Proceedings: Indoor Air 2002
COMPARISON OF METHODS FOR MEASURING
CONCENTRATIONS OF SEMI VOLATILE PARTICULATE MATTER
JJ letter1*, Z Guo1, JA McBrian2 and MR Flynn3
'indoor Environment Management Branch, Air Pollution Prevention and Control Division,
National Risk Management Research Laboratory, Office of Research and Development,
U.S. Environmental Protection Agency, Research Triangle Park, NC, USA
2ARCADIS Geraghty & Miller, Inc., Research Triangle Park, NC, USA
31 department of Environmental Sciences and Engineering, University of North Carolina,
Chapel Hill, NC, USA
ABSTRACT
The purpose of this study was to compare methods for measuring concentrations of
semivolatile particulate matter (PM) from indoor-environment, small, combustion sources.
Particle concentration measurements were compared for methods using filters and a small
electrostatic precipitator (ESP). Particle size distributions were measured using an electrical
low-pressure impactor (ELPI). Semivolatile PM was produced from cigarette and incense
smoke, and, for comparison, a relatively nonvolatile aerosol was produced from
polyalphaolefin oil. Results showed that semivolatile PM concentrations measured using the
ESP method were significantly higher than those measured using the filter method. The ESP
method was shown to be a more effective means of measuring semivolatile PM under the test
conditions.
INDEX TERMS
Particulate matter, Measurement methods, Semivolatile particles, Cigarette smoke, Incense
smoke, '
INTRODUCTION
Concentrations of PM in ambient and indoor environments are typically measured by
gravimetric analysis of particles collected on filters. This method is suitable for measuring
concentrations of nonvolatile PM, but may not be accurate for measuring concentrations of
semivolatile PM. Semivolatile compounds have vapor pressures between 10"11 and I0"4 atm
over the ambient temperature range, and these compounds may exist in both gas and particle
phases in air. Semivolatile particles trapped in a filter matrix present a large surface area to
the flowing airstream, and mass transfer between phases can occur. If gas-phase semivolatile
concentrations are below equilibrium levels, mass may be transferred from the liquid particles
collected on a filter to the gas phase, and concentrations may be underestimated (Volckens et
al. 1999). If gas-phase semivolatile concentrations are above equilibrium levels, gas-phase
semivolatiles may adsorb to the filter surface, and concentrations may be overestimated
(McDow and Huntzicker 1990).
Methods for measuring concentrations of semivolatile PM have been developed that use
denuders and filters, but these methods are cumbersome and are also subject to error from
several sources (Volckens et al. 2000). An easier method of collecting semivolatile particles
has been developed using a small ESP (Leith et al. 1996). As sampled air passes through the
* Contact author email: ietter.jim@epa.gov
132
-------
Proceedings: Indoor Air 2002
tubular ESP, particles are charged by ions generated by a corona with positive polarity, and
charged particles are deposited on an aluminum foil substrate lining the wall of the ESP. The
mass of particles collected on the substrate is determined by gravimetric analysis. Mass
transfer between phases is minimized, because particles collected on the substrate present a
smaller surface area to the flowing airstream, compared to particles collected on a filter, and
the air velocity in the boundary layer near the substrate is less than the air velocity through a
filter (Volckens et al. 2000).
In a previous study, PM emission rates for burning incense were determined using a filter
method, an ESP method, and an ELPI method (Jetter et al. 2002). Emission rates determined
by the ESP method were consistently higher than the other methods, most likely because the
ESP captured and retained more semivolatile particle mass than the filters or the cascade
impactor in the ELPI.
The objective of this study was to compare methods for measuring concentrations of
semivolatile PM from indoor-environment, small, combustion sources. Particle concentration
measurements were compared for the filter and ESP methods. Size distributions were
determined using an ELPI. Semivolatile PM was generated from cigarette and incense
smoke, and, for comparison, a relatively nonvolatile aerosol was generated from
polyalphaolefin oil.
METHODS
A test system specifically designed for measuring PM emissions from indoor-environment
sources was used for the experiments. Air flow through the system was automatically
controlled with a proportional/integral/derivative (PID) controller, a variable-frequency drive
that supplied electrical power to a blower, and an air-flow measuring station that provided
feedback to the controller. Air from the blower passed through a plenum, prefilters, and ultra-
low penetration air (ULPA) filters to a test chamber where aerosols were generated. Air
flowed through the chamber, through a converging section, through a 180° bend, and then
through a long, straight length of 25.4-cm (10-in.) diameter duct where isokinetic sampling
probes were located. Air was exhausted through a stack to outside of the laboratory space. A
detailed description of the test system, with a diagram and particle loss calculations, was
included in a previous publication (Jetter et al. 2002).
