.
Denuder Difference Method: In order to avoid both positive and negative
nitrate artifacts, a sampling system developed by Shaw et al. (52) has been
tested and has been used in a number of EPA field studies. The system, termed
the denuder difference method (ODM), is shown in Figure 11 a. Figure lib is a
photograph of a combined DOM and fine particle collection device to Measure
sulfate, fine particle mass, and SO-. The experimental set-up for the DDM
consists of Teflon cyclone to remove the coarse particles. The ambient air
passes through the cyclone at - 30 L/min, and into a manifold where two
parallel samples are collected at 3 L/min downstream of two tubes. One of the
tubes, the denuder, is coated with MgO; the other tube, constructed of Teflon,
is uncoated. The residence time of the gas and fine particles in the denuder
is 0.2 s. After the air sample exits the tubes, it passes through a 25-mm
diameter, 1-um pore size Membrana nylon filter. The MgO removes the HNO, and
the true fine particle nitrate is collected on the nylon filter. On the other
n>lon filter HNO., + fine particle nitrate is collected. The difference
between the HNO, •»• nitrate sample and the fine nitrate is HMO,, hence the name
of the method.
The nylon filters are removed and stored in a sealed dessicator at 5°C
until they are extracted. The nylon filters are extracted in 1 x 10 NaHCOj
solution in an ultra:- -«ic bath. The extract is analyzed by ion exchange
chromatography for nitrate content. The precision of the method for 19
replicates is: nitrata = 0.1 ug/m3, nitric acid - 0.2 ug/m3. The precision
refers to an ambient air sample size of 2.0 m3. Figure 12 is a bar graph that
39
-------
CYCLONE
(2.5 Jim CUT POINT)
MANIFOLD
281/mte
t> l/min
PUMP
FLOW RESTRICTORS, 3 l/min EACH ____
1. DENUDER, UgO COATED (REMOVES HN03)
2. FILTER, NYLON (COLLECTS NITRATE AEROSOLS)
3. FILTER, NYLON (COLLECTS HNOg, NITRATES, AEROSOLS)
4. DIFFERENTIAL FLOW CONTROLLER
Figure 11s. Denuder difference method (DOM) sampler.
Figure lib. Photograph of ODM samples and fine particle collector.
40
-------
ou
64
^
E
a.
z'
o
P 48
ec
z .
UJ
z
S32
oc
z
16
o
I
I
1
1
I 1
OENUOER DIFFERENCE EXPERIMENT
O ARTICULATE
-
"
-
-
.
-
i
17 18
>
20
p
^
%
'%%
21
-
MiHM
I
•%
tt
^
1
^
-
-
i
22 23
DAY OF MONTH (AUGUST. 1980)
Figure 12. Bar graph of denuder difference results for samples collected
in N.C., 1982.
41
-------
compares measurements of gaseous nitrate and particle nitrate obtained using
the DDM during the summer of 1980 in Research Triangle Park, NC.
Annular Denuders: During the past two years a new design for denuders has
been developed and tested at the Laboratory for Atmospheric Pollution of
C.N.R., Rome, (Italy), which has led to a device described by Possanzini et
al. (54) that is suitable for the simultaneous measurement of different
species contributing to deposition.
The C.N.R. denuder is an annular tube configuration (Fig. 13). During
sampling, air is drawn under laminar flow conditions through the annular space
between two concentric glass cylinders that have been coated with a chemical
that reacts with selected trace species. As the sample stream passes through
the annular space, the gaseous trace species travel by molecular diffusion
from the bulk gas to the reactive surface and are collected.
the reader is referred to the work of Pozzanini, et al. (54) for a
discussion of annular denuder tube theory. A description of the ADM design
characteristics and operating parameters used to collect ambient samples is
shown in Table 5.
For a given flow the annular denuder method (ADM) can achieve equivalent
collection efficiency in M./30 of the length required for an open-tube
denuder, or for a given denuder length, the annular denuder can sample at ~ 30
times the flow of an open-tube denuder.
The high operating flow rate (10-30 2min ) makes the annular denuder
very useful for experiments where collection of low concentration of certain
gases are required over short term sampling (1-4 h) periods. The high flow
rate also permits more material to be collected on membrane filters downstream
of the denuder. Previously, collection of large quantities of particles while
simultaneously preserving the integrity of the particles through removal of
42
-------
CASES AND
PARTICLES
CASES DEPOSITED ON WALLS
fANO-OFF
STRIPPED CASES
AND PARTICLES
Figure 13. Annular denuder
43
-------
reactive gases (eg, HN03> NH3, and S02) required that the filter be preceded
by a parallel multitube denuder assembly similar to that described by Stevens,
et al. (4).
For a denuder to be effective, the system needs to ensure separation of
the gases and particles. However, diffusional and inertial deposition at the
inlet can result in particle uptake which has been determined experimentally
(Possanzini et al., 54) to be not larger than about 3%. The transit time of
air through the denuder is <0.1 second, reducing the opportunity for
substantial disturbance of the atmospheric gas-particle equilibrium existing
in the atmosphere. The walls of the denuder are etched by sand blasting with
100 urn sand particles. This feature increases the surface area available for
chemical coating, and, as a result, the capacity of the denuder to collect the
pollutants of interest can be increased to several milligrams. A diagram of
the annular denuder system which was evaluated by the EPA in Research Triangle
Park, NC is shown in Figure 14.
The use of a water soluble and 1C compatible substrate to coat the walls
of the annular tube (e.g., ^CO-) shown in fig. 14 simplifies the extraction
and the analysis of the sample. Other substances deposited on the denuder may
give rise to the formation of the same ions; for example, deposition of
particulate matter containing chlorides, sulfates, and nitrates interfere with
the measurement of HC1, SO. and HNO,. The absorption of N02 and PAN on Na-CO,
yield nitrites which interfere with the measurement of HNO-. However, the
efficiency for the collection of these interfering species is relatively small
(about 1 to 3%) (Perm and Sjodin, 55). Therefore, the ADM uses two annular
denuders connected in series as shown in Figure 14 in field studies where SO.,
HNO., nitrate sulfates and related species are to be measured. Thus, in this
configuration the amount of relatively unreactive interferents collected in
44 .
-------
HNOj,HCI.S02.HONO<
PNEUMATIC FLOW CONTROLLER
,NYLON FILTER
^TEFLON FILTER
^—Benuder #2
PUMP
-Connector
>Ni2COj- GLYCERINE
COATING
Senuder fl
• TOTAL FLOW
ADJUSTER
TEFLON CYCLONE
Ill/Mi*
Figure 14. Annular denuder: a filter oack system used to collect HNO,, HC1,
S02, N03, HONO, S04, and H .
45
-------
the first denuder will be approximately equal to that found in the second
denuder. T! is feature can be used to correct data obtained from the analysis
of the first denuder. The use of two denuders in series will then permit the
simultaneous analysis of several acidic compounds, even though the ratio of
analytes in the gas phase and participate matter is extremely low. For
instance, the technique would be is valuable for the measurement of trace
levels (< 0.1 ug/ma) of SO- and HN03 in the presence of large quantities of
sulfates and nitrates in particulate matter, and, in addition, the use of two
denuders will permit the measurement of small amounts of HNO-.
A typical chromatogram obtained from 1C analysis of annular denuder
extracts is shown in Fig. 15. There is no visual evidence of SOT or N03 in
the second dnuder; this shows the high collection efficiency of the first
denuder for HNO- and SO- and implies that deposition of pirticles in the
denuder is low. Absence of NO- in tne second denuder is interpreted as an
indication that the NO- in the first denuder is due to a very reactive species
such as HNO- rather than PAN or NO-, which react much more slowly with the
Na-C03 denuder surface (Perm and Sjb'din, 55).
A sample collection system based on annular denuders followed by
particulate filters appears to be a very promising system for measurement of
HN03, N03 S02, SO^. NH3, NH* and H*. The system's irain features are: (a)
operation at a relatively high flow rate while maintaining a collection
efficiency greater than 95%; (b) use of denuder coatings which are extractable
in water and compatible with conventional 1C analysis; (c) single flow train
and correspondinn reduction in the number (and cost) of flow control devices;
(d) relatively easy to set up and operate; (e) all gases of interest are
removed from the sample stream by denuders prior to passage of the sample
stream through any filter medium. In this configuration the acidification or
46
-------
S04 (S02)
REPLICATE 1
MARCH 27, 1985
1035 -1 605 hrj
15 l/min
$02* • 6-
HNO3-1.55j;g/m3
HNQ2 • 0.4 pg/m3
I ru
REPLICATE 2
URST N*2CO3 DENUD*R
SECOND N«2COj OENUOER
Figure 15.
Typical ion chromatogram of parallel annular denuders which
have collected ambient air samples. The second denuder after
the first denuder does not contain measurable amounts of SOT or
N03 indicating the e.ficiency of the first denuder.
47
-------
neutralization of particles on filters is minimised. The configuration of an
annular denuder assembly: filter pack system built and being tested by the CNR
Research Group in Italy to measure the aforementioned species is shown in Fig.
16. .
Table 5. Characteristics of an Annular Denuder System.
ADM
Denuder:
Annular Space:
Coating:
Flow Rate:
Filters:
Species Measured:
Analysis:
381 x 38 mm
1.5 mm
Na2C03: glycerine
15 LPM
47 mm 2 urn Teflon
47 mm 1 urn Nylon
HN03, S02, N0~, S
NH3, (HN02), (H+)
1C, Colorimetry
48
-------
(
200 mm
f
q
H
i
r
U ol
i
t
2
i i
c
) F
HC
h{XjI_j
^ SHUTO
VALVE
)UAL
ILTER
)LDER
A
120mm
i
1
b 01 :
i
.<
j_
1
^ •!-•
I-"'"1 '^
') 0 1
i
i
J
1 — A
V~ 7
i
J
i
)
Txrlt^
!£>Mfa
s.
1
\
\s
.J
c
1
c
c
ll
90 mm
90m
r -;
DENUDE R
TRANSPORT
CONTAINER
cL
) o fl
1 ~ " * r
)
3
c f
1
m
TEFLON MANIFOLD
-Li
-
y~
) o Q
"r
— •£
>o 6
1
1
) ,
VEFLON CYCLONE
COATING [POLLUTANT \
1 ''
1 ?
Kl. NtAi ||
1 i
ii
.. NYLON [j
j" TEFLON ||
,|NK
Ii
ll
t
CITRIC
ACID I
1 '
* !!
t
N.jC03 1 NC
'l
1
1
1
i "
II
1'
, 1
ll
II
II
"!MC
N.2C03 J,HN
Jjl
.1
1
i * - !•
i i!
PUMP ( (.
Til
i \
NO, 1 1
H*. 1
44.S04'[
1
?! !
i ! i
: i
i
i
i /
2. HN02> /
i |
I 1
1
1. MNO2. l /
°3- ~°2 ' 1
1-! /
RtMOTC
PROGRAMMABLE
FLOW CONTROL
Figure 16. US-Italy annular dehuder acid deposition sampling system.
49
-------
Conclusions
The major conclusions that can be drawn from this review of sampling and
analytical methods to measure the composition of ambient particulate samples
are as follows:
1. Collection of aerosols into two size ranges, fine and coarse fractions,
simplifies the chemical analysis and preserves chemical integrity of acid
species present in fine particles by removing the alkaline coarse
particles before the particles are deposited onto the appropriate filter.
2. Two-micron pore size Teflon filters are recommended for particle
collection when mass, elemental and ionic composition, and x-ray
diffraction properties of the particles are to be measured.
3. Energy dispersive X-ray Fluorescence (XRF) Analysis and Neutron
Activation Analysis are ideal analytical methods to measure the elemental
composition of most of the elements present in ambient particles. XRF
methods are preferred, because the analysis is completely
non-destructive, highly sensitive, and can measure to within ± 10% such
elements as Pb, Br, Cl, S, Cu, Si, Fe, Cu, K, As, Se and Cd at levels
typically found in urban atmospheres.
4. Quartz filters are recommended when the carbon content of the particulate
matter is to be measured. Combustion of the particulate samples at 650°C
and 850°C differentiates the elemental or soot carbon from condensable
organics present in ambient aerosols.
. 50
-------
5. Sampling aerosols for microscooic or scanning electron microscopic
analysis requires much shorter collection periods than for bulk elemental
composition measurements.
6. Aerosol nitrate and nitric acid can only be reliably measured by use of
HNO- denuders and nylon (or equivalent) filter materials.
7. Ion exchange chromatographic methods are recommended for the measurement
of most anions and cations present in ambient particulate samples. If
the appropriate precautions are followed, sulfites that may oe if present
in ambient particles can also be measured by ion chromatographic
procedures.
8. X-ray diffraction can measure mineral content of particulate if the mass
of particles per unit area on the filter is >_ 200 ug/cm and soil
standards from the airshed are available.
51
-------
Acknowledgements:
The author wishes to express his gratitude to Dr. Glen Gordon of the
University of Maryland and Dr. John Watson of Desert Research Institute for
permitting use of portions of their reports on receptor modeling as part of
this review. I also wish to thank Dr. Thomas Dzubay of the EPA and Dr. Briant
Davis, South Dakota School of Mines, for their contributions. I wish to thank
Gloria Gallant for typing the manuscript.
Disclaimer:
Although most of the research described in this review article has been
conducted at the U.S. Environmental Protection Agency, it has not been
subjected to Agency review and therefore does not necessarily reflect the
views of the Agency, and no official endorsement should be inferred.
52
-------
References
1. Stevens R.K. and Pace T. (1984) Overview of mathematical and empirical
receptor models workshop. Atmospheric Environment. In press.
2. Stevens R.K. (1979) Current methods to measure atmospheric nitric acid
and nitrate artifacts. Report EPA-600/2-79-051, U.S. Environmental
Protection Agency, Research Triangle Park, NC.
3. Liu B. , Pui D. , Rubow K. and Kuhley G. (1978) Research on air sampling
filter media. Progress Report — Grant Report R804600, Univ. of MN,
Minneapolis, MN.
4. Stevens R. K. , Dzubay T. G., Russwurm G. M. and Rickel D. (1978) Sampling
and analysis of atmospheric sulfate and related species. Atmospheric
Environment 12, 55-68.
5. Huntzicker J.J., Johnson P. L. , Shah J. J. and Coy R. A. (1982) An
evolution of organic and elemental carbon in ambient aerosols by a
thermal-optical method. In Atmospheric Life Cycle 77-78 (Edited by Wolff,
G.T. and Klirnisch, R. L.) Plenum Press, New York.
6. John W. and Reischl Georg (1980) A cyclone for size-selective sampling of
ambient air. J_._ Air Poll. Control Assoc. 30, 872-876.
7. Lippmann M. and Chan T. Cyclone sampler performance. (1979) Staub
Reinhart. Luft 39: 7-11.
8. Jaklevic J. M., Gatti R. C., Goulding F. S and Loo B. W. (1981) A p-gauge
method applied to aerosol samples. Envir. Sci. Technol. 15, 680-686.
9. Courtney W. J., Shaw R.W., Jr. and Dzubay f. G. (1S82) Precision and
accuracy of a p-gauge for aerosol mass determination. Envir. Sci.
Technol. 16, 236-239.
10. Scott D. R., Loseke W. A., Holbroke L. E. and Thompson R. J. (1976) Appl.
Spectrosc.. 30 (4) 392.
11. Jaklevic J. M. , Landis D. A. and Goulding F. S. (1976) Energy dispersive
x-ray fluorescence spectrometry using pulsed x-ray excitation. Adv j_n
X-ray Anal. 19, 253-265, Kendall/Hunt, Dubuque, 10.
12. Currie L.A. (1977) Detection and quantisation in x-ray fluorescence
spectrometry. In X-ray Fluorescence Analysis £f Environmental Samples,
pp. 289-306, Ann Arbor, Ann Arbor, MI.
13. Jaklevic J.M., Goulding F.S., Jarrett B.V. and Meng, J.D. (1974)
Application of x-ray fluorescence techniques to measure elemental
composition of particles in the atmosphere. Analytical Methods Applled to
Air Pollution Measurements (Edited by Stevens, R.K.), pp. 123-146, Ann
Arbor, Ann Arbor, MI.
53
-------
14. Kowalczyk G.S., Gordon G.E. and Rheingrover S.W. (1982) Identification of
atmospheric participate sources in Washington, DC using chemical element
balances. Envir. Sci. Techno!. 16, 79-83.
15. Gordon G.E., Rheingrover S.W. and Aras N.K. (1982) Determining
compositions of particles from major sources from ambient samples. Trans.
Am. Nucl. Soc. 41, 199-.
16. Watson J.G. (1979) Chemical element balance receptor model methodology
for assessing the sources of fine and total suspended particulate matter
in Portland, OR. Ph.D. dissertation, Oregon Graduate Center, Beavevton,
OR.
17. Core J.E., Hanrahan P.L., and Cooper J.A. (1981) Air particulate control
strategy development: a new approach using chemical mass balance methods.
In Atmos. Aerosol.: Source/Air Quality Relationships (Edited by Macias
and Hopke) ACS Sym. Series No. 167, American Chemical Society,
Washington, DC.
18. Brosset C. and Perm M. (1976) Man-made airborne acidity and its
determination. IVL Publications B-34B, Gothenburg, Sweden.
19. Mulik J. , Puckett R. , Williams D. and Sawicki E. , (1976) Ion
chromaiographic analysis of sulfate and nitrate in ambient aerosol.
Analyt. Lett. 9, 653-663.
20. Abbas, R. and Tanner, R. (1981) Continuous Determination of Gaseous
ammonia in the ambient atmosphere using fluorescence derivatization.
Atmospheric Environment 15, 277-281.
21. Stevens R.K., McClenny W.A., Dzubay T.G., Mason M.A. and Courtney W.J.
(1982) Analytical methods to measure carbonaceous content of aerosols. In
Atmospheric Life Cycle. (Edited by Wolff, G.T. and Klimisch, R.L.), pp.
111-129, Plenum Press, New York.
22. Cadle S.H., Grohlicki P.O. and Mulawa P.A. (1981} Problems in the
sampling and analysis of carbon particulate. General Motors Research
Publication, GMR-3935, ENV #19; available from GM Research Laboratories,
Warren, MI 48090.
23. Currie L.A. (1982) Contemporary particulate carbon. In Particulate
Carbon. Atmospheric Life Cycle (edited by Wolff G.T. and Klimisch R.L.),
pp. 245-260, Plenum Press, New York.
23. Currie L.A., Klouda G.A., Continetti R.E., Kaplan I.R., Wong W.W., Dzubay
T.G. and Stevens R.K. (1983) On the -irjin of carbonaceous particles in
American cities: results of radiocarbon 'dating1 and chemical
characterization. Proc. llth Int. Radiocarbon Conference. Vol. 25,
(edited by Stuiver, M. and Kra, R.).
25. Stevens R.K., Dzubay T.G., Lewis C.W. and Shaw R.W., J,\ (19S4) Source
apportionment methods applied to the determination of the origin of
ambient aerosols that affect visibility in forested areas. Atmospheric
Environment 18, 261.
54
-------
26. Mamane Y. (1977) A quantitative method for the detection of individual
submicron sulfate particles. Ph.D. thesis, 142 pp. Pennsylvania State
University, University Park, PA 16802.
27. Mamane Y. and de Pena R. (1978) A quantitative method for the detection
of individual submicrometer size sulfjte particles. Atmospheric
Environment 12, 69-82.
28. Mamane Y. and Pueschel R.F, (1980) A method for the detection of
individual nitrate particles. Atmospheric Environment 14, 629-639.
29. Mamane Y. and Noll K. (1984) The use of micro-spot techniques to provide
evidence of heterogeneous sulfur oxide and nitrogen oxide chemistry.
Presented at National Symposium on Recent Advances in Pollutant
Monitoring of Ambient Air and Stationary Sources, Raleigh, NC
30. Bradway R. M., Record F. A. and Pace T. G. (1976) Application of
polarizing microscopy to the characterization of ambient suspended
particulate. Proceed. Ann. Meet. Fed. Analy. Chem. Spectres. Soc. Phila,
PA Standard Oil Co., Cleveland, OH.
31. Bradway R.M. and Record F.A. (1976) National assessment of the urban
particulate problem II: particle characterization Report EPA-450;
3-76-025, Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency.
32. Adler H.H. and Kerr P.F. (1965) Variations in infrared spectra, molecular
symmetry, and site symmetry of sulfate minerals. Am. Min. 50, 132-147.
33. Moharram M.A. and Sowelim M.A. (1980) Infrared study of minerals and
compounds in atmospheric dust fall in Cairo. Atmospheric Environment 14,
853-856.
34. Biggins P.D.E. and Harrison R.M. (1979) Atmospheric chemistry of
automotive lead. Envir. Sci. Techno!. 13, 558-565.
35. Thompson A.C., Jaklevic J.M., O'Connor B.H. and Morris C.M. (1982) X-ray
powder diffraction system for chemical speciation of particulate aerosol
samples. Nuclear Instruments and Methods, 198. 539-546.
36. O'Connor B.H. and Jaklevic J.M. (1980) X-ray diffractometry of airborne
particulates deposited on membrane filters. X-ray Spectrometry 9, 60-65.
37. Brosset C., Andreasson K., and Ferm M. (1975) The nature and possible
origin of acid particles observed at the Swedish west coast. Atmospheric
Environment 9, 631-642.
38. Davis B.L., Johnson L.R., Stevens R.K., Gatz D.F. and Stensland G.J.
(1982) Observations of sulfate compounds of filter substrates by means of
X-ray diffraction. In Am. Geoph. Union Monogr. 26, 149-156.
39. Davis B.L. and Johnson L.R. (1982) Sample preparation and methodology for
x-ray quantitative analysis of thin aerosol layers deposited on glass
fiber and membrane filters. Adv. jn X-ray Analysis 25, 295-300.
55
-------
40. Davis B.L. and Johnson L.R. (1979) X-ray diffraction analysis in aerosol
provenance studies. Preprints of the 72nd Annual Meeting, A.I.Ch.E., San
Francisco, paper 118c.
41. Davis B.L., Johnson R.L., Stevens R.K., Courtney W.J. and Safriet D.W.
(1984) The quartz content and elemental composition of aerosols from
selected sites of the EPA inhalable particulate network. Atmospheric
Environment 18, In press.
42. Suggs J.C., Rodes C.E., Evans E.G. and Baumgardner R.E. (1981) Inhalable
particulate network annual report; operation and data summary (mass
concentration only). April 1979-June 1980, EPA-600-4-81-037 (available
from NTIS, Springfield, VA 22161]).
43. Lewis W.M. and Grant M.C. (1980) Acid precipitation in the Western United
States. Science 207. 176-177.
44. Huebert B.J.. and Lazrus A.L. (1980) Tropospheric gas-phase and
particulate nitrate measurements. J. Geophys. Res. Section C85, No.
C-12, 7322-7328.
45. Spicer C.W. (1977) Photochemical atmospheric pollutants derived from
nitrogen oxides. Atmospheric Environment 11, 1089-1095.
46. Kelly T.J., Steadman, D.H., Ritter, J.A. and Harvey R.B. (1980)
Measurements of oxides of nitrogen and nitric acid in clean air.
J. Geophys. Res. Section C, 85, No. C. 7117-7125.
47. Pierson W.R., Butler J.W. and Trayser D.A. (1974) Nitrate and nitric acid
emission from catalyst equipped automotive systems. Envir Lett. 7,
267-273.
48. Appel B.R., Tokiwa Y. and Haik M. (1980) Sampling of nitrates in ambient
air. Atmospheric Environment 15, 283-289.
49. Spicer C.W. and Schumacher P.M. (1977) Interferences in sampling
atmospheric particulate nitrate. Atmospheric Environment 11, 873-876.
50. Stelson A.W. , Friedlander S.K. and Seinfeld J.H. (1979) A note on the
equilibrium relationship between ammonia and nitric acid and particulate
ammonium nitrate. Atmospheric Environment 13, 369-371.
51. Tang I.N. (1980) On the equilibrium partial pressure of nitric *cid and
ammonia in the atmosphere. Atmospheric Environment 14, 819-828.
52. Chang D.P.Y. and Hill R.C. (1980) Retardation of aqueo-'s droplet
evaporation by air pollutants. Atmospheric Environment, 803-807.
53. Shaw R.W. , Jr., Stevens R.K. , Bowermaster J., Tesch J.W. and Tew E.
(1982) Measurements of atmospheric nitrate and nitric acid: the denuder
difference experiment. Atmospheric Environment 16, 845-853.
54. Possanzini M. , Febo A. and Liberti A. (1983) New design of high
performance denuder for the sampling of atmospheric pollutants.
Atmospheric Environment 17, 2605-2610.
56
-------
55. Sjodin A. and, Perm M. (1985) Measurements of nitrous acid in an urban
area. Atmospheric Environment 19, 985-992.
57
-------
-------
------- |