-------
isolating any contributions from the supporting substrate. Rosasco
et al. (25), in recording spectra from single micron-size particles,
have demonstrated that heating effects caused by the high excita-
tion laser power densities associated with extreme focusing of the
probing light beam can be minimized if the particle is in good ther-
mal contact with a substrate, such as sapphire or LiF2. Thus, to
characterize a distribution of particles with a single Raman scan,
the particulate sample should be distributed as a monolayer on a
substrate to maximize the thermal contact so that high field inten-
sities can be employed to increase the Raman signal. An obvious
approach is to use a relatively high power laser focused directly
on the collected sample and a conventional spectrometer which could
probe a sample area having a diameter of 100 Mm or larger. However,
adaptation of the techniques of internal reflection spectroscopy
(30) to enhance the excitation electric field intensity at the
particulate-substrate interface may provide significant advantages.
Carniglia et al. (31) have investigated, both theoretically
and experimentally, the excitation of a monolayer of fluorescent
molecules by evanescent or exponentially decaying waves generated
upon total internal reflection as well as the emission of light by
the excited molecules. They find that absorption is proportional
to the square of the exciting field whether the incident light is
homogeneous or evanescent and that the emission process follows a
reciprocity principle. Lee et al. (32) recently have completed
similar experiments to determine the polarization and angular de-
pendence of fluorescence from a thick liquid dye layer in contact
with the substrate excited by evanescent waves in the configuration
shown in Figure 10. Light entered a hemicylindrical sapphire prism
at an angle of incidence, #INC, which exceeded the critical angle
for the prism-dye solution interface; thus, excitation of the dye
molecules was by an evanescent wave. The emitted fluorescence was
observed with an angular resolution of 0.3° upon its passing through
the prism at an observation angle, 0OBS , as defined in Figure 10.
The observed fluorescence intensity as a function of 0OBS for the
different polarization cases (as defined above) for an angle of in-
cidence of 60° is shown in Figure 11. In good agreement with the
work of Carniglia et al. (31), as well as with Fresnel theory, the
fluorescence in all cases was observed to peak at a specific angle
which is equal to the critical angle had the light of the fluore-
scence wavelength been incident from the prism to the dye solution.
It should be noted that, because of dispersion in the prism refrac-
tive index, the fluorescence does not peak at the critical angle
corresponding to the excitation wavelength. The fluorescence
intensity dependence on #]NCis shown in Figure 12, which indicates
that the largest fluorescence intensity is observed when the exci-
tation and fluorescence radiations are at their respective critical
angles.
122
-------
AIR
SOLUTION
ATTENUATED
REFLECTION
CO
FLUORESCENCE
SCATTERING
EXCITATION
INC
Figure 10. Hemicylindrical sapphire prism with dye solution on top. The
angle of incidence, #u\j(-, exceeds the critical angle for the
prism-dye solution interface, causing the evanescent wave in
the dye solution. The fluorescence scattering is detected as
a function of Q.
OBS
Different combinations of polarization
for incident and scattered radiation can be selected with polarizers.
-------
50
4O
V)
J—
z
3O
J22O
I O
e,NC=
v-v
30° 60"
ANGLE OF OBSERVATION 9
90°
ots
30° 60°
ANGLE OF OBSERVATION 8OBS
90°
Figure 11. The observed fluorescence intensity as a function of
for different polarization combinations. V and H designate
vertical and horizontal polarization directions, respectively,
with regard to the scattering plane shown in Figure 10. GINC
was greater than the critical angle. The fluorescence inten-
sity peaked at the prism-dye solution interface critical angle
for fluorescence wavelength.
124
-------
v-v
CO
ex.
eINC=60°
z
LU
X 20 -
10 -
30° 60° 90C
ANGLE OF OBSERVATION 6OBS
Figure 12. The fluorescence intensity as a function of 0 for
several 0INC . Note that the largest fluorescence
intensity is observed when O^c an<^ ^OBS are at their
respective critical angles. The polarization com-
bination is V-V.
125
-------
This internal reflection configuration applied to observing
Raman scattering from a monolayer of particulates offers several
advantages. First, the electric field amplitude at the interface
can be enhanced by a factor two because of superposition of the
incoming and reflected field amplitudes (30), which causes a factor
4 increase in the excitation intensity. Correspondingly, utiliz-
ing the reciprocity of the emission process, an additional factor
4 enhancement (at maximum) can be attained when the evanescent
fluorescence is converted to a homogeneous wave upon entering the
prism. Second, because of the prism dispersion, the reflected inci-
dent light is spatially separated from the angular peak in the
Raman signal intensity and need not enter the spectrometer. In
addition, localization of the laser amplitude within the monolayer
can be achieved since the penetration depth of the evanescent wave
can be on the order of the monolayer thickness, depending upon the
refractive index ratio of the substrate and particulate, the exci-
tation wavelength, and the angle of incidence. However, isolation
of the desired Raman signal from fluorescence, elastic, and Raman
scattering initiated within the substrate may be difficult to
obtain.
By replacing the liquid dye layer in the configuration of
Figure 10 with a layer of monodispersed spherical particles contain-
ing dye molecules, the fluorescence angular distributions (32) given
in Figure 13 were obtained. The packing density of particles on
the substrate was found to affect the fluorescence profile. With
a layer of particles on a substrate, there are effectively two
interfaces, one defined by the contact area between the prism and
the spheres and another formed in the space between neighboring
particles as an air-prism boundary. A dense packing of particles
led to the angular distribution of curve (c) shown in Figure 13,
which peaked at an effective critical angle corresponding to the
particle-prism interface. In curve (a), for which the spheres were
sparsely packed, the air-prism critical angle dominated the distri-
bution. Curve (b) represents an intermediate packing density. Other-
wise, the results obtained were analogous to those of the liquid
dye configuration.
The preliminary data obtained utilizing internal reflection
techniques to enhance the signal are encouraging and indicate that
additional studies should be completed. Application of resonant in-
ternal-reflection prism approaches, as recently reviewed by Hjorts-
berg et al. (3), should result in much larger signal enhancements
than achieved in the ordinary internal reflection geometry- For
example, utilizing a surface plasmon enhancement in the geometry
of Simon et al. (34) could increase the effective interface excita-
tion intensity by a factor 50 or greater. By choosing the obser-
vation angle to be at resonance, a similar additional enhancement
126
-------
50
40
30
20
10
(a)
(0
30° 60°
ANGLE OF OBSERVATION 6OBS
90C
Figure 13. By replacing the dye-solution shown in Figure 10 with
monodispersed spherical particles (0.81 jura) containing
dye molecules, the fluorescence intensity as a function
of $Qg<; is shown, Dense packing of particles is shown
in (c), while the sparsely packed case ±- shown in (a).
Curve (b) represents an intermediate packing density.
Two effective interfaces (prism-air and prism-particle)
are present, and thus the fluorescence emission is
expected to peak at the corresponding critical angles.
127
-------
should be achievable in coupling the Raman signal back to the prism.
Furthermore, the pronounced dip in elastically reflected light in-
tensity associated with excitation of the surface plasmon mode should
yield excellent isolation from the laser wavelength. Although the
optimum configuration will depend largely on the difficulty of
minimizing unwanted signals generated in the substrate materials,
these resonance methods offer a prospective means of increasing the
sensitivity of the Raman method in analyzing sulfur-containing
particulates.
INSTRUMENTAL TECHNIQUES
While laboratory Raman systems similar to that depicted in
Figure 8 have become highly refined since the advent of the laser,
specialization of equipment to the problem of obtaining spectra
from atmospheric gases, aerosols, and particulates could yield
improvement in signal/noise ratios. Figure 14 is a schematic
representation of a Raman system currently being developed in our
laboratory to investigate atmospheric constituents. In common
with the apparatus of Figure 8, the new system will use a 3W con-
tinuous argon laser for excitation and employ a multipass optical
light trap to increase the total excitation intensity within the
focal volume. Two concave mirrors with dielectric coatings will
be mounted with multiple degrees of freedom to insure optimum align-
ment. Collection optics include the lens L2, and an f/1.2 camera
lens in conjunction with a third spherical mirror. After collec-
tion, the Raman scattered light passes through an interference
filter (IF) chosen to block light at the laser wavelength (5145
A*) with a rejection of 10 4and pass Raman shifted signals (Au —
500-3050 cm~^) with an average transmission exceeding 70%. The
scattered light is focused, commensurate with an f/8, 1 meter
focal length diffraction grating, onto the spectrograph entrance
slit. To determine the frequency components present in the
scattered light, the system uses a single concave holographic
grating which has a high rejection (^10 ) and requires no additional
components inside the spectrometer. Use of a single grating of high
rejection rather than the usual double grating approach signifi-
cantly improves system throughput. As indicated in Figure 14, the
instrument can be used as a conventional spectrometer with an exit
slit, photomultiplier, and photon-counting electronics or as a
Raman spectrograph by imaging a selected portion of the entire
diffracted spectra (about 500 cm"1) on a silicon intensifier tar-
get (SIT) vidicon having approximately 500 channels of resolution.