For all tests, the target air-volume flow rate through the test system was 8495 L/min (300
cfm). The air flow rate was measured for each test, and the average flow rate for all tests,
corrected to standard conditions, was 8463 L/min (298.9 cfm) with a standard deviation of 76
L/min (2.7 cfm). At this flow rate, the air velocity in the test section where the aerosol was
generated was approximately 6 cm/s (12 fpm), an air velocity typically found in indoor
environments with air circulation. The air velocity in the 25.4-cm (10-in.) nominal diameter
duct was 253 cm/s (499 fpm), with turbulent flow, indicated by a Reynolds number of
approximately 45000, for mixing of the air stream. A traverse of the cross-section of the duct
at the location where the sampling ports were located with the sampling probe for the ELPI
showed that particles were well-mixed in the air stream.
For each test, a sufficient amount of aerosol was generated to provide greater than 600 :g of
particle mass collected on each filter and ESP substrate for accurate measurement. Cigarette
and incense samples were placed on a remotely operated turntable, and a remotely operated
lighter was used to ignite the samples to minimize particle generation not associated with the
burning samples. Before the samples were ignited, the test chamber door was sealed, and
133
-------
Proceedings: Indoor Air 2002
clean air was supplied to the test chamber for at least 15 minutes to reduce background
concentrations of particles. Data were collected with the ELPI before each test to ensure that
the background concentration of particles was less than 1 :g/m3 and 1000 particles/cm3. After
each incense sample was ignited, a remotely operated air jet was used to blow out the flame
so the samples smouldered as they would in normal use. During the lighting and burning of
the cigarettes and incense, the samples were observed through a window in the test chamber.
For comparison, a relatively nonvolatile aerosol of Emery 3004 polyalphaolefin oil was
generated using a TSI Model 9302 atomizer.
Two University Research Glassware (URG) cyclone/filter assemblies were used to collect PM
less than 2.5 :m in aerodynamic diameter (PM2.5) and PM less than 10 :m in aerodynamic
diameter (PM10). Critical orifices were used to maintain the sampling flow rate for the
cyclone/filter assemblies, and dry gas meters were used to measure the flows. A Dekati ELPI
Model 97-2E real-time particle size spectrometer was used to measure particle size
distributions. A data acquisition system was used to record output signals from the ELPI. An
Aerosol Associates ESP was used to collect particles on aluminum foil substrates. A URG
cyclone excluded particles larger than approximately 10 :m from the ESP, operating at a
sampling flow rate of 3.0 L/min. A mass flow controller was used to control the sampling
flow rate of the ESP. Filters and ESP substrates were weighed with a Sartorius MC-5
microbalance before and after experiments to determine the masses of particles collected.
Teflon® filters for the cyclone/filter assemblies and aluminum foil substrates for the ESP were
equilibratecTat 21.5 ± 0.3°C and 34 ±2% relative humidity for 24 hours in an environmental
chamber, and tare weights were obtained before tests were performed. Following each test,
filters and substrates were recovered and stored at a temperature of -40°C to reduce the loss of
semivolatile particle mass. Filters and substrates were equilibrated for 24 hours at the same
conditions before they were weighed again to determine the mass of PM collected.
As aerosol was generated, sampling pumps for the two cyclone/filter assemblies and the ESP
were simultaneously activated, and sampling pumps were simultaneously deactivated at the
end of the test period. The run time of the sampling pumps was measured and recorded. Data
from the ELPI were recorded as 60-second averages before and during each test. Average
PM concentrations were determined by dividing the measured mass of particles collected by
the measured air sampling volume for each device. For each type of aerosol that was
generated, five test replications were performed. Mean values and standard deviations were
calculated for PM concentrations from the five replicates. For each replicate, the relative
difference, D, between the PM10 concentration obtained by the ESP method, Ce, and the PM10
concentration obtained by the filter method, Cf, was calculated as: D = Ce-Cf / [(Ce+Cf)/2],
Statistical analysis was used to verify that differences in concentration measurements were
due to the test methods rather than random error.
RESULTS
Table 1 shows measured PM concentrations in mg/m3 for the three aerosol types and for the
two test methods. Values are shown for the five replicates and for means and standard
deviations of the replicates. Concentrations are shown for PM2.5 and PM10 measured with the
filters. For the cigarette and incense smoke, PM2.5 and PM10 concentrations measured with
the filters were similar, because the smoke consisted of particles less than 2.5 (im in
aerodynamic diameter. However, for the polyalphaolefin aerosol, PM2.5 and PM10
concentrations measured with the filters were not similar, because the aerosol contained some
particles larger than 2.5 p.m.
134
-------
Proceedings: Indoor Air 2002
Table 2 shows relative differences between filter and ESP measurements of PMio
concentrations for the three aerosol types. Values are shown for the five replicates and for
means and standard deviations of the replicates.
Figure 1 shows the filter versus ESP measurements of PMio concentrations for the three
aerosol types.
Table 1. Measured PM concentrations (mg/m3)
Aerosol Type
Test
Method
Replicate
Mean ± Std. Dev.
1
2
3
4
5
Polyalphaolefin
aerosol
(nonvolatile)
Filter, PM2.s
6.8
6.7
6.6
6.6
6.6
6.6 + 0.1
Filter, PMio
9.3
9.8
10.5
10.7
10.9
10.2±0.7
ESP, PMio
j
00 :
00 !
9.5
9.1
9.4
9.3
9.2 ±0.3
Cigarette
smoke
(semivolatile)
Filter, PM2.5
12.1
10.1
10.3
11.0
9.6
10.6 ± 1.0
Filter, PMio
11.4
10.0
10.3
11.0
9.7
10.5 ±0.7
ESP, PMio
15.2
12.1
13.0
14.0
12.5
13.3 ± 1.2
Incense
smoke
(semivolatile)
Filter, PM2.5
2.3
2.8
3.2
3.3
2.8
2.9 + 0.4
Filter, PMio
2.3
2.8
3.2
3.4
2.8
2.9 ±0.4
ESP, PM,0
2.6
3.1
3.6
3.7
3.0
3.2 ±0.4
Table 2. Relative differences between filter and ESP measurements of PMio
Aerosol Type
Replicate
Mean ± Std. Dev.
1
2
3
4
5
Polyalphaolefin aerosol
-0.05
-0.03
-0.14
-0.14
-0.16
-0.10 ±0.06
Cigarette smoke
0.28
0.19
0.23
0.23
0.25
0.24 ± 0.03
Incense smoke
0.12
0.10
0.11
0.10
0.07
0.10 ±0.02
16.0
—. 14'°
CO*
E
12.0
c
•| 10.0
rc
l.
§ 8.0
C
o
o
i-~ 6.0
il
4.0
2.0
I
i
!
¦
1
i
*~
-
1
, ¦
~
•1
¦
~ Polyalphaolefin
¦ Cigarette smoke!
a Incense smoke j
Equality j
2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
ESP, concentration (mg/m3)
Figure 1. Filter versus ESP measurements of PMio concentrations
135
-------
Proceedings: Indoor Air 2002
Figure 2 shows the PM2.5 size distributions by mass measured with the ELPI instrument for
the three aerosol types. Mean values are shown with error bars representing plus or minus
one standard deviation for the five values from replicated tests. The ELPI measurements, as
well as the filter measurements, indicated that the polyalphaolefin aerosol contained particles
larger than 2.5 (j.m in aerodynamic diameter, but the cigarette and incense smoke consisted of
particles less than 2.5 (J.m. Size distributions were more consistent between replicates for the
polyalphaolefin aerosol than for the cigarette and incense smoke. Size distributions for the
incense smoke were consistent with those measured for incense smoke in a previous study
(Jetter et al. 2002).
£
O
o
re
re
5
~ Polyalphaolefin
H Cigarette smoke
~ Incense smoke
0.0 -U
& O ^
Cv Cv Cv o- <->' Of
-------
Proceedings: Indoor Air 2002
Statistical analysis was performed on the PMio data in Table 1. Differences in measured
PMio concentrations between the two test methods were analyzed with t-tests for paired data.
Differences were calculated for each replicate by subtracting the PMio concentration obtained
by the filter method from the PMio concentration obtained by the ESP method. For the
polyalphaolefin aerosol, the mean difference was -1.0 mg/m with a 95% confidence interval
from -1.8 to -0.2 mg/m3. For the cigarette smoke, the mean difference was 2.9 mg/m3 with a
95% confidence interval from 2.1 to 3.6 mg/m3. For the incense smoke, the mean difference
was 0.3 mg/m3 with a 95% confidence interval from 0.2 to 0.4 mg/m3. Measurements with
the ESP and filter methods were significantly different with observed significance levels (p-
values) of less than 0.025 for the polyalphaolefin aerosol and less than 0.001 for both the
cigarette and incense smoke.
CONCLUSIONS AND IMPLICATIONS
Results showed that semivolatile PM concentrations measured using the ESP method were
significantly higher than those using the filter method. Although all methods for measuring
concentrations have associated measurement error, the ESP method was substantially more
effective than the filter method for capturing semivolatile PM in cigarette and incense smoke.
When PM concentrations are lower than those measured in this study, the ESP method may
require a longer sampling time to collect sufficient PM mass for accurate gravimetric analysis.
A possible drawback of the ESP method is that the corona produces ions that can react with
the sampled air and form a metal oxide on the substrate surface over a long sampling period.
However, gravimetric methods can correct for this phenomenon. To account for the metal
oxide formation, a substrate with deposited PM is weighed, the PM is then removed from the
substrate with solvents, and the clean substrate is weighed again. The difference in the two
measurements provides an accurate indication of the collected mass, since the metal oxide is
not removed from the substrate by the solvents. Another possible drawback of the ESP
method is that the corona produces a small amount of ozone that can react with some
compounds in sampled air. However, this does not appear to be a problem for PM
measurements, since the particles are deposited on the substrate before coming into contact
with the ozone plume inside the device.
Results from this study and from other studies have shown that the ESP method is an effective
means of measuring PM in certain applications in the indoor environment, and the ESP
method shows promise for other applications, such as in the ambient environment. Further
testing and development of ESP methods are underway at U.S. EPA research laboratories and
at the University of North Carolina at Chapel Hill.
REFERENCES
Jetter JJ, Guo Z, McBrian JA, et al. 2002. Characterization of emissions from burning
incense. The Science of the Total Environment. Accepted for publication.
Leith D, Leith F and Boundy M. 1996. Laboratory measurements of oil mist concentrations
using filters and an electrostatic precipitator. AIHA Journal. Vol. 57, pp. 1137-1141.
McDow BJ and Huntzicker JJ. 1990. Vapor adsorption artifact in the sampling of organic
aerosol: Face velocity effects. Atmospheric Environment. Vol. 24A (10), pp. 2563-2571.
Volckens J, Boundy M, Leith D, et al. 1999. Oil mist concentration measurements: A
comparison of sampling measurements. AIHA Journal. Vol. 60, pp. 684-689.
Volckens J, Tolocka M, Leith D, et al. 2000. Design and development of a semivolatile
aerosol sampler. University of North Carolina at Chapel Hill. Website:
http://www.sph.unc.edu/baitylab/iv/ivproposal.html, last updated 20-09-2000.
137
-------
w™ r, rrc TECHNICAL REPORT DATA
NHMRL—Klr-r-DJ J (Please read Instructions on the reverse before completing)
—
l. REPORT NO. 2.
1PA/600/A-02/081
3. R6CI
4. TITLE ANO SUBTITLE
Comparison of Methods for Measuring Concentrations of
Semivolatile Particulate Matter
S. REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. authors JiJetterj z>Guo (epa)« J.McBrian (ARCADIS);
and M.Flynn (UNC)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME ANO AOORESS
ARCADIS Geraghty & Miller, Inc., RTP, NC.
Univ. of North Carolina, Chapel Hill, NC.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12, SPONSORING AGENCY NAME AND AOORESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Published paper; 10/1/01-Pres.
14. SPONSORING AGENCY COOE
EPA/600/13
15. supplementary notes project officer is J.J. Jetter, Mail Drop 54, 919/541-4830.
For presentation at Indoor Air '02, Monterey, CA, 6/30-7/2-02.
16. abstract paper gives results of a comparison of methods for measuring concentra-
tions of semivolatile particulate matter (FM) from indoor-environment, small, combus-
tion sources. Particle concentration measurements were compared for methods using
filters and a small electrostatic precipitator (ESP). Particle size distributions
were measured using an electrical low-pressure impactor (ELPI). Semivolatile PM pro-
duced from cigarette and incense smoke was compared with a relatively nonvolatile
aerosol produced from polyalphaolefin oil. Results showed that semivolatile HI con-
centrations measured using the ESP method were significantly higher than those mea-
sured using the filter method. The ESP method was shown to be a more effective means
of measuring semivolatile PM under the test conditions.
17. KEY WORDS ANO DOCUMENT ANALYSIS
a. DESCRIPTORS
b.lOENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Air Pollution
Tobacco
Particles
Aerosols
Smoke
Measurement
Stationary Sources
Particulate
Cigarettes
Incense
13B
06C,02D
14G
07D
21B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
6
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
CPA Form 2220-1 (9-73)
------- |