Session" AB-lc: Continuous Measurement Techniques for PM Components
Laboratory and Field Evaluation of Instrumentation for the Semi-Continuous
Determination of Particulate Nitrate (and Other Water Soluble Particulate Components)
Russell W. Long and William A. McCIenny
U.S. EPA, National Exposure Research Laboratory
Research Triangle Park, NC 27711
Studies conducted at the EPA facility in Research Triangle Parle, NC and at a field study
in Southern California have demonstrated the capability for the semi-continuous determination of
particulate nitrate (and other water soluble ionic species). Two instruments, a R&P 8400N
particulate nitrate monitor and an ion chromatography (IC)-based prototype monitor developed at
Texas Tech University (TTU), were evaluated both in the laboratory using a simulated ambient
aerosol and in the field (Rubidoux, CA) during a three week joint ambient comparison study with
Brigham Young University (BYU) and the South Coast Air Quality Monitoring District
(SCAQMD). During the initial laboratory studies, both instruments were responsive to changes
in the simulated aerosol concentration. However, potential problems were discovered involving
both instruments during the laboratory based studies and these problems are currently being
addressed. Both instruments were then transported to the SCAQMD Rubidoux field site and
operated for a period of three weeks (July 1- 21, 2003). Due to manufacturer's quality assurance
issues associated with IC components of the TTU prototype instrument, limited data were
obtained from this instrument during the three week sampling period. Initial comparisons show
general agreement between the R&P and IC-based prototype instruments for the semi-continuous
determination of ambient particulate nitrate at lower nitrate concentrations (<15 jig/m3) and an
under determination by the R&P instrument at higher concentrations (>15 {ig/m3). During the
three week study period, 15-minute average particulate nitrate concentrations approaching 30
jig/m3 were observed. Semi-continuous results obtained from the EPA-operated instruments
were averaged and compared to integrated sampler results obtained by BYU and SCAQMD.
Results of the laboratory and field studies will be addressed.
Disclaimer.- This is an abstract of a proposed presentation and does not necessarily reflect the
United States Environmental Protection Agency (EPA) policy. The actual presentation has not
been peer reviewed by EPA. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.

-------
Laboratory and Field Evaluation of Instrumentation for the
Semi-Continuous Determination of Particulate Nitrate (and
Other Water-Soluble Particulate Conipoiients)
Paper Number: 587
Mussel! W, Long and William A. McClenny
U.S. EPA, National Exposure Research Laboratory, Research Triangle Park, N€ 27711
ABSTRACT
Studies conducted at the EPA facility in Research Triangle Park, NC and at a field study in
Southern California have demonstrated the capability for the semi-continuous determination of
particulate nitrate (and other water-soluble ionic species). Two instruments, a R&P 8400N
particulate nitrate monitor and an ion chromatography (lC)-based prototype monitor developed at
Texas Tech University (TTU). were evaluated both, in the laboratory using aqueous standards and
a simulated ambient aerosol and in the field (Rubidoux, € A) during a three week joint ambient
comparison study withBrigham Young University (BYU) and the South Coast Air Quality
Monitoring District (SCAQMD). During the initial laboratory studies, both instruments were
responsive to changes in the simulated aerosol concentration. However, potential problems were
discovered involving both instruments during the laboratory based studies and these problems are -
currently being addressed Both instruments were then transported to the SCAQMD Rubidoux
field site near Riverside, CA and operated for a period of three weeks (My 1- 21.2003). Due to
malfunctioning IC components (concentrator columns) of the TTU prototype monitor, limited
data were obtained from this instrument during the three week sampling period. Initial ambient
comparisons show general agreement between the R&P and IC-based prototype instrument? for
the semi-continuous determination of ambient particulate nitrate at lower nitrate concentrations
(<15 jig/m3) and an under determination by the R&P instrument at higher concentrations (>15
jig/m3). During the three week study period, 15-mmute average particulate nitrate
concentrations approaching 30 (lg/m3 were observed. Semi-continuous results obtained from the
EPA-operated instruments were averaged and compared to integrated sampler results obtained by
BYU and SCAQMD,
INTMOBU CTION
Human health endpoints associated with exposure to airborne particulate matter (PM)
include increased mortality and morbidity from respiratory and cardiopulmonary disease.1"3 The
observed exacerbation of health problems is believed to be associated more closely with exposure
to fine particles (PM2.s) than coarse particles. As a result in 1997, the U.S. Environmental
Protection Agency (U.S. EPA) promulgated revised standards for PM, which establishes new
annual and 24-hour fine particulate standards with PM2.5 measured according to the Federal
Reference Method (PM2,j FRM) as the indicator.4 However, ambient fine particulate matter is
not a single pollutant, but a mixture of many chemical species, dominated by primary and
1

-------
secondary aerosols from combustion emissions. Major components include: sulfate, nitrate,
ammonium, and hydrogen ions; trace elements (including toxic'and transition metals); organic
material: and elemental carbon (EC), Stable species such as sulfate can-be accurately measured by
single filter samplers such as the PM2 5 FRM,4"5 Semi-volatile fine particulate species such as
ammonium nitrate are not accurately determined by these techniques.6'"5 As stated previously, the
current national standard for PM2.5 is monitored using the PM2-5 FRM, This method provides a
24-hour averaged PM2.5 concentration. With the FRM, losses of semi-volatile ammonium nitrate
from the particles can occur during sampling and equilibrationlW2 according to the following
equation;
NH4 NO, (p)	(g) + HNO, (g)
Filter samples collected over longer time periods, lack sufficient temporal resolution to track
short-term diumal events (i.e., impact of traffic and photochemistry events on particulate nitrate
concentration).13 In addition, sampled filters require off-line analysis techniques which result in
data availability delays. Therefore, it is desirable to obtain artifact free, PM chemical composition
data with high temporal resolution. Short-term diurnal characterization of mq'or PM2.5
components will provide previously unavailable information about changes in concentration, air
mass movement and source contributions. This information benefits areas such as epidemiological
and environmental studies, model development and evaluation, visibility degradation and climate
change.13"14
This paper demonstrates the capability for the semi-continuous determination of
particulate nitrate. Two instruments, a R&P 8400N particulate nitrate monitor and an ion
chromatography (IC)-based prototype monitor developed at Texas Tech University (ITU), were
evaluated These evaluations were,performed in the laboratory using standard injections and a
simulated ambient aerosol, and also during a three week ambient comparison study in Rubidoux,'
CA. ,	' .
EXPERIMENTAL METHODS
Sampling Sites
Laboratory Studies. Laboratory based evaluation of the R&P 8400N and the TTU prototype
monitors was conducted at the new EPA campus located in Research Triangle Park, NO.
Summer 2003 Field Study. During the Summer 2003 field study both the R&P 840GN and the
TTU prototype instruments were' operated at the SC AQMD Rubidoux sampling site in Southern
California. Rubidoux is a residential community on the east side of the L. A. Basin (near
Riverside). In addition to local sources, (mobile sources, stock yards, etc.). the Rubidoux site is
frequently impacted by pollution transported from the L. A*, metropolitan area by the prevailing
wind (west to east) patterns associated with this area PM2_5 composition at Rubidoux during
summer months is expected to be dominated by organic material and ammonium nitrate. '5~'G
Average (24-hr) particulate nitrate concentrations observed in the Rubidoux/Rlverside area are
documented to be the highest in the State of California during summer months (>5 |ig/m3).15-16
Thus the reason for choosing this site for study. Short term (10 min) nitrate concentrations have
been reported to be in excess of 30 [ig/m3.17	.
2

-------
Sampling Periods
Laboratory Studies. Initial laboratory evaluation of both the R&P 840QN and the TTU prototype
instrument began in Februaiy 2003. Upon completion of the Rubidoux. C A field study, both
instruments were returned to the EPA facility in Research Triangle Park. NC where further
laboratory-based evaluation is on-going.
Rubidoux, CA Field Study. Both instruments were transported to and operated at the SCAQMD
Rubidoux sampling site during My 1- 21, 2003.
Sampling Methods
R&P Series 8400NParticulate Nitrate Monitor. The R&P Series 8400N Ambient Particulate
Nitrate Monitor is composed of two components: (1) pulse generator and (2) pulse analyzer.
Ambient air samples are pulled into the pulse generator through a sharp cut cyclone (SCC),
operated at 5 L/min, to remove particles with diameters greater than 2.5 |xm Following the 2.5
fim cut, the flow is split into a 4 L/min bypass and a 1 L/min sample flow. The 1 L/min sample
flow passes through an activated charcoal honeycomb denuder to remove potential gaseous
interferences (ie.. HN03). The sampled particulate nitrate is collected by critical impaction onto
a Ni-chrome impaction/flash strip mounted in a collection and vaporization cell. Prior to passing
into the collection and vaporization cell, the particle containing sample stream is passed through a
Nation™ humidifier. The' humidifier increases the sample relative humidity (RH) above the
deliquescence point18 of the nitrate salt (i.e., NH4NO3, NaN03 and KNO3) which causes the
particles to grow thus achieving a higher collection (impaction) efficiency. In addition to
increasing the collection efficiency, humidifying the sampled air stream also results in decreased
evaporative losses of semi-volatile nitrate (i.e., aIrimoniumDitrate).,7 At the end ofthe sample
collection phase, the monitor diverts the sample flow from the collection and vaporization cell
while maintaining flow through the sample line, denuder and humidifier, and purges the cell with
nitrogen (N2) gas. The nitrogen flows through the cell and into a nitrogen oxide NO* analyzer
. (pulse analyzer). The impaction strip is then flash heated by current from a battery until reaching
an infrared cutoff (-350 °C). Typical heating times are 70-90 milliseconds (ms). The
vaporization/decomposition process converts the particulate nitrate contained in the sample to
NOx (a combination of NO and N02). The evolved NOx is transported by the nitrogen carrier
gas into the pulse analyzer, where the NO; is subsequently reduced to NO by a heated
molybdenum converter, and detected (along with the initial NO present) by ehemihxminesGence.
The pulse analyzer'output is integrated to yield the nitrate concentration Additionally, the
analyzer baseline is read prior to each anafysis flash and subsequently subtracted from the
integrated result, to yield the final, corrected pulse. At the end of the analysis period the system
returns to sample collection. The pulse generator components (cyclone, denuder, humidifier and
collection/vaporization cell) are housed in an enclosure which is ventilated with outside air to
maintain sampling temperatures close to ambient For the purpose of this paper the sampling
(impaction) period was 13.5 min followed by a 1.5 min analysis period. This results in anew
particulate nitrate measurement in Jig/m3 every 15 min. The above mentioned system is based on
integrated collection and vaporization cell technology developed by. Stolzeriburg and Bering.17
Flow and span audits may be programmed into the monitors sampling program or
performed manually. Analyzer flow audits are done during the sample collection step, without
cycle interruption. The results of the flow audit are used to set the N2 carrier gas flow during the
3

-------
analysis phase of the 15 min cycle. Prior to sampling, the pulse analyzer is calibrated with. 5.0
ppmNG inN2 (02 free) span and N2 (99.995%) zero gases. Daily span audits are used to •
monitor the calibration. Span audits were performed daily during the Rubidoux field study
beginning at 12:00 AM each day. For the laboratory-based studies, span audits were performed
at the beginning of each laboratory session. In addition, the system (pulse generator + pulse
analyzer) is calibrated manually using aqueous standards applied directly to the collection
substrate. The collection surface is then manually flashed and the instrument response recorded.
Control standard injections were used to monitor the aqueous calibration.
TTU IC-Based Soluble Particulate Component Monitor (prototype). The TTU prototype
instrument consists of two main components: (1) particle collector with, parallel plate denuder and
.peristaltic pump and (2) ion chromatograph (including concentrator columns, column switching
valves, analytical columns and conductivity detector) for sample analysis. Ambient samples are
pulled through a cyclone operated 'at 1,0 L/minute to remove particles above 2.5. pm After the
cyclone inlet, the 10 L/minute flow is spit with 5 L/minute going to waste. The remaining 5
L/minute sample flow passes through a wet-wall parallel plate denuder to remove potential
gaseous interferences (positive artifacts). Ambient particles pass through the wet-wall' parallel
plate denuder and into a cylindrical charter through an orifice. A hydrophobic Teflon filter
(Fiuoropore. Millipore Corp., Billcrica, MA) is placed at the chamber exit to prevent particles
from escaping the chamber. At the chamber entrance orifice, the sampled air comes in contact,
with a stainless steel capillary tube through which de-ionized (DI) water is pumped. Acceleration
of the sampled air through the orifice aspirates the water into a fine mist and places the sampled
PM in a water matrix. Any nitrate (or other water soluble ion) contained in the particles goes into
solution thus effectively reducing any evaporative losses '(negative artifacts). The sample
containing mist then continues through the chamber, encounters the hydrophobic filter, and
condenses into droplets which then fall to the bottom of the chamber. The chamber bottom is
cone shaped causing the water droplets to coiect at a point An additional stainless steel capillary
is positioned at the cone point and extends through the chamber wall. The collected water is
removed from the chamber bottom by means of the peristaltic pump and directed to the IC for
subsequent analysis. Upon reaching the IC portion of the monitor, the ion containing water is
pumped into a short concentrator column (TAG-ULPl, Dionex Corp., Sunnyvale, CA) filled with
resin. The resin binds the nitrate and concentrates the sample over a period of 15 min At the end
of the 15 minute sampling period, the effluent from the mist chamber is switched to a second,
alternate column of identical design. The original concentrator column is then back-flushed with.
IC eluent (-25 mMNAOH) causing.the nitrate ions to be released to the analytical column for
analysis. Thus, the two concentrators are alternately switched every 15 min for continual
collection and analysis of nitrate. The IC analysis (including concentrator column switching) is
automated under computer control through the use of an IBM Thinkpad™ computer and the
standard Dion® software package (PeakNet™).
Calibration of the TTU prototype instrument is obtained by preparation of nitrate solutions
of known and varying concentration. These solutions are subsequently injected into the sample
loop (known volume) of the IC and analyzed. Using the injection loop volume and the calibration
standard concentrations, the mass (fig) of nitrate injected into "the IC for analysis is calculated.
Upon completion of the calibration standard analysis, the area count response is recorded and
4

-------
plotted versus the corresponding injected nitrate mass in fig. After all calibration standards are
injected and analyzed a calibration curve is plotted using linear regression analysis with
independent variable jxg nitrate. During ambient monitoring, an area count is obtained for each 15
minute cycle. The corresponding \ig reading on the calibration curve can then be found and
recorded. This mass of nitrate collected can then be divided by the total volume sampled (ie.. 5
Ipm * 15 irrin = 75 L = 0.075 ni3) to give a 15 min nitrate concentration in Jig/ffi3. Calibration of
the TTU prototype instrument was monitored by daily control standard injections. The TTU
prototype instrument and its operation are described in Al-Oorr et al. instrumentation based on
similar principles is described by Weber et al.20 and Simon et ciL a'a Due to manufacturer's
quality assurance issues associated with IC components of the TTU prototype instrument limited
data were obtained from this instrument during the three week Rubidoux field sampling period.
However, these issues have been addressed and the prototype instrument has performed without
incidence during post-Rubidoux laboratory evaluation.	' -
Simulated Ambient Aerosol Generation System. A TSI (Shoreview, MN) Model 3941
Supermicrometer Monodisperse Aerosol Generation System was used to generate a simulated
ambient aerosol of uniform size and shape (monodisperse). The complete Model 3941 System
consists of a Model 3450 Vibrating Orifice Aerosol Generator (VO AG), a Model 3054 Aerosol
Neutralizer, and a Model 3074B Filtered Air Supply. AnHPLC solvent delivery system (pump)
was used in conjunction with the TSI Model 3941 to facilitate longer operating periods than were
normally obtainable with the system as configured from the manufacturer. In addition to the size
and shape uniformity, the model 3941 allows the researcher to change and control the chemical
composition and concentration of the generated aerosol Both sodium nitrate and ammonium •
nitrate aerosols were produced by the Model 3941 and used in the laboratory evaluation of the
previously described monitors. A TSI Model 3321 Aerodynamic Particle Sizer Spectrometer
(APS) was used to characterize (particle size and concentration) the laboratory generated aerosol
prior to sampling by the instruments being evaluated. Figure 1 shows a schematic representation
of the overall simulated ambient aerosol generation system.
Figure 1. Schematic representation of the simulated ambient aerosol generation system used in
the laboratory based studies.
Jii
Aerosol Flow
a-
¦M
N
VOAG

I
¦3
m
A
r
Dilution Air In
APS
Pump

TTU
— 8400N
Waste
5

-------
Integrated Fiber Samples. During the Rubidoux field study, independent integrated nitrate filter
samples (3-, or 24-hr) were collected by Brigham Young University using their PC-BOSS
sampler.8'UM2 In addition, 24-hr Integrated nitrate filter samples were collected by the SCAQMD
using a speciation sampler similar to the Met-One SASS. Nitrate in samples collected by both the
PC-BOSS and SCAQMD sandier was determined by IC analysis in each corresponding lab,
RESULTS AND DISCUSSION
Laboratory Studies Results
8400NLaboratory-Based Evaluation. Prior to monitor comparisons, the R&P 84GGN was
evaluated using aqueous nitrate standards injected directly onto the collection/vaporization
surface. Initial aqueous calibration studies involved the "One-drop, two-drop" method in which a
single concentration standard (100 xig/fiL N03) was used and the mass of nitrate deposited on the
collection surface was varied by changing the injection volume. With this calibration method;
both the amount of nitrate and water increase with an increase in injection volume. Typically,
injection volumes ranged from 0.25 -1.5 |iL (25-150 ng N03). This calibration method is
analogous to that recommended by the manufacturer.23 The zero point was obtained by injecting
0.5 pL DI water. Figure.2 shows a comparison of mass deposited vs mass measured using the
'"One-drop, two-drop" method on two separate occasions. .As indicated, a non-linear
Figure 2. Comparison of Mass Deposited vs Mass Measured Using the "One-drop, two-drop"
Method on Two Separate Occasions.
8405 Aqueous Standard Cal]08/14>2003)
* . Raw Avsraga - - ¦ Siope=t —1—Lmssr (Avsrsgs)
150
y= assise* itse?
; ~ s?=fesis»
125
M0-
SO 75 100
Mass Deposited (ng)
150
25
125
8400 Aqueous Standard Cal (05,29,<2003)
• Raw A- Average - " 1 S(ops=1
1®
m ¦

%
£
0
,20
'« m
Mass Deposited (ng)
response was obtained when large injection volumes were used (i.e., >0.75 pL. corresponding to
an ambient nitrate concentration of 5 fig/m3 for a 15-min average). This observation suggests two
hypotheses (1) at high nitrate concentrations the instrument response is non-linear, or (2) the
increase iti the amount of water matrix results in the instrument response becoming non-linear.
To test these hypotheses, a second calibration procedure (constant volume procedure) was used
in which-a constant volume (0.5 or 1.0 p.L) of standard with varying concentrations was deposited
on the collection surface. The concentration range of the standards used in this method was from
50-300 ng/pL N03. With this method, the deposited nitrate mass changes while the water
amount remains relatively constant Figure 3 shows a comparison of mass deposited vs mass

-------
measured using the constant volume procedure. The figure on the left represents a low- water
content calibration procedure in which masses of 0, 50,100. and 150 ng of nitrate were deposited
on the collection surface using 0.5 pL each of D1 water (0 ng), and 100 ng/|iL (50 ng), 200 ng/|LtL
(100 ng), and 300 ng/^iL (150 ng N03 standards, respectively. The figure on the right
Figaro 3. Comparison of Mass Deposited vs Mass Measured Using the Constant ¥olume
Procedure.
84flDi5queous Standard ^low water)
A, Average— "SIopeH • Raw	Linear(fe/erage'
OjQf*-	1	1		r	—r-*	1
0 25 50 75 100 125 150
Deposited Mass (ng)
8400 Aqueous Stanriairt (Mrjfi \valsr)
A Average ¦
"Sfoj>e=1 ~ Raw	LSrear tftsrerajpj
O
0
s
0
s	us
K-i»asfe ~
25 50 75 100 125
Deposited Mass («sl
represents a high-water content calibration) procedure where identical masses were deposited
using 1.0 jxL ,(2x low-water injection volume) injection volumes and standards half as
concentrated as those used in the low-water procedure. As indicated in both plots, a linear'
response is obtained for both the low- and high-water procedures. However, the response
(calibration slope) is decreased when the high-water procedure is used as compared to that of the
low-water. - A factor of two increase in the amount of water injected onto the flash strip while
holding the mass of nitrate injected constant, results in as much as a 35% decrease in the
calibration slope. The linear response of both curves (even at high nitrate loading) and the
decrease in instrument response associated with an increase in deposited water suggests that the
previously observed non-linear response with the manufacture suggested calibration procedure is
due to a matris (water) interference.
Three hypotheses were proposed to explain the effect of the water matrix on the
instrument response during the aqueous standard calibration studies. First, it was hypothesized
that excess water on the collection surface could result in significant amounts of water vapor
being transported to the ehemilummescence detector after'flash vaporization. The evolved water
vapor may then have a quenching effect on the chermtominescence signal measured by the pulse
detector. To test this hypothesis, aNafion dryer (Perma Pure Inc., Toms River, NJ) was placed in
line between, the pulse generator and pulse analyzer. The presence of the drier will effectively
remove any evolved water vapor from the sample stream prior to its entering into the
chemiluminescencc detector. The constant volume calibration procedure was then repeated with
both the low and high water standards. Figure 4 shows the comparison of deposited nitrate mass
vs measured mass for both the low- and high-water procedures. As indicated in the figure, the
presence of the Nafion dryer in-line between the pulse generator and pulse analyzer did not
7

-------
improve the slope degradation with increased water content. Therefore it is assumed that the
decrease in instrument response in the presence of excess water is not due to quenching effects by
water vapor in the pulse analyzer.	• .
Figure 4. Comparison of Deposited Nitrate Mass vs Measured Mass for Both the Low- and
High-Water Procedures with the Nalon Drier In-Line between the Pulse Generator and Pulse
Analyzer.
8400 Aqueous Standard {{owyralei)
A An Mage — «S!oper 1 • Raw	Linear (Average;
Deposited Mass fug)
846S AqusMis Stzntet! flag};watei)
A Average •— -Slcpe=1 » Raw	Lmear Ifimrage,
0 25 50 75 100 125 150
Deposited Mass (ntjJ
The second hypothesis was that the presence of water on collection surface (flash strip)
results in a decreased N03 to NO conversion efficiency. Ideally, the flash vaporization process
would result in complete conversion of the particulate nitrate to NO which is subsequently
detected by the clienaluminescence analyzer. The presence of water on the flash strip will result
in less energy being available during the vaporization process to quantitatively convert the nitrate
to NO. This incomplete vaporization will result in the production of significant amounts of NOa.
Under normal operating conditions, the presence and amount of N02 will not have an effect on
the instnment response.24 However, a reduced molybdenum converter efficiency (<95%) in
converting N02 to NO in the pulse analyzer will result in under-detcrrrination of nitrate when the
flash vaporization step results in a high NOz/NO ratio. The molybdenum converter efficiency was
determined by first spanning the pulse analyzer with aNIST traceable NO calibration gas of
known concentration (5.01 ppm, Scott Specialty Gases, Plumsteadville, PA). A known
concentration of NOz calibration gas was then passed through the pulse analyzer and the
instrument response recorded. The above mentioned procedure resulted in a molybdenum
converter efficiency of less than 50%. The decreased molybdenum converter efficiency may
explain the non-linear response observed during the "One-drop, two-drop" calibration method
(Figure 2). At higher injection volumes (Le., >0.75 (.si.) the excess water present results in a
higher N02/NO ratio where only a fraction of the N02 is converted to NO and detected In
addition, the decrease in slope associated with an increase in water during the constant volume
procedures (Figures 3 and 4) can be attributed to a decreased molybdenum converter efficiency.
A new molybdenum converter (>95% efficiency, see procedure above) was installed in the
instrument and the "One-drop, two-drop" calibration procedure repeated. Figure 5 shows the
8

-------
results oftMs calibration on two separate occasions. As indicated in the figure. linearity of the
calibration curve is improved with the new molybdenum converter installed in 1he'8400N pulse
analyzer. A slight departure from linearity is still observed at high injection volumes. However,
Figure 5. Comparison of Mass Deposited vs Mass Measured Using the "One-drop, two-drop"
Method on Two Separate Occasions after Replacement of Pulse Analyzer Molybdenum
Converter.
8400 Aqueous standard cal (G2SS2KH)
• Raw A Average" " 'Slope=l —— Linear Coverage)
Mas Deposited {ng|
8400 Aqueous Standard Caf (62.273004)
• Raw. a Average * ¦ ¦Sicpr i Linear (ftyersge)
15D
*»
125
I
ICO
s
150
0
Mess Deposit®! ^ng)
the volume required to produce a decrease in response is increased by a fee tor of 2 (i.e.. -150 |iL)
as compared to the volume required (i.e.. -75 P-L) to produce a similar instrument response with
use of the faulty molybdenum converter. The constant volume calibration procedure was
repeated to determine the change in calibration slope with an increase in water on the flash strip
with the new molybdenum converter installed in the pulse analyzer. As indicated in Figures 6A
and 6B. no observable change in slope is observed in going from the low-water to the Mgh-water
procedure. Figures 6C and 6D illustrate an addition constant volume calibration procedure where
both the deposited nitrate and water masses were increased by a factor of 2 by doubling the
injection volumes (Le., 1.0 and 2.0 |il for low- and Mgh-water procedures, respectively). This
procedure resulted in a decrease of the calibration slope by approximately 50 % when excess
water was placed on the flash, strip (see Figures 6C and 6D). The similarity of the calibration
slope in Figure 6C (high-nitrate, medium-water) to those in Figures 6A (low-nitrate, low water)
and 6B (low-nitrate, medium-water) further strengthen the assumption that high nitrate loading is
not responsible for the decreased instrument response. Although improved with addition of a
functioning molybdenum converter, a decrease in instrument response is still observed when'
excess water is placed on the flash surface (see Figures 5 and 6).
The third hypothesis investigated is that under conditions of heavy water loading, some
but not all sampled nitrate is vaporized during the flash process. The residual nitrate remains in
the particulate form and is removed from the sample stream by an in-line Teflon filter prior to
entering the pulse analyzer. This assumption was tested by placing a 25 mm Tefto™ (PaR Corp.
Ann Arbor, MI) filter in-line between the pulse generator and the pulse analyzer to collect any
9

-------
Figure 6. Comparison of Deposited Nitrate Mass vs Measured Mass for Both the Low- and
High-Water Procedures After Replacement of Pulse Analyzer Molybdenum Converter. In Figures
C and D, Both the Deposited Nitrate and Water Mass Were Increased by a Factor of 2 by
Doubling the Injection Volumes,
8400 Aqueous Standard Cal flowwater)
• Raw & Average" " * S!ope=1
•Lhssr i^iversga)
50 73 100
Msbs Deposited (ng)
150
8400 igneous Standard Ca! flewwater)
Raw a Average'
'S!ope=1
-LJnsar Ct.-zersga)
250-
2 2^0
1 150
S 50
0
-yCL74«x-t.9B5;
- f^-asaes
• TOO 150 200
Mass Deposited (ngj
B MOT Aqnsous Standard CM (high, water)
) « Raw Average" * 1 Blopgsl
- Linear (Average)
R^SSSSS.
50 75 .100
Mess Dej>osii«iI (ng)
125
150
D 8400 fiijiisous Standard Cal (Mgii water)
Raw a. Average
:Iops=1
-LM8a(ft.Va'^8)
SCO
¦55?^ ^ .
= 250-
® 2CO
7=&398£x-t£P
i iw
1C0 ISO 200 250
.Mas Deposited fog)
particulate nitrate remaining after the flash vaporization step. The filter can then be extracted and
the deposited nitrate mass determined by IC analysis. Repeated 1.0 flL irgections were made onto
the collection surface using a concentrated (300 ng/p.L fromNaN03) N03 standard solution
After each injection, the collection surface was flashed and the instrument response recorded.
Upon completion of the final injection/flash step, the nitrate collected on the in-line filter was
determined. The sum of the filter collected nitrate mass and the 84O0N measured mass was then
obtained and compared to the total mass deposited on the "collection surface. The results of this
comparison on two separate occasions axe given in Table 1. The presence of nitrate on the in-line
filter (up to 35% of the total mass deposited) suggests that the presence of excess water does
have an effect on the NO3 to NO conversion efficiency and hence the instrument response.
As stated previously, major " components of ambient fine particulate matter include:
sulfate, nitrate, ammonium, and hydrogen ions; trace elements (including toxic and transition
10

-------
Table 1. R&P 840QN Aqueous Calibration Mass Recovery Results/Water Interference).
Date:
June 10,2003
July 07,2003
Deposited Mass (ng)
3600
6000
8400M Measured Mass (ng)
2057.9,
4574
In-line Filter Mass (ng)
1253.5
1182.8
Measured + Filter Mass (ng)
3311.4
5756.8
% Mass Measured by 84D0N
57.2%
76.2%
% Mass of Mass 012 Filter
34.8%
19.7%
% Mass Recovered
(Measured +• Filter Mass)
92%
. 95.9%
metals); organic material; and elemental carbon (EC). It is reasonable to assume that the presence
of these other PM components, 'in addition to water, will also have an effect on the NO3 to NO
conversion efficiency and hence, instrument response. This assumption was investigate by again
placing a 25 mm Teilo filter in-line between the pulse generator and the pulse ansJyzerto collect
any residual particulate nitrate after the flash vaporization step. Repeated 1.0 fiL injections were
made onto the collection surface using a 250 ng/jiL N03 standard solution prepared from
NH4NO3. After each injection, the collection surface was flashed and the instrument response
recorded. Upon completion of the final injection/flash step, the nitrate collected on the in-line
filter was determined by IC analysis. The above process was then repeated using a 250 ng/|iL
NO3 + 250 ng/pL SO4 standard solution prepared fromlMHsNOs and (NELX2 S04, respectively.
The results of this comparison for the two separate standard solutions are given in Table 2. The
increased nitrate mass on the in-line filter for the NH4NO3 + (Mi^SCX-. injections as compared to
fee single component NH4NO3 injections indicate that the presence of other PM components on
the flash strip may result in a decreased instrument response. The low percentage of mass
recovered for both the single- and double-component standards suggests additional loss
mechanisms that warrant further evaluation.
In addition to effecting the slope and shape of the calibration curve (which may result in
significant errors in ambient nitrate deterrainations), water has the potential to interfere with the
actual ambient measurement. The hygroscopic nature of nitrate PM results in significant amounts
of water, being associated with the particles as they exist in the ambient air." In addition, the
sample stream is humidified once inside the pulse generator resulting in even further hydration.
Therefore, it is reasonable to assume that a situation similar to a "high-water'7 aqueous standard
calibration may exist during periods of extremely high RH or high nitrate concentrations resulting
in a decreased NO3 to NO conversion efficiency and therefore an under determination of ambient
nitrate concentrations by the 8400N. The presence of additional PM components may result in a
¦similar decrease in instrument response under ambient conditions.
11

-------
TaMe 2. R&P 8400N Aqueous Calibration Mass Recovery Results (PM Component
Interference).			.	.
Standard:
NH4NO3

Deposited Mass (ng)
8000
8000 •
IVT^^ (irccr^
L i A.* JL 4-1 i XYjL1hL2 w ^ J
3937.9
3026.1
In-line Filter Mass (ng)
690.8
- _ 2595.7
Measured + Filter Mass (ng)
4628.7
5621.8
% Mass Measured by 8400N
49.2%
. 37.8%
% Mass of Mass on Filter
8.6%
32.4%'
% Mass Recovered
(Measured + Filter Mass)
57.8%
70.2%
This decrease in response may be confounded further with multi-component aerosols such as
exist in ambient air.
Comparison of8400N and TTUPrototype Results (SimulatedNitrate Aerosol). The 15-min
average TTU prototype and 8400N monitor nitrate results obtained while sampling a laboratory
generated ammonium nitrate aerosol (1.2 pan average particle diameter) are given in Figure 7,
.Changes 'in aerosol concentration were obtained by adjusting the dilution air flow rate into the
dilution chamber of the particle generation system (see Figure 1), During these laboratory
studies, precautions were taken to provide a pure, dry (RH<20%) aerosol to the instruments for
sampling to eliminate any potential matrix effects. Both monitors show general agreement with
respect to changes in nitrate concentrations. However, the 8400N shows a more rapid response
to increases and subsequent decreases in nitrate consecrations (see Figure 7, open circles). The
slight delay in response associated with the TTU prototype monitor can be attributed to the time
needed for droplet formation and dislocation in the TTU particle collector. In addition, a small
portion of the dislodged droplets are held up on the chamber walls, thus adding to the delay. The
residence time of the droplets on the chamber walls can be reduced considerably by treating the
interior surface of the particle collector with a wetting agent (e.g., Rain-X). Figure 8 shows a
comparison (with regression statistics) of TTU vs 8400N nitrate results obtained while sampling
the laboratory generated ammonium nitrate aerosol. The highlighted data points (white circles,
see Figure 8) represent samples taken immediately following dramatic increases or decreases in
nitrate concentration (see Figure 7, open circles). During controlled laboratory conditions (low
RH, single component mono disperse aerosol), similar results were obtained by both .the TTU
prototype and R&P 8400N particulate nitrate monitors with a regression slope near unity
(1.02x4-0.67. n=43) and a R2 value near 0.95. .
12

-------
Figure 7. TTU Prototype and 840ON Monitor Nitrate Results (15-min average) Obtained While
Sampling a Laboratory Generated Ammonium Nitrate AerosoL Open Circles Represent Time
Periods Immediately FoEowing Dramatic Changes in Nitrate Concentration.
Simulated NH 4NO3 Aerosol (Dec, 2003)
- - - -mi	8400K
50
40
3D
2©
10
0
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43
• Sample No.
Figure S. Comparison of TTU and 8400N Nitrate Results (15-min average) Obtained from
Sampling a Laboratory Generated Ammonium Nitrate Aerosol The Highlighted Data Points
(white circles) are Due to Differences in Instrument Response Time to Changes in Nitrate
Concentration.
December 2003 Laboratory Studies
• N03
Linear (N03)
veiasa
« 40 -
1	—1	
20	30
TTU HO , trcferogramsftn s)
5fi
Rubidoux, CA Field Study Results
13

-------
Comparison of8400Nand TTUPrototype Results. The 15-min average TTU prototype ^
8400N monitor nitrate results obtained during two. 2-day periods (12:00 My 4-12:00 My 6 and
21:00 My 18-21:00 My 20) from the My 3003 study at Rubidoux, CA are given in Figures 9A
and 9B. These figures illustrate changing particulate nitrate concentrations during the respective
figure 9. The TTU Prototype and 8400N Monitor Nitrate Results (15-min average) Obtained
Dining: A) 12:00 July 4-12:00 My 6 andB) 21:00 My 18-21:00 My 20, from the My 2003
Study at Rubidoux, CA
Rubidoux, CA (04-Gg July 2003}
——Rap	TTU I
2®
5=
I «
E
o
12:00 18:00 0:00 6:00 12:08 1S:« 0:00 6:00
Time of Day
B
Rufcldctix, CA (1S-20 July 2003)
£
-TTU
20
15 -¦
• 10
FS
5
fi
Sv«-i
£'
\
*
\J
o
21:00 3:®t 9:00 ' 15:00 21:00 3:00 S:O0 15:0!)
Time of Day
periods oyer the Rubidoux/Riverside area Typically, peaks in particulate nitrate concentrations
14

-------
are observed around noon (12:00) each day. However, a large nitrate peak is observed beginning
in the late evening on July 4 and extending to the early morning hours of My 5 (see Figure 9A).
This peak is attributed to a local fireworks display and resulting brush fire near the'Rubidoiix site.
The data shown in Figure 9 indicate that both instruments'are capable of tacking rapid (Le., 15-
30 min) changes in ambient particulate nitrate concentrations. Figure 10 shows a comparison
(with regression statistics) of'I TU vs 8400N particulate nitrate results obtained during the
Rubidoux; CA field study. As stated previously, limited data were obtained from the the TTU
Figure 1®. Comparison (with regression statistics) of TTU vs 8400N Particulate Nitrate Results
(15-min Average) Obtained During the Rubidoux, CA Field Study.
July 2003 Rubidoux, CA
Linear 15 |ig/m3), the
8400N measures lower than the TTU instrument. Linear regression of TTU vs 8400N particulate
nitrate results for the entire study period gives y=0.79x+0.64 (n=451) with an R4 value of 0.86.
Limiting the data set to those values at or below 15 Jlg/m3 results in a regression slope closer to
unity (y= 0.94x-0.30, n= 403, R2=0.83) as shown in Figure 11. As stated in the previous section,
divergence at high nitrate concentrations was not observed when the TTU prototype and 84G0N
instruments were compared under controlled laboratory conditions (see Figure 7). It is therefore
assumed that the divergence (under-meas urement) of the 8400N at higher particulate nitrate
concentrations is due to a matrix type interference (i.e.. water, other particulate components, etc.)
that exists under ambient conditions but is eliminated Under controlled conditions.'
15

-------
Figure 11. Comparison (with regression statistics) ofTTU vs 8400N Particulate Nitrate Results
(15-min Average) <15 p,g/m3 Obtained During the Rubidoux, CA Field Study. ' • '. '
July 2003 Rubidoux, CA
¦ N03	SIope=1	Linear (N03)
15
___
* j»J
12
9
6
3
0
6
9
12
15
TTU NO 3 {ir.icrograns'm 3y
Comparison of Semi-continuous and Integrated Sampler Results. The results obtained by the
8400N and the TTU prototype for the semi-conlinuo us determination of particulate nitrate were
averaged (3-, or 24-hr) during the study period for comparison with results obtained from the PC-
BOSS and SCAQMD-samplers. The results of this comparison (with regression statistics) are
given in Figure 12. The abscissa (filter nitrate data) in Figure 12 contains both PC-BOSS (3- and
24-hr)-and SCAQMD (24-hr) results. Due to the previously mentioned IC associated issues with
the TTU monitor, very limited data were obtained with this instrument for comparison with the
integrated filter samples (Figure 12, white triangles). Al-llorr el al. , gives a more complete
comparison of PC-BOSS and semi-continuous nitrate results obtained with a newly
commercialized version of the TTU prototype instrument25 Comparison of filter based results
with semi-continuous 8400N results show general agreement at concentrations at or below 10
}lg/m3. Above this concentration, the 8400N measures lower than the filter based methods,
possibly due to the previously described matrix effects.
SUMMARY •
Studies conducted at the EPA facility in Research Triangle Park. NC and at a field study in
Rubidoux. C A have demonstrated the capability for the semi-continuous determination of
particulate nitrate. However, these studies have indicated that the presence of water and other
matrix materials (i.e., other PM components) on the collection surface of the 84Q0N results in a
decreased instrument response during both calibration procedures and ambient determinations.
16

-------
Figure 12, Comparison (with regression Statistics) of Filter (3- and 24-hr average) vs Semi-
continuous Particulate Nitrate Results Obtained During the Rubidoux. CA Field Study.
July 2003 Rubidoux, CA

^LifissrCrFtJ
Jsjai=l~*
25
V = 64215+ 2.2055
f: nVoj^S - 1

10	15	20
Filter Nitrate (micrograms.^)
The decrease in instrument response can result in significant under-detennioaticm of particulate
nitrate at higher concentrations. Comparison of 11IJ prototype and 8400N results obtained while
sampling a laboratory generated nitrate aerosol show good agreement at both low and high nitrate
concentrations. The response time to changes in nitrate concentration for the TTU prototype was
delayed as compared to the 8400N. This delay is associated with the time needed for droplet
formation and dislocation in the particle collection chamber of the TTU instrument. Treating the
interior surface of the particle collection chamber with a wetting agent such as Rain-X was shown
to reduce the delays associated with droplet hang-up on the chamber walls. During the Rubidoux
field study, both instruments were capable of tracking short term changes in particulate nitrate '
concentrations. Data ftomthe July 2003 study period show good agreement between the 8400N
and TTU instruments at lower nitrate concentrations (i.c.. <15 }ig/m3). At higher nitrate
concentrations (i.e., >15 jxg/m3) the 8400N was observed to measures lower than the TTU
instrument. Similarly, comparison of filter based and 84GGN results show good agreement at
nitrate concentrations at or below710 jig/m5 and divergence (under-measuremsit by the 840ON) at
concentrations above this level The under-determination at higher nitrate levels under ambient
conditions may reflect the matrix (water, other PM components) interference observed during the
laboratory studies. Further investigation (laboratory and field) is needed to elucidate and
eliminate the interference associated with the R&P monitor and to identify and address any
problems associated with the IC based instrument (TTU prototype and Dioncx commercial units).
In addition, further comparison studies should be performed with both semi-continuous and filter
17

-------
based methods. These studies axe necessary to validate the semi-continuous methods for the
determination of ambient particulate nitrate.
ACKNOWLEDGMENTS	-
The technical assistance of Rida Al-Horr ofDionex Corp., Sandy Dasgupta of Texas Tech.
University and of Ruppreeh and Patashnick Co., Inc. in lie research is gratefully acknowledged.
Cooperation by Rudy Eden of the South Coast Air Quality Monitoring District and Brett Grover,
Michael Heinman, Norman L. Eatough and Delbert 1 Eatough of Brigham Young University in
providing the integrated filter nitrate data is also gratefully acknowledged.
DISCLAIMER
This paper has been reviewed in accordance with the United States Environmental. Protection
Agency's peer and administrative review policies and approved for presentation and publication.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
REFERENCES
1.	Pope C. A. III. 2001. Epidemiology of fine particulate air pollution and human health;
Biological mechanisms and who's at risk?. Environ. Health Perspect, 108(Sup 4), 713-
.723.
2.	Schwartz. I, Dockeiy. D.W., arid Neas, L.M., 1996. Is daily mortality associated
specifically with fine particles? ./. Air Waste Manage.. 46, 927-939.
3.	U.S. Environmental Protection Agency 1996. Air Quality Criteria for Particulate Matter.
Environmental Protection Agency, Research Triangle Park, NC, EPA/6QQ/P-95/Q0IaF.
4.	Schaefer, G.. Hamilton, W., and Mathai C.V., 1997. Implementing the revisedNAAQS
and the FACA subcommittee for ozone, particulate matter and regional haze. Environ.
Manage., 1997,22-28.
5.	Mustek D.. 2000. A summary of the ambient air program for PM2.s. Environ. Manage
February, 17-20.
6.	Eatough, D. J., Obeidi, F., Pang, Y., Ding, Y., Eatough, N.L., and Wilson W.E., 1999.
Integrated and real-time diffusion denuder samplers for PM2.5 based on BOSS, PC and
TEGM technology. Atmos. Environ., 33,2835-2844.'
7.	Herring, S., Cass, G., 1999, The magnitude of bias in the measurement of PM2.5 arising
from volatilization of particulate nitrate from Teflon filters. J. Air Waste Manage., 49,
725-733. '	'
8.	• Lewtas, J., Booth, D., Pang, Y., Reimer. S., Eatough, D.J., Gundel. L., 2001.
18

-------
Comparison of sampling methods for semi-volatile organic carbon (SVOC) associated
withPM2j. Aerosol Set. Technol, 34, 9-22.
9.	Pang. Y., Eatough, D.J., and Eatough, N.L., 1999. Determination of PM2.5 sulfate and
nitrate with a PC-BOSS designed for routine sampling for semi-volatile particulate matter.
J. Air Waste Manage., 49, 69-75.
10.	Eatough, D.J., Long, R.W., Modey, W.K., Eatough, N.L., 2003. Semi-volatile secondary
organic aerosol in urban atmospheres: meeting a measurement challenge. Amos. Environ.,
37, 1277-1292.
11.	Long, R.W., Smith, R., Smith, S.,. Eatough, N.L. Mangelson, N.F., Eatough, D.J., Pope,
C.A. and Wilson. W.E.. 2002. Sources of fine particulate organic material along the
Wasatch front Energy & Fuels, 282-293.
12.	Long. R.W., Eatough, N.L., Mangelson, N.F., Thompson, W.,Fiet, K., Smith, S., Smith,
R., Eatough, D J., Pope, C.A, Wilson, W.E., 2003. The measurement of PM2.j,
including semi-volatile components, in the EMPACT program; results from the Salt Lake
¦ City study. Aimos. Environ., 37, 4407-4417.
13.	Long, R.W., Modey, W.K., Smith, P.S., Smith, R., Merrill C., Pratt, J., Stubbs, A, "
Eatough, N.L.. Eatough, D.J., Malm, W.C., Wilson, W.E., 2003. One- and three-hour
PM2.5-characterization, speciation and source apportionment using continuous and
integrated samplers. Aerosol Sci. Technol, submitted.
14.. McClenny, W.A, Kronmiller, KG., Oliver, K.I).. Jacumin, Jr.,-H.H., Daughtrey, Jr., E.H.,
2003. Status of semi-continuous monitoring instruments for gas and water soluble particle
components needed for diagnostic testing of ambient air quality simulation models.
EP A/600/R-02/072 (NITS PB2004-100894).
15.	Qbeidi F., Eatough, N.L., and Eatough, D.J., 2002. Continuous measurement of semi-
volatile fine particulate matter at Riverside and Bakersfield, California. Aerosol Sci.
Technol, 36, 204-216. ¦
16.	Pang, Y., Eatough, N.L., and Eatough, D. J., 2002. PM2.j semi-volatfle organic material at
Riverside, California: implications for the PMz 5 Federal Reference Method sampler. .
Aerosol Sci. Technol. 36, 277-288.
17.	Stolzenburg. M.R., and Heririg, S.V., 2000. Method for the automated measurement of
fine particulate nitrate in the atmospphere. Environ. Sci. Technol., 34,907-914.
18.	Lightstone, J. M., Gnascfa, T. B., Imre, D., and Oatis, S., 2000. Deliquescence,
efflorescence, and water activity in ammonium nitrate and mixed ammonium
19

-------
nitrate/succimc acid microparticles, J. Phys. Chem,, 104, 9337-9346.
19.	Al-Horr, R., Samanta, G., Dasgupta, P.K, 2003. A continuous analyzer for soluble
anionic constituents and ammonium in atmospheric particulate matter. Environ. Sci.
Techno!., 37, 5711-5720.
20.	¦ Weber, R.J.. Orsini, D., Daun, Y., Lee, T.-N.. Klotz, P.J., Brechtel, F., 2001. A Particle -
into-liquid collector for rapid measurement of aerosol bulk chemical composition. Aerosol
Set TechnoL, 35, 718-727.
21.	Simon, P.K, Dasgupta. P.K, 1995. Continuous automated measurement of gaseous
nitrous and nitric- acids and particulate nitrite and nitrate. Environ. Sci. TechnoL, 29.
1534-1541.
22.	Simon, P.K, Dasgupta, P.K, 1995. Continuous automated measurement of the soluble
fraction of atmospheric particulate matter. Anal Chem., 67, 71-78.
23.	Hering, S., Kirby, B., 2002. (Draft) Standard operating procedure for routine operation
of the Rupprecht and Patashnick 8400N ambient particulate nitrate monitor.
24.	Rupprecht &'Patashnick Co., Inc., 2003. Operating Manual: Series 8400N Ambient
Particulate Nitrate Monitor, Revision A R&P Part Number 42-007038.
25.	Al-Horr. R_, Later, D.W., Grover. BD. Watson, J., Chow, J., Eatougl^ D.J., 2004. A
new gas-particle ion chromatographic, system for the continuous monitoring of soluble
gases and ionic constituents of particulate matter. Proceedings, A&WMA 97th Annual
¦ Conference & Exhibition. Indianapolis IN, (submitted).
KEYWORDS
Particulate Nitrate
Semi-Continuous Instruments
Ambient Measurements
Semi-Volatile Material
20

-------
TECHNICAL REPORT DATA
1. Report No. 2.
q
4. Title and Subtitle
Laboratory and Field Evaluation of Instrumentation for the Sena-
coolmuous Determination of Particulate Nitrate (and Other Water-Soluble
Particulate Components)
5.	Report Date
March 2004
6.	Performing Organization Code
7. Author(s)
Mussel! W. Long, NERL, MD D2O5-03, RTP ,NC 27711
William A. McClenny, NERL, MD D205-03, RTP ,NC 27711
8. Performing Organization.
Report No.
9.PerfortDing Organization Name and Address
U.S. Environmental Protection Agency
National Exposure research Laboratory
Research Triangle Park, NC 27711
10.	Program Element No.
11.	Contract/Grant No.
12.Sponsoring Agency Name and Address
U.S.- Environmental Protection Agency
National Exposure research Laboratory
Research Triangle Park, NC 27711
13.	Type of Report and Period
Covered
14.	Sponsoring Agency Code
15. Supplementary Notes . '
16. Abstract
(see- attached)
17. KEY WORDS AND DOCUMENT ANALYSIS
A. "Descriptors
semi-continuous, particulate nitrate, water soluble
particulate components
B. -HeMifiefs/'OpefiEai®! -¦
Terms
P ATT

18. Distribution Statement
19. Security Class (This
Report)
21. No. of Pages .
21
20. Security Class (This
Page)
22. Price
Form Available: Network Neighbdrhood\IQ)ight\Groups\HEASDWorms\Technical-Report-Data-222(M

-------