The parallel-channel approach to detection (35) offers many
advantages. It provides the capability of simultaneously detec-
ting numerous constituents having different Raman shifts within
128
-------
RAMAN SPECTROGRAPH
AR LASER
CONCAVE HOLOGRAPHIC GRATING:
F = 980 MM
o
LINEAR DISPERSION: 5A/MM
2000 G/MM
PHOTON
COUNTING
X
X
1 ^^^^1
i
VIDICON
1
COMPl
1
GRATING
Figure 14. Schematic diagram of a high throughput spectrometer/
spectrograph containing a single concave holographic
grating. For single-channel detection, a specific
Raman wavelength is passed through the slit and then
onto the photomultiplier. For parallel-channel
detection, a range of Raman wavelengths is deflected
by a mirror onto the front face of a vidicon camera.
129
-------
an interval of 500 cm 1 and therefore decreases recording time by a
factor 500. More important, since the dark count rate is comparable
to that of a conventional photomultiplier, shorter collection times
result in larger signal/noise ratios. In addition, spectral ano-
malies and relative peak height variations caused by fluctuations in
laser power and ambient concentration levels which occur as a
spectrometer is scanned through wavelength can be minimized by
parallel-channel detection. Camera scanning and^ grating rotation
(in order to diffract other intervals of 500 cm ) are controlled
by a minicomputer, which can also perform molecular concentration
calculations based on recorded Raman intensities.
A disadvantage of the ORE approach to characterizing sulfur
containing molecules is its inability to probe a wide geographic
area. Unlike LIDAR, which can probe samples distant from the
spectrometer, ORE requires that the species of interest be contained
within the focal volume of the multipass light trap. As a partial
solution, the feasibility of using fiber optic cables to guide
excitation laser light to a remote scattering cell and to return
the Raman scattered light to the spectrometer is also being investi-
gated in our laboratory. A multipass light trap can be used with
each pair of fiber cables to increase the Raman signal. The low
transmission losses (^10 dB/km) of present optical fibers should,
at the least, enable remote sensing using ORE with sufficient
sensitivity to detect toxic species in hostile environments. For
these applications, a single laser and spectrometer could be used
with many pairs of fiber cables to sample numerous locations with
the ultimate sensitivity being limited mainly by the ability to
couple light into and out of the fibers and by the background
scattered signals generated in the transmitting light cables.
CONCLUSION
The Raman effect of sulfur oxides in gaseous, liquid, solid,
and aerosol forms has been summarized, and recent developments in
the understanding of inelastic scattering from small particles have
been discussed. Many specific values of the Raman cross sections,
lineshifts, and linewidths of some important sulfur oxides are still
undetermined. Consequently, it is not possible at this time to
state unequivocally that the Raman technique can be used to dis-
tinguish the chemical species and their phases or to measure the
concentration of each species contained in the emission from com-
bustion sources. The Raman technique does have the potential of
being a viable approach for monitoring simultaneously and in situ
many chemicals in their various phases.
130
-------
This work was supported in part by the American Gas Association
Grant No. GRI 5009-362-0044/AGA BR-142-1, by the National Science
Foundation Grant No. ENG77-07157, and by the Northeast Utilities
Service Company, Hartford, Connecticut.
131
-------
REFERENCES
1. Irish, D. E., and H. Chen. Equilibria and Proton Transfer
in the Bisulfate-Sulfate System. J. Phys. Chem., 74(21):
3796-3801, 1970.
2. Fouche, D. G., and R. K. Chang. Relative Raman Cross Section
for O3, CH4, C3H8, NO, N20, and H2. Appl. Phys. Lett.,
20(7):256-257, 1972.
3. Fouche, D. G., and R. K. Chang. Observation of Resonance
Raman Scattering Below the Dissociation Limit in I2 Vapor.
Phys. Rev. Lett., 29(9):536-539, 1972.
4. Penney, C. M. Remote Measurement of Ozone by Resonance Raman
Scattering and Differential Absorption. IEEE J. Quantum
Electron., QE-11:36D, 1975.
5. Wang, Charles C. , and L. I. Davis, Jr. Measurement of
Hydroxyl Concentrations in Air Using a Tunable UV Laser
Beam. Phys. Rev. Lett., 32(7):349-352, 1974.
6. Penney, C. M., W. W. Morey, R. L. St. Peters, S. D. Silverstein,
M. Lapp, and D. R. White. Study of Resonance Light Scattering
for Remote Optical Probing. NASA Report No. CR-132363, 1973.
7. Rousseau, D. L., and P. F. Williams. Resonance Raman
Scattering of Light from a Diatomic Molecule. J. Chem.
Phys., 64(9):3519-3537, 1976.
8. Fouche, D. G., A. Herzenberg, and R. K. Chang. Inelastic
Photon Scattering by a Polyatomic Molecule: N02. J. Appl.
Phys., 43(9):3846-3851, 1972.
9. Tolles, W. M., and R. D. Turner. A Comparative Analysis of
the Analytical Capabilities of Coherent Anti-Stokes Raman
Spectroscopy (CARS) Relative to Raman Scattering and Absorp-
tion Spectroscopy. Appl. Spectry., 31(2):96-103, 1977.
10. Hill, R. A., and D. L. Hartley. Focused, Multiple-Pass Cell
for Raman Scattering. Appl. Opt., 13(1):186-192, 1974.
11. Stafford, Richard G., and Richard K. Chang. Absolute Raman
Scattering Cross Sections of Sulfate and Bisulfate and Their
Application to Aqueous Aerosol Monitoring. In: Internation-
al Conference on Environmental Sensing and Assessment,
Vol. 2, IEEE, New York, 1976. pp. 23-5/-1-4.
132
-------
12. Woodward, L. A., and R. G. Homer. Changes in the Raman
Spectrum of Sulphuric Acid on Dilution. Proc. Roy. Soc.
(London), 144A(851):129-143, 1934.
13. Hibben, James H. The Raman Effect and Its Chemical Applica-
tions. Reinhold Publishing Corp., New York, 1939. 544 pp.
14. Zarakhani, N. G., N. B. Librovich, and M. I. Vinnik. Homo-
geneous Catalytically Active Solutions. VII. The Equilibrium
Compositiion of the Sulphuric Acid-Water System. J. Phys.
Chem., 45(7):981-983, 1971.
15. Lindstrom, Richard E., and Henry E. Wirth. Estimation of
the Bisulfate Ion Dissociation in Solutions of Sulfuric Acid
and Sodium Bisulfate. J. Phys. Chem., 73(1):218-223, 1969.
16. Young, T. F., and G. E. Walrafen. Raman Spectra of Concentrated
Aqueous Solutions of Sulphuric Acid. Trans. Faraday Soc.,
57(l):34-39, 1961.
17. Chew, H., P. J. McNulty, and M. Kerker. Model for Raman and
Fluorescent Scattering by Molecules Embedded in Small Particles.
Phys. Rev., A 13(1):396-404, 1976.
18. Kratohvil, J. P., M.-P. Lee, and M. Kerker. Angular Distributior
of Fluorescence from Small Particles. To be published, Appl.
Opt., 1978.
19. Lee, El-Hang, R. E. Benner, J. B. Fenn, and R. K. Chang.
Angular Distribution of Fluorescence from Monodispersed
Particles. To be published, Appl. Opt., 1978.
20. Stafford, R. G., R. K. Chang, and P. J. Kindlmann. Laser-Raman
Monitoring of Ambient Sulfate Aerosols. In: Methods and
Standards for Environmental Measurement, Proceedings of the
8th Materials Research Symposium. NBS Special Publication
464. William H. Kirchoff, ed. U.S. Government Printing
Office, Washington, D.C., 1976. pp. 659-667.
21. Anathanarayanan, V. A Comparative Study of the Raman Spectra
of Anhydrous Sulphates and Estimation of Crystalline Forces.
Indian J. Pure Appl. Phys., 1(2):58-61, 1963.
22. Blomer, F., and H. Moser. Absolute Intensitatsmessungen an
Raman-Kristallpulverspektren von Nitraten und Sulfaten. Z.
angew. Phys., 27(5) :302-306, 1969.
133
-------
23. Wright, M. L., and K. S. Krishnan. Feasibility Study of In-
Situ Source Monitoring of Particulate Composition by Raman
or Fluorescence Scatter. EPA-R2-73-219, U.S. Environmental
Protection Agency, Washington, B.C., 1973. 117 pp.
24. Radian Corp. Literature Evaluation of Methods Available for
the Collection and Measurement of Sulfur Dioxide, Sulfur
Trioxide, Sulfuric Acid, and Particulate Sulfate in Ambient
Air. RC No. 200-077. Austin, Texas, 1974. 80 pp.
25. Rosasco, G. J., E. S. Etz, and W. A. Cassatt. The Analysis
of Discrete Fine Particles by Raman Spectroscopy. Appl.
Spectry., 29(5):396-404, 1975.
26. Rosen, H., and T. Novakov. Identification of Primary
Particulate Carbon and Sulfate Species by Raman Spectroscopy.
Topical Meeting of the Optical Society of America. Aerosols:
Their Optical Properties and Effects. Williamsburg, Va., 1976
27. Van de Hulst, H. C. Light Scattering by Small Particles.
John Wiley & Sons, New York, 1962. 470 pp.
28. Kerker, Milton. The Scattering of Light and Other Electro-
magnetic Radiation. Academic Press, New York, 1969. 666 pp.
29. Byer, R. L. Review, Remote Air Pollution Measurement.
Optical & Quantum Electron., 7(3) :147-177, 1975.
30. Harrick, N. J. Internal Reflection Spectroscopy. Inter-
science Publishers, New York, 1967. 327 pp.
31. Carniglia, C. K., L. Mandel, and K. H. Drexhage. Absorption
and Emission of Evanescent Photons. J. Opt. Soc. Am.,
62(4):479-486, 1972.
32. Lee, El-Hang, R. E. Benner, J. B. Fenn, and R. K. Chang.
Angular Profile of Fluorescence from Monodispersed Spherejs
by Evanescent Wave Excitation. To be published.
33. Hjortsberg, A., W. P. Chen, and E. Burstein. Resonant
Internal-Reflection Prism Spectroscopy Using Surface, Guided,
and Fabry-Perot EM Waves. Appl. Opt., 17(3):430-434, 1978.
34. Simon, H. J., R. E. Benner, and J. G. Rako. Optical Second
Harmonic Generation with Surface Plasmons in Piezoelectric
Crystals. Opt. Commun., 23(2):245-248, 1977.
134
-------
35. Black, P. C., and P. J. Kindlrnann. Parallel-Channel
Detection System for Low Light Level Spectroscopy. In:
Proceedings of the Technical Program, Electro-Optical Systems
Design Conference-1973, New York, 1073. pp. 206-21o.
135
-------
Report of the Working Group on
Measurement of Gaseous Sulfur Oxides
Emissions
Russell N. Dietz, Reporter
This working group, charged primarily with the responsibility
for examining the pertinent aspects of sampling and analysis for the
gas phase sulfur constituents in flue gas, made recommendations in
the following areas:
• Definition of terms
• Essential measurements
• Units for data presentation
• Sampling methodology
• Specific research needs
DEFINITION OF TERMS
Molecular Formulation
Correct molecular formulation for the presentation and reporting
of constituents is essential. For example,
• sulfur dioxide - S02
• sulfur trioxide - S03
• sulfate salts or sulfates - MS04
(do not use S04 only)
137
-------
Sulfuric Acid
H2SO4 can exist as a vapor, adsorbed vapor, or liquid aerosol.
The H2S04 vapor exists as a gas in the flue gas with a certain fraction
existing in the adsorbed state on particulate matter in the flue gas.
If the temperature of the flue gas is below the dewpoint temperature
for the sulfur acid (not the usual case), some of the H2S04 could be
present as liquid aerosol.
Sulfate Salts
The balance of sulfate material collected on a filter medium
is considered to be comprised of sulfate salts. Some of the remain-
ing sulfate fraction on the particulate matter may be chemisorbed
H2SO4, but there would be no simple means for distinguishing that
from true sulfate salts.
Total Sulfates
Sulfuric acid and sulfate salts may be referred to collectively
as total sulfates. But in every case of usage, the authors should
explicitly define their meaning.
Artifact Sulfate
Sulfate that is generated by the measurement technique, and
that is determined as sulfate because of either an interference
problem or because of conversion of SO2 to sulfate during the col-
lection and/or determination, may be considered to be artifact
sulfate and is, in fact, a measurement error.
ESSENTIAL MEASUREMENTS
Separation of H2SO4 and Sulfate Salts
A filter should be utilized to adequately separate the flue gas
particulate matter from the sulfuric acid in the vapor phase. The
filter should first be extracted with an appropriate solvent (e.g.,
anhydrous acetone, isopropyl alcohol) for the recovery of adsorbed
H2SO4 or liquid aerosol H2S04. This is followed by extraction with
water for recovery of the water soluble sulfate salts. The vapor
phase fraction of the H2S04 should be appropriately collected
downstream of the filter.
138
-------
Sampling Location, Temperature, Water Vapor, and Oxygen Level
The location of the sampling point should be clearly indicated.
In addition, the temperature and water vapor content at the sampling
location should be specified in order to determine the physical and
chemical state of the sulfuric acid (i.e., vapor H2S04, liquid
aerosol H2SO4) or gaseous S03). Sampling location of the oxygen level
should also be determined in order to correct for leaks in the flue
gas ducting downstream of the boiler.
Furnace Oxygen
The furnace or boiler 02 level has been shown to directly affect
the levels of H2S04 and sulfate salts formed within the combustion
unit. It is essential that the amount of furnace or boiler oxygen
be determined and that this quantity not be confused with the oxygen
level existing at the sampling location.
UNITS FOR DATA PRESENTATION
Gaseous Species
Volumetric units such as ppm (parts per million), denoted by
vol. ppm or by vol. %, may be used for reporting the concentration
of gaseous species.
Liquid or Solid Species
Mass units reported as mass per unit volume of flue gas is
preferred for liquid and solid species. The formula name of the
specie being reported must be specified since units of mass are be-
ing usea. In addition, the conditions at which the gas volume is
being reported must be indicated (i.e., temperature and pressure).
In both cases above, the other conditions of the flue gas such
as wet or dry basis, and whether or not corrections were made for
flue gas ducting leaks, should be reported.
139
-------
SAMPLING METHODOLOGY FOR H2SO4
Isopropyl Alcohol Methodology
With regard to the application of methods utilizing isopropyl
alcohol (IPA) for the determination of sulfuric acid, there are
problems associated with the use of a small static amount of IPA in
contact with a significant volume of flue gas. At sulfuric acid con-
centrations in flue gas less than 3 to 5 vol. ppm, significant errors
in H2SO4 determination can occur because of H2S04 aerosol collection
inefficiency, residual dissolved S02, or oxidation of S02 to H2S04 .
It is not possible to measure H2S04 concentrations less than 1 vol.
ppm in flue gas using IPA methodology (e.g., EPA Modified Method 6 or
Method 8).
If utilizing IPA methodology, an appropriate reagent grade, free
of oxidant, should be used.
Controlled Condensation Methodology
The controlled condensation methodology is the preferred ap-
proach for manual determination of H2S04. There is a critical need
for the specification of equipment design and sampling practice, in-
cluding approaches addressing the problems of pre-filtering, final
condenser filtering, and pressure drop considerations.
Acid Dewpoint Methodology
Such monitors have limited application; other direct methods
such as controlled condensation are preferred.
SPECIFIC RESEARCH NEEDS
Particulate and Vapor Separation
Research must be directed towards separation methods for
adequately distinguishing between the H2SO4 vapor and the sulfate
salts. Potential separation techniques include filtering, electro-
static precipitation, and cyclone separation.
140
-------
Optical Methods
Research should be directed towards studying optical techniques
because of their potential for in-stack determinations for multiple
compounds.
Continuous IPA H2S04 Monitor
The continuous automatic analyzer for H2SO4 developed by the
Central Electricity Generating Board in England, based on co-current
absorption in IPA solution followed by reaction with barium chlor-
anilate for colorimetric determination as 535 nm, should be tested
and evaluated.
Automatic Controlled Condensation
The EPA approach should be studied in more detail.
Validation of Methods
There should be a continual examination of reference and field
methods for latest improvements and up-dating of these techniques
as field experience is acquired.
141
-------
Section 2
Paniculate Sampling
and Analysis
143
-------
Collection Methods for the Determination of
Stationary Source Particulate Sulfur and Other
Elements
Kenneth T. Knapp
Roy L Bennett
Robert J. Griffin
Raymond C. Steward
U.S. Environmental Protection Agency
ABSTRACT
The growing concern over the chemical composition of the
particulate emissions from stationary sources has led
to the improvement oi sampling methods, so that re-
presentative samples compatible with chemical analysis
can be obtained. Improvements include the development
of techniques for obtaining particle-sized samples.
Analytical techniques used should be sensitive; require
minimal sample preparation and small samples, and work
quickly. One technique that offers these qualities is
x-ray fluorescence spectrometry (XRF). Some XRF systems
can determine 30 elements in less than 10 minutes. In
addition, these systems are non-destructive and require
little preparation for filter samples. For light elements
such as sulfur, the best results are obtained using samples
collected uniformly on the surface of thin substances
with a low mass, such as filters made of organic polymers.
The collection of samples on thin membranes at many
sources has been the source of several major problems.
The most severe problems are related to the temperature
limitation of the filters and their degradation by hot
sulfuric acid emissions. However, good samples for chemi-
cal analysis can be obtained in the field using temperature
controls and careful handling. Since, in general, only
small amounts of material are collected, the samples must
be handled carefully in the laboratory to avoid sample
loss, contamination, and moisture effects.
145
-------
The use of impactor-sized samples for XRF analysis
creates additional problems, including non-uniform
collection, many small piles of material, and sub-
strates that usually are not compatible with the XRF
spectrometer. Several techniques are presently being
used to transfer these types of samples to compatible
substances. Another approach being pursued is the
development of a new system that will yield uniform-
sized samples on compatible substrates.
INTRODUCTION
The growing concern for information on the chemical composi-
tion of particulate emissions from stationary sources has resulted
in research to improve sampling methods so that representative
samples are compatible with the analytical techniques. The infor-
mation sought includes the chemical analysis of both bulk and sized
particulate emissions. The analytical techniques used should have
good sensitivity, require small samples, and need only minimal
sample preparation. X-ray fluorescence spectrometry (XRF) is one
technique that offers these qualities. It is fast and non-destruc-
tive, and filter samples require little or no preparation. The
XRF system used by the Particulate Emissions Research Section
(PERS) of the Environmental Sciences Research Laboratory of EPA can
determine 30 elements including sulfur in a filter sample in less
than ten minutes. Since XRF analysis of source emissions and the
PERS system have been described in several publications, no detailed
discussion of them will be presented in this paper (1)(2)(3).
The sampling techniques described in this paper have been
designed for optimum XRF analysis of the emissions from stationary
sources. Special attention has been given to sulfur analysis.
Two types of sample collection techniques are discussed, those for
bulk or total sample analysis and those for particle-sized sample
analysis.
METHODS
Systems that are used to obtain samples for chemical or
other analysis from stationary sources have basically three parts.
The first part transfers the emitted sample from the source to the
collecting device. This transfer section, sometimes referred to as
the sampling interface, can be a simple nozzle or a complex boun-
dary layer quantitative transport system. The collecting device,
the second part of the sampler, can be open cups, impaction sur-
faces or filters. Filters have several advantages over the other
146
-------
devices, including low cost, ease of handling, and efficiency. The
final part of the sampler is the gas flow section which contains
the gas pump and a flow measuring device. For particle-sized
samples, the collecting device generally performs the sizing,
Obtaining a gas flow system that will provide a reasonably
accurate flow over a wide range is no major problem. Therefore,
no further discussion of it will be given. Many investigators have
worked on the problems of sample transport, and several publica-
tions are available on this subject (4)(5)(6); therefore, it
too will not be discussed further.
Filters
One of the most critical problems in measuring the particulate
emissions from sources is getting a truly representative sample
to the measuring device. Of equal importance, when the chemical com-
position of the emissions is to be determined, is the proper selec-
tion of the collecting surface or device. Filters are generally used
to collect the samples. They are usually simple to use and are
compatible with many analytical techniques such as XRF. Listed in
Tables 1 and 2 are the filters which PERS has used for chemical
analysis of source emissions. Table 2 is a list of fluorinated
polymer filters tested. Figure 1 shows the pressure drop at various
flows for these filters.
The most commonly used filter for air pollution work is the
glass fiber filter. While it has several advantages for measuring
total particulate mass such as low pressure drop, high capacity,
ease of weighing, and cost, it has several major drawbacks for
XRF analysis. Even the best grade of glass fibers has high x-ray
backgrounds due to trace element contamination. The filters also
scatter large amounts of x-rays. In addition, these filters are
depth f?Iters where the particles penetrate deep into the filters
and cause attenuation of the characteristic x-rays from the light
elements such as sulfur. This makes corrections necessary. Theo-
retical corrections are hard to determine; therefore, empirical
corrections are generally used.
In an attempt to obtain a filter with a lower background, EPA
contracted for the development of a filter made of quartz fibers.
Such a filter was developed and is described in an EPA report (7).
However, the filters tr.ad<- of quartz fibers • .ce not strong enough to
be useful in general field sampling.
147
-------
Table 1. Characteristics of Filters Used for Source Sampling
00
Filter
(pore size)
Glass fiber
Quartz
Millipore AA
(0.8 fim)
Nucleopore
(0.8 Mm)
Composition
Glass fiber
Si02 fibers
Mixed esters
of cellulose
Polycarbonate
Upper Temp.
Limit °C
>250
>250
105
130
Ease of
Weighing
Good
Fair
Poor to good
(high static
change)
Good
Ease of
Handling
Good
Poor
Good
Poor
Remarks
High capacity, good ex-
tractive filter, poor
for XRF , poor for in-stack
Too fragile for general
field use
Good for XRF, decomposed
by hot H2SO4
Excellent for XRF, high
pressure drop, low capacitj
Gelman HT 650
(0.65 jim)
Unknown sulfur 135
containing polymer
Good
Good Not usable for sulfur-
sulfate analysis by XRF
-------
Table 2. Polytetrafluoroethylene* and Other Fluorinated Polymer Filters
(Ł>
Filter
(pore size)
Fluoropore
(1 Mm)
Millipore FSLW
(3 urn)
Ghia Dual
Teflon (1 fim)
S&S TE36
(0.5 fim)
S&S TE37
(1 fim)
Mitex LC
(10 fim)
Millipore
457-55
(1 nm)
Composition
PTFE on
polypropylene
PTFE on non-
woven PP
PTFE on PTFE
PTFE on non-
woven PP
PTFE on non-
woven PP
FTF^
Unknown
f luorinated
polymer
Upper Temp.
Limit °C
150
(PP)
150
>250
150
150
>250
200
Ease of
Weighing
Fair
Good
Good
Good
Good
Good
Good
Ease of Hand-
ling in Field
Poor-Fair
Poor
Poor
Poor
Poor
Fair
Fair
Remarks
Low capacity, backing
changes shape in hot gase
Hard to center in filter
holder, low pressure drop
Hard to center in filter
holder, low capacity
Pressure drop too high
Hard to center in filter
holder, low capacity
Filter efficiency too low
Pressure drop too high
*PTFE
-------
-=, 5
CFM 0.2
LPM 5.7
U DUAL TEFLON
O MILLIPOREAA
A DUAL TEFLON
0 NUCLEPORE 0.8/j
O GLASS FIBER A
O MITEX 1< LCWP
DUAL TEFLON 3/j=<1" Hg
FSLW3u=1"Hg
Figure 1. Pressure drop vs. flow for 47 mm filters.
150
-------
Many of the synthetic polymeric fiber filters have low levels
of trace element contamination and low x-ray scattering. In
addition, some of these filters collect the material on their
surfaces and minimize the x-ray absorption problem. The Nucleopore
filters have these qualities but have the drawback of high pressure
drop for the high efficiency filters. They also have a temperature
limitation of 130°C and have low capacity for particulate loading.
These filters are used for collecting extractive samples where the
temperature can be maintained below 130°C and when low loading is
desired such as for microscopy.
The polymeric fiber filter used most extensively in the past
field work by PERS for collecting samples for chemical analysis
by XRF is Millipore. Generally, the 47 mm 0.8 jim pore size AA
filters were used because of their ease of handling in the field,
low XRF background, and adequate capacity for particulate loading.
However, several major problems have been discovered with these
filters in source sampling. One problem is their thermal insta-
bility. The filters will totally disintegrate at temperatures above
125°C. At most sources, this limits the use of these filters to
carefully controlled extractive sample systems. Exposure of these
filters to gas streams with large amounts of hot sulfuric acid
(temperature above 100°C) results in the saponification of the
acetate esters and the loss of filter weight. These filters are
sensitive to the moisture content of the air and, therefore, give
different and sometimes changing weights with different air moisture
content. These weight changes may be as high as 100 micrograms.
The static charge of these filters presents a problem that affects
weighing them on microbalances. Weighing in a carefully controlled
balance room has minimized these weighing problems. Even with these
serious disadvantages, the AA filters are the most frequently used
for source characterization by PERS, since no known good replacement
is currently available.
The Gelrnan HT G50 filter was considered since it has good
temperature stability, ease of weighing and handling, and sufficient
capacity. However, the filter cannot be used in the PERS source
characterization work since it contains sulfur and gives too high
an XRF sulfur background.
The fluorinated polymeric fibers have good temperature stability
and low XRF background. However, because these fibers do not hold
together well, good filters of only fluorinated polymers have been
difficult to make. Most fluorinated polymer filters are made with
151
-------
some type of backing, usually polypropylene. Both woven and non-
woven backings are used. Several new filters which have the fluori-
nated polymer backing have become available and have a high (>250°C)
temperature limit. All the fluorinated polymer filters are hard
to handle in the field, and most have low capacity for particulate
loading and high pressure drop. The 10 pm Mitex LC filter has too
low efficiency to be useful. With improved handling techniques,
the fluorinated polymer filters may emerge as the best overall fil-
ter for source characterization.
Table 3 gives the element sensitivity for the PERS XRF sys-
tem. The background counts for selected elements of three fil-
ters are given in Table 4.
Sizing Device
The two types of sizing devices used in the source charac-
terization work at PERS are the cascade impactors and series cyclone,
The series cyclone was used to get bulk-sized material that
was not deposited on some surface. This material was used for
carbon and other analyses that are done by methods other than
XRF. The system used consisted of a three-cyclone unit with
cut points at 3.3, 1.8, and 0.8 /im. This cyclone unit could
be used in-stack; however, it was more practical to use it
with the same extractive probe used with the extractive im-
pactor described below. Several investigators have continued
to work on the development of series cyclones for analytical
use and a group at Southern Research Institute has developed a
five-series in-stack system (8).
Good in-stack impactors are available. However, all of
them collect the samples on surfaces that are not compatible
with the PERS XRF system. Two approaches have been developed
that present the sized particles in a suitable form for x-ray
analysis. One approach is to use the Andersen Model 0203, one-
cfm ambient sampler with polycarbonate films for collection
surfaces as an extractive impactor. The other approach is to
redeposit the collected material of the in-stack impactors on
a suitable substrate for XRF analysis. The sample in the
Andersen System is extracted from the sources with a heated
probe into the impactor which is held in a special heated samp-
ling box. The collected sized samples are analyzed by XRF on
the collection films. Besides the drawbacks of losing material
in the probe, the results suffer from the non-uniformity of
the material deposits. Not only are the deposits not uniform,
but each impactor stage gives a different material deposit
152
-------
Table 3. Emissions Characterization Data for Coal-Fired
Power Plants with ESP's
en
CO
Date
Plant PC
7/19/77
7/20/77
7/22/77
Plant MCd
7/26/77
7 /27 /77
7/28/77
7/28/77
Part.
Time (mg/Nm3)
1551-1636
1020-1120
0939-1039
1430-1545
0915-1015
0900-1COO
1110-1210
840
1,100
360
330
2,800
2,100
450
S02
(mg/m3 )
7,200
6,000
7,300
5,900
5,900
6,500
6,500
S04
(mg/m3)
250
330
180
210
190
240
240
SO 4
(
3
5
2
3
3
3
3
/SOX
%)
.4
.2
.4
.4
.1
.6
.6
Q
Plume
Opac. (%)
27
23
32
18
86
80
20
In-Stackb
Opacity (%)
21
28
31
21
81
72
17
2
2
2
1
3
3
Remarks
fields
fields
fields
field
fields
fields
off
off
off
off
off
off
Normal operatioi
d
Observer measurements.
r\
Tra ismissometer measurements.
'8% Ash, 3.3% S, 99 ppm V, 100 MW.
14% Ash, 3.9% S, 35 ppm V, 330 MW.
-------
Table 4. Magnitude and Uniformity of Background Count
for Three Types of Membrane Filters (2)
Ol
Nucleopore
0.8 n
(Mass = 1.1
mg/cm2)
Fluoropore
Type FA
(Mass = 2.7
mg/cm2)
Millipore
Type AA
(Mass = 5.0
mg/cm2)
F
Na
Mg
Si
P
S
Cl
K
Ca
Ti
V
Fe
Ni
Cu
Zn
Cd
Ba
Na
1.32
0.22
0.64
66.22
17.77
48.54
86.49
64.48
4.43
2.91
7.95
44.28
2.10
553.46
12.14
1.87
2.35
ab
0.09
0.06
0.08
4.44
0.62
0.82
3.18
2.67
1.17
0.56
0.80
4.21
0.36
9.03
0.81
0.08
0.35
N
549.78
0.15
0.75
1.25
17.63
40.21
54.44
63.49
8.85
7.67
16.10
50.41
3.51
567.36
26.31
2.12
5.00
a
28.20
0.05
0.10
0.25
2.16
2.23
4.81
2.46
3.15
1.20
1.93
4.86
0.67
14.13
3.49
0.11
0.57
N
1.96
0.28
1.43
2.29
16.34
54.52
200.24
212.40
333.84
12.74
25.71
62.51
5,52
577.69
41.34
3.18
8.27
a
0.13
0.05
0.16
0.27
0.35
0.63
5.59
2.40
7.13
1.62
1.65
4.79
1.12
21.61
6.82
0.17
0.94
,N is the mean value in cps for ten blank specimens.
a is the standard deviation in background for ten blank specimens,
-------
distribution, i.e., there are different sizes and numbers of
piles of material. In spite of these problems, usable results
based on elemental ratios have been obtained. The results from
some of the PERS studies are given in another paper of this
workshop (9) .
Since better sizing of source-emitted particles is obtained
with the in-stack impactors due to the elimination of probe
losses and collection of the material at stack conditions, a
technique for obtaining XRF analysis on these sized materials was
desired. An extraction redeposition technique has been developed
which removes the material from the impactor collection plates
and redeposits it on suitable substrates. In this technique, the
collected sized material is washed from the impactor plates into
glass evaporation chambers with suitable substrates at the bottom.
The extraction solvent is then evaporated under a stream of dry
nitrogen. During the evaporation step, the sides of the cham-
bers are frequently washed with clean solvent. Good transfer
of the sized material can be obtained; however, much care must
be used in the removal of the material from the impactor plate,
the washing down of the chamber walls, and the handling of the
redeposited sample. The results from three runs of sulfur distri-
bution among sized material from an oil-fired power plant by the
redeposition technique are given in Table 5. Table 6 shows the
results for all the important elements detected in one of these
samples. The data given in these two tables illustrate the type
that can be obtained by this technique. The vanadium is enriched
in the fine particles with iron, calcium, silicon, and aluminum
present in higher percentages in the larger particles.
SUMMARY
When sulfur and other elemental analyses are needed on emis-
sions from stationary sources, the collection substrate must be
carefully chosen. Each of the filters now in use has some draw-
back such as temperature limitation, low capacity, or poor XRF
background; therefore, compromises must be made in filter selec-
tion. In spite of all the problems, good chemical analysis can
be obtained from XRF analysis of samples collected on Nucleopore,
Millipore AA, and some of the fluorinated polymer filters.
As with filters, compromises must be made in selecting devices
for analysis of sized particulate emissions. With in-stack impac-
tors, the sized material must be transferred to a suitable sub-
strate. Extractive impactors suffer from probe losses and non-
uniform deposits.
155
-------
With careful work, good representative samples can be obtained
that will yield good chemical analysis of emissions from sources.
Table 5. Percent Sulfur in Sized Fractions
Oil-Fired Power Plants
Stage
DSO i fm
23
10
4.7
1.9
1.0
0.52
0.27
Filter
Boiler Excess 02
0.2%
-
1.6
2.0
-
3.9
8.9
2.6
2.4
Levels
0.25
2.1
2.9
3.5
3.2
5.3
3.1
1.3
2.4
0.6%
5.3
3.6
1.9
4.6
3.5
4.9
2.4
1.7
156
-------
Table 6. Distribution of Important Elements in
Sized Fractions - Oil-Fired Power Plant
Stage
DBO , nm
23
10
4.7
1.9
1.0
0.52
0.27
Filter
Elements, Weight Percent
S
5.3
3.6
1.9
4.6
3.5
4.9
2.4
1.7
V
2.8
2.0
0.4
0.6
0.5
0.3
0.007
5.1
Ni
0.3
0.2
0.03
0.02
0.03
0.02
NF
0.8
Fe
2.7
1.4
0.2
0.2
0.3
0.2
0.02
0.1
Ca
0.4
0.1
0.05
0.04
0.1
0.02
0.009
NF
Si
0.2
0.3
0.6
0.6
0.3
0.6
0.1
NF
Al
0.1 <
0.03
0.009
0.008
0.02
NF
NF
NF
157
-------
REFERENCES
1. Bennett, R. L. , J. Wagman, and K. T. Knapp. The Application
of a Multichannel Fixed and Sequential Spectrometer System
to the Analysis of Air Pollution Particulte Samples from
Source Emissions and Ambient Air. In: Advances in X-ray
Analyses, Vol. 19, Kendall/Hunt Publishing Company, Dubuque,
Iowa, 1976. pp. 393-402.
2. Wagman, J., R. L. Bennett, and K. T. Knapp. X-ray Fluorescence
Multispectrometer for Rapid Elemental Analysis of Particulate
Pollutants. EPA-600/2-76-033, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina, 1976. 44 pp.
3. Wagman, J., R. L. Bennett, and K. T. Knapp. Simultaneous Multi-
wavelength Spectrometer for Rapid Elemental Analysis of
Particulate Pollutants. In: X-ray Fluorescence Analysis of
Environmental Samples, T. G. Dzubay, ed. Ann Arbor Science
Publishers, Inc., Ann Arbor, Mich., 1977. pp. 35-55.
4. Ranade, N. B. Sampling Interface for the Quantitative Trans-
port of Aerosols, Field Prototype. EPA-600/2-76-157, U.S.
Environmental Protection Agency, Research Triangle Park,
North Carolina, 1976. 68 pp.
5. Lundgren, D. A., and M. D. Durham. Aerosol Sampling in Tur-
bulent or Tangential Flow. EPA Report to be published, U.S.
Environmental Protection Agency, Research Triangle Park,
North Carolina, 1978.
6. Lundgren, D. A., M. D. Durham, and K. W. Mason. Sampling
Tangential Flow Streams. J. of Am. Industrial Hygiene
Association, Summer 1978 (in press).
7. Benson, A. L., P. L. Levins, A. A. Massucco, and J. R.
Valentine. Development of a High-Purity Filter for High
Temperature Particulate Sampling and Analysis. EPA-650/2-73-
032, U.S. Environmental Protection Agency, Washington, D.C.
1973. 71 pp.
158
-------
8. Smith, W. B., and R. Wilson, Jr. Development and Laboratory
Evaluation of a Five-Stage Cyclone System. EPA-600/7-78-008,
U.S. Environmental Protection Agency, Washington, D.C., 1978.
66 pp.
9. Bennett, R. L., and K. T. Knapp. Sulfur and Trace Metal
Particulate Emissions from Combustion Sources. In: Measure-
ment Technology and Characterization of Primary Sulfur Oxides
Emission from Combustion Sources, Southern Pines, North Caro-
lina, April 1978.
159
-------
A Stack Gas Sulfate Aerosol Measurement
Problem
Dale A. Lundgren
Paul Urone
University of Florida
Thomas Gunderson
Los Alamos Scientific Laboratory
ABSTRACT
Measurement of the emission of sulfur-containing par-
ticulate matter from a combustion source is complicated
by the presence of high concentrations of relatively
reactive sulfur gases, namely, S02 , SO3 , and H2SC>4 .
As stack gas temperature changes, the gases may react,
change phase (condense), or adsorb out on surfaces —
especially particulate collection surfaces such as
filters. If appropriate precautions are not taken, it
is not uncommon or improbable for gaseous sulfur com-
pounds to contribute more sulfur to a particulate
catch than the "precollection" stack gas particulate
matter.
This problem can be magnified when attempts are made to
determine the in-stack particle size distribution of a
sulfur-containing aerosol. Ga.s sorption-reaction on
surfaces such as glass fiber filter-covered impaction
plates can be a major source of error in measuring par-
ticles at in-stack conditions. Particle size measure-
ments made out-of-stack frequently require cooling of
the extracted gas stream, which, in turn, can cause
reaction-condensation of SO3-H2S04 tc ^articulate or
sorption of water vapor by hygroscopic sulfate particles
causing size change.
These problems will be discussed as they relate to sul-
fate particle concentration and size distribution
measurement.
161
-------
INTRODUCTION
A major problem in stack gas sampling and analysis is how to
classify gases which condense or react in the sampling train or
on the particle collection medium to form what may be considered
particulate matter. One substance which falls into this cate-
gory is sulfur trioxide (SO3), formed in equilibrium with sulfur
dioxide (SO2) when sulfur-containing fossil fuels are burned.
Up to 5% of the total sulfur in the fuel is converted to S03,
yielding from 5 ppm to 50 ppm SO3 in the flue gas (1)(2). The
SO3 is in equilibrium with water (H2O) vapor in the flue gas,
and, depending on temperature and gas component concentrations,
various amounts of sulfuric acid (H2S04) vapor will be formed.
This H2SO4 can be collected on filters and weighed as particulate.
The importance of establishing whether or not condensed
SO3/H2SO4 is to be considered particulate matter is pointed
out by a 36% average contribution of this material to total
measured particulate grain loading (oil-fired boiler emissions)
as reported by Jaworowski (3). As fly ash emission levels are
reduced by air pollution control equipment, this amount of con-
densed SO3/H2S04 may equal or surpass the dry particulate sulfate
contribution.
SOX , H2O AND H2SO4 EQUILIBRIA IN FLUE GAS
Most sulfur in power plant flue gases appears as S02 (Table
1) (4), with typical SO3 levels ranging from 1.0% to 2.5% of the
SO2 . However, as Figure 1 shows, the equilibrium constant for
the reaction:
S02 + 1/2 C2 IŁSO3
strongly favors the formation of SO3 at temperatures below about
540°C (1000°F). This graph was calculated from data cited by
Hedley (5). Kinetics of the reaction are unfavorable in the
absence of a catalyst, but thermodynamically the SO3 concentra-
tions could exist at levels much greater than those normally
encountered. Ratios of S03 to S02 as high as 0.1 have been
reported (6). Since formation of SO3 is controlled by catalytic
effects as well as amount of excess air present, concentration
of S03 resulting from combustion of a particular fuel can only
be estimated in absence of direct measurements.
Reaction between H20 vapor and S03 is given by:
H20 -t- S03 + H2S04 .
162
-------
c*>
I
ro
O
CO
O
-P
04
O
•n
O
c
O
•H
en
O
u
100
•H
s
tr
w
80
60
40
20
0
600
3% 0.
700
800
900 1000 1100 1200
Temperature - °F
Figure 1. Equilibrium conversion of SO to SCL [from Hillenbrand,
Engdahl, and Barrett (A)].
163
-------
Figure 2 shows equilibrium conversion of SO3 to H2S04 as a
function of temperature for a typical flue gas H2O vapor concen-
tration of 8 vol%. At temperatures below 204°C (400°F), essen-
tially all S03 present is converted to H2SO4 at equilibrium.
In contrast to formation of SO3, formation of H2S04 occurs
rapidly in the thermodynamically feasible temperature range (7).
Table 1. Typical Exhaust-Gas Composition from Coal-Fired Boiler
[From Hillenbrand, Engdahl and Barrett (4)]
Component
H20
CO,
Concentration
Volume
Percent
4.0
15.0
a
g/m3
30
273
Fly ash before
precipitator
Fly ash after
precipitator
NO
SO2
S03
Hydrocarbons
0.050
0.20
0.0030
(30 ppm)
(0.0010)
9.16
(4 gr/ft3)
0.458
(0.2 gr/ft3)
0.63
5.3
0.10
a
At 21°C (70°F), 1 atmosphere
DETERMINATION OF H2SO4 DEWPOINT
Fly ash particles can influence the apparent dewpoint
(saturation temperature of H2S04 in flue gas), but one commits
practically no error by neglecting the presence of other gases
and considering only the H2S04-H20 system (8). Thermodynamic
164
-------
100
c*>
o
w
c
a
o
<*>
o
to
c
o
D
•H
•H
^H
•r4
80
60
40
20
200
300
400
500
600
700
Temperature - °F
Figure 2. Equilibrium conversion of S03 to H2S04 at 8.0 vol% in flue
gas [from Hillenbrand, Engdahl, and Barrett (4)).
165
-------
analysis of the H2SO4-H20 flue gas system, ignoring fly ash
effects, provides a theoretical basis for predicting acid dew-
points and condensate composition from vapor-liquid equilibria
data.
Abel (9) was the first to derive a relationship enabling
calculation of H2S04, H20 and S03 partial pressures from enthalpy,
entropy, free energy, and heat capacity values. From his H2SO4
partial pressures and Greenewalt's H2O partial pressures over
11^804 solutions, H2S04-H2O dewpoint charts were prepared (Figures
3 and 4). The range of uncertainty indicated by Abel is on the
order of 5°C (9°F) at 10 vol% H20 vapor.
Information contained in Figure 3 can be used to predict
dewpoint temperature from an analysis of H2S04 and H20 vapor
content. If gas is cooled below its dewpoint, condensate equili-
brium concentration and mass can be obtained. Condensate mass
predicted from use of the dewpoint chart is actually a prediction
of the amount available for condensation. The actual amount of
condensate depositing on a fiber or metal surface may differ from
the chart prediction because of mass transfer considerations.
As an example of the use of the chart, consider a flue gas
containing 10 ppm H2S04 and 10 vol% H20 vapor. Condensation
would occur at about 135°C (275°F), and condensate composition
at that point would be about 79 wt% H2S04. If the gas were
cooled to 121°C (250°F), 85% of the H2SO4 would be removed from
the gas phase and an insignificant amount of H20 vapor would
also condense. Condensation, therefore, follows the 10 vol% water
line, resulting in a condensate which would be the equilibrium
composition of the condensate at 121°C (250°F), assuming the vapor
phase is in equilibrium with the total liquid condensed. Compo-
sition change of the liquid is small over the temperature interval
given in this example, ranging from 79 vol% at 135°C (275°F) to
75 vol% at 121°C (250°F).
Large changes in H2O vapor content of flue gases cause only
slight changes in acid dewpoint. Variation of dewpoint with
H2S04 content of gases having different H20 vapor concentrations
is shown in Figure 5, where the range from 0.5% to 15 vol% H20
vapor changes dewpoint only 17°C (30°F) to 22°C (40°F) for the
medium-to-high acid concentrations indicated.
In addition to the procedure based on calculated partial
pressure, a number of efforts have been made to determine H2SO4
dewpoints from instrumental and chemical procedures. Figures
6 and 7 present results obtained for flue gas dewpoints as a
166
-------
I
(-(
a
04
100
80
60 —
40 —
20
10
8
6
2 -
6 8 10
20
40 60 100
R 0 Vapor - vol%
Figure 3. Dewpoint and condensate composition for vapor mixtures
of H-O and H2S04 at 760 mm Hg total pressure [from Abel
(9) and Greenewalt (10)].
167
-------
&
04
n
a
10
o
W
CM
100
80
60
40
20
10
8
6
4
220
240
260
280
300
320 340
Dew Point - °F
Figure 4. H2S04 dewpoint for typical flue gas moisture concentrations
[from Abel (9) and Greenewalt (10)].
168
-------
400
300
-P
G
•H
O
Ai
O
Q
200
100
or content of
ses - %
0.000
0.005
0.010
0.015
0.020
Concentration in Dry Gas - %
Figure 5. Variation of dewpoint with H2S04 content for gases having
different H20 contents [from Matty and Diehl (11)].
169
-------
O
O.
O
CO
CM
S3
60
50
40
30
20
10
8
6
4
3
Muller Calculated
Data Points
10%
Abel and
Greenewalt
10% H^O
Lisle
6.9-9.4%
H20
I I
140 160 180 200 220 240 260 280 300 320
Dew Point - °F
Figure 6. H2S04 dewpoint obtained by various investigators [Abel
(9) and Greenewalt (10), Gmitro (14), Lisle (13), Muller
(8), and Taylor (12)].
170
-------
370
CM
o
4J
C
•H
O
0)
Q
350 —
330
310
290
270
250
230
210
190
Taylor (dew point meter)
Muller (calculated)
Experimental partial
pressure measurements
Ill I till
i i A
l I i I i
0.01
0.1
1.0
10
100
in Flue Gas - ppm
Figure 7. Dewpoint as a function of H2S04 concentration [from
Taylor (12) and Mueller (8)].
171
-------
function of H2SO4 content by various investigators. To make
an exact comparison, all curves should be for a gas of the
same vol% H20 vapor. However, reference to Figure 3 will in-
dicate that a variation in H2O vapor concentration from 7 vol%
to 10 vol% can cause only about 1°C (2°F) to 2°C (4°F) change
in dewpoint. Taylor's (12) results were obtained from an
electrical dewpoint meter which is inaccurate at low acid partial
pressures. Lisle and Sensenbaugh1s (13) data were obtained with
a spiral condenser. Dewpoint curves of Gmitro (14), Muller (8),
Abel (9), and Greenewalt (10) were based on calculated partial
pressures.
In view of the difficulties with calculation based on liquid
phase thermodynamic properties and inaccuracy of dewpoint meters
at low acid partial pressure, the most reliable method of
correlating H2S04 dewpoints with H20 and H2SO4 vapor concentra-
tion is the experimental condensation method employed by Lisle
and Sensenbaugh (13). Their data correlate best with Muller's
calculated dewpoints and are the basis for ASME Power Test Code
19.10.
Rendle and Wilson (15) have also published some data on
relation of the S03 content of combustion gas and of gas dew-
point to sulfur content of fuel oils (Figure 8). Results of
several other investigators have also been plotted. The type of
oil, ash content, and combustion conditions differ for the various
sets of points. Although the plot of SO3 content shows consider-
able scatter, it is apparent that with more than 0.5% sulfur in
the oil, S03 content of the gas does not increase proportionately
to the fuel sulfur %.
Figure 8 indicates the following:
1. There is a rapid initial rise in dewpoint with the
first increment of sulfur in the fuel. For an estimated
dewpoint of 38°C (100°F) with no sulfur, an increase to
127°C (260°F) (H2SO4 dewpoint) is found with 1% sulfur.
2. There is a relatively small rise in dewpoint as
sulfur in the fuel oil increases from 1% to 6%.
EXPERIMENTAL DATA
Hillenbrand et al. (4) sampled flue gas from a coal-fired
boiler with quartz filters maintained at 205°C (400°F) and 138°C
172
-------
<#>
rH
o
01
nJ
O
>i
H
Q
<4-l
O
•P
c
(U
-p
on
O
-P
fi
•H
O
P4
(U
a
0.004
0.003
0.002
0.001
0 0.000
300
200
100
e
X
I
Rendle and Wilsdon
Corbett and Fireday
Flint et al.
Taylor and Lewis
I I I
2345
Sulfur Content of Oil - %
Figure 8. Relation of dewpoint and S03 content of combustion gases
to sulfur content of oil [from Corbett and Fireday (18),
Flint et al. (17), Taylor and Lewis (16), and Rendle and
Wilson (15)] .
. 173
-------
(280°F) and found that filter temperature significantly affected
the amount of H2SO4 found on the filter. At 138°C (280°F),
45% and 41% of the total H2S04 catch (total H2S04 mass in
probe, filter, and impingers) were found on the filter in two
trials. At 205°C (400°F), 24% and 8% of the total H2S04 catch
were found on the filter in two trials. The greater amount of
H2SO4 found on the cooler filter was interpreted by them to mean:
1. A considerable portion of H2S04 collected on the
filter resulted from both condensation and reaction of
particulate with the S02 and S03.
2. Condensation and consequent reaction is favored at
lower temperature.
Jaworowski (3) sampled flue gas from several oil-fired boilers
with three different sampling methods: EPA Method 5 sampling
train, ceramic thimble apparatus, and a high-volume sampling
system. In all three sampling methods, temperature of the filter
was kept between 120°C (250°F) and 150°C (300°F). His results
(Table 2) show the magnitude of H2S04 contribution to total par-
ticulate grain loading ranged from 18% to 78% and averaged 36%
of the total measured emissions.
Experiments with H2S04 aerosol were conducted by Lundgren
and Gunderson (19) at two temperatures [120°C (248°F), 205°C
(401°F)], 25 cm/sec filtration velocity, 8.5 vol% H20 vapor, and
140 ppm H2SO4 , At these concentrations of H2SO4 and H2O vapor,
<,he acid dewpoint is about 170°C (338°F). The results in Table
3 show at 120°C (248°F), below the H2S04 dewpoint, most of the
H2S04 was found in the coil and on the test filter. At 205°C
(401°F), above the H2SO4 dewpoint, most of the H2S04 was found
after the test filter in the impinger contents and on the backup
filter. Simple calculations based on typical stack sampling
data from oil-fired boilers and on these experimental data indicate
H2S04 could account for more than 50% of the total particulate
catch at a 120°C (248°F) sampling temperature (or at a temperature
below the H2S04 dewpoint in the stack gas), but only for about
9% at a 205°C (401°F) sampling temperature (or at a temperature
above the H2S04 dewpoint in the stack gas).
CONCLUSION
A gas such as H2S04 (SO3) can condense at normal sampling
temperatures and condense out on sampling lines or be collected
out by a particulate collection device. It is not improbable
174
-------
Table 2. Amount of H2S04 Found in Particulate
Matter by Various Stock Sampling
Methods [From Jaworowski (3)]
Location
Plant A
Plant A
Plant A
Plant A
Plant B
Plant C
Plant C
H2S04
Filter ppm
Thimblea 8.1
8.1
10.8
9.5
9.9
9.5
9.5
9.5
Hi-volumea 6.9
6.0
8.8
8.1
EPA/APCOa 14.9
13.8
7.5
11.6
6.1
9.5
9.5
8.8
8.8
EPA/APCOa 11.0
9.9
9.9
EPA/APCO 5.0
5.0
6.7
5.7
5.9
5.4
o
Hi-volume 4.7
2.8
EPA/APCOb 3.5
4.2
H2S04
gr/SCF
0.0147
0.0147
0.0196
0.0172
0.0180
0.0172
0.0172
0.0172
0.0125
0.0109
0.0147
0.0050
0.027
0.025
0.0135
0.021
0.011
0.0172
0.0172
0.0159
0.016
0.020
0.018
0.018
0.0092
0.0092
0.0123
0.0104
0.0104
0.0098
0.0085
0.0050
0.0064
0.0076
Total
gr/SCF
0.0694
0.0344
0.107
0.366
0.0645
0.0329
0.0688
0.0405
0.0548
0.0540
0.0255
0.0292
0.151
0.0321
0.0308
0.0388
0.0242
0.0403
0.0643
0.0659
0.033
0.033
0.077
0.0757
0.022
0.023
0.036
0.030
0.031
0.033
0.0111
0.0076
0.0212
0.0200
H2S04 %
of Total
21
43
18
47
28
52
25
42
23
20
62
50
18
78
44
54
45
43
27
24
48
61
23
24
42
40
34
35
34
30
77
66
30
38
a
b
BaCl2 precipitation
NaOH titration
175
-------
Table 3. H S04 Distribution in Sampling Train (19)
120°Ca 205°CŁ
% of Total % of Total
H2S04 Catch0 H2S04 Catch0
S.S. Coil at Test
Temperature
Filter at Test
Temperature
Two Impingers +
Backup Filter
Total
61
32
7
100
8
11
81
100
aH2S04 Dewpoint 170°C
Average of two trials
that this contribution of particulate sulfate is greater than
the "dry" particulate sulfate that exists at "pre-collection"
conditions—if appropriate precautions are not taken.
ACKNOWLEDGMENT
This work was supported by Environmental Protection Agency
Research Grant No. 803126-01-0.
176
-------
REFERENCES
1. Danielson, J. A. Air Pollution Engineering Manual, Los
Angeles County Air Pollution Control District, Los Angeles,
California, 1967, p. 536.
2. Hemeon, W. C. L., and A. W. J. Black. Stack Dust Sampling:
In-Stack Filter or EPA Train. J. Air Poll. Control Assoc.,
22(7):516, 1972.
3. Jaworowski, R. J. Condensed Sulfur: Trioxide Particulate or
Vapor? J. Air Poll. Control Assoc., 23(9):791, 1973.
4. Chemical Composition of Particulate Air Pollutants From
Fossil-Fuel Combustion Sources. Battelle-Columbus Labora-
tories, p. II-2, March 1, 1973.
5. Hedley, A. B. In: The Mechanism of Corrosion by Fuel
Impurities, H. R. Johnson and D. L. Littler, eds. Butterworth,
London, 1963, p. 204.
6. Cuffe, S. T., R. W. Gerstle, A. A. Orning, and C. H.
Schwartz. J. Air Poll. Control Assoc., 14:353, 1964.
7. Snowden, P. N. , and M. H. Ryan. Sulfuric Acid Condensation
from Flue Gases Containing Sulfur Oxides. J. Inst. Fuel,
42:188, 1969.
8. Mueller, P. Study of the Influence of Sulfuric Acid on the
DewPoint Temperature of the Flue Gas. Chemie-Ing.-Tech.,
31:345, 1959.
9. Abel, E. The Vapor Phase Above the System Sulfuric Acid-
Water. J. Phys. Chem., 50:260, 1946.
10. Greenewalt, C. H. Partial Pressure of Water Out of Aqueous
Solutions of Sulfuric Acid. Ind. and Eng. Chem., 17:552-553.
11. Matty, R. E. , and E. K. Diehl. New Methods for Determining
SO2 and SO
Dec. 1953.
SO2 and S03 in Flue Gas. Power Engineering, 57:87,
12. Taylor, A. A. Relation Between Dew Point and the Concentra-
tion of Sulfuric Acid in Flue Gases. J. In^t. Fuel, 16:25,
1942.
177
-------
13. Lisle, E. S., and J. D. Sensenbaugh. The Determination of
Sulfur Trioxide and Acid Dew Point in Flue Gases.
Combustion, 36(1):12, 1965.
14. Gmitro, J. I., and T. Vermuelen. Vapor-Liquid Equilibria
for Aqueous Sulfuric Acid. Univ. of Cal. Radiation Lab.,
Report 10866, Berkeley, California, June 24, 1963.
15. Rendle, L. K., and R. D. Wilson. The Prevention of Acid
Condensation in Oil-Fired Boilers. J. Inst. Fuel, 29:372-
380, 1956.
16. Taylor, R. P., and A. Lewis. Sulfur Trioxide Formation in
Oil Firing. In: Proceedings of Fourth Inst. Congress on
Industrial Heating, Group II, Sec. 24, No. 154, Paris,
France, 1952.
17. Flint, D., A. W. Lindsay, and R. F. Littlejohn. The
Effect of Metal Oxide Smokes on the S03 Content of Com-
bustion Gases from Fuel Oils. J. Inst. Fuel, 26:122-127,
1953.
18. Corbett, P. F., and F. Fireday. The S03 Content of the
Combustion Gases from an Oil-Fired Water-Tube Boiler. J.
Inst. Fuel, 26:92-106, 1953.
19. Lundgren, D. A., and T. C. Gunderson. Filtration Char-
acteristics of Glass Fiber Filter Media at Elevated
Temperatures. EPA-600/2-76-192, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina,
July 1976, pp. 13-72.
178
-------
Sulfur Oxide Interaction with Filters Used for
Method 5 Stack Sampling
Edward T. Peters
Jeffrey W. Adams
Arthur D. Little, Inc.
ABSTRACT
An experimental program was conducted to study the
conditions under which sulfur oxide-containing gases can
interact with high efficiency filters to form "false"
particulate, e.g., particulate formed as a result of the
sample collection process. Evaluation filters were
exposed to particulate-free gas streams containing air,
water, sulfur dioxide and/or sulfuric acid vapor at
elevated temperatures. After exposure, the filters were
leached in hot water which was analyzed for sulfate
content. Filters tested included two quartz and seven
glass compositions corresponding to four classes - high
titania, medium titania, high barium, and borosilicate.
Sulfate pickup on clean and on Mn*2 spiked filters
exposed to streams containing 500 ppm or 2000 ppm S02
and up to 25 volume percent moisture at 205°C was less
than 1 mg/m3, comparable to sulfate blank analyses for
these filters. However, exposures at 205°C with sulfuric
acid vapor at 10 ppm to 40 ppm and moisture contents
up to 25 volume percent (acid moisture dewpoint of 120°-
150°C) lead to collection of 1-40 mg/m3 sulfate, depend-
ing upon steam parameters and stream concentration. As
a result, a series of exposures (in triplicate) was
carried out with all commonly used stack sampling filters;
stream conditions included 10 ppm sulfuric acid vapor
and 10% to 15% moisture content at 200°C in simulation
of a typical combustion source. Levels of 1 to 14 mg/m3
were encountered. The results and implication to Method
5 stack sampling will be given.
179
-------
INTRODUCTION
General
In response to the provisions of the Clean Air Act of 1970,
the Environmental Protection Agency has identified stationary
sources categories that contribute significantly to air pollution.
Standards of performance applicable to new and modified sources
within these industrial categories have been promulgated for twenty-
four classifications and have been proposed for more. These
regulations provide emission limits for various pollutants and
specify the reference methods that are to be utilized in determining
compliance with the standards.
At present, fifteen of the stationary source categories are
regulated with respect to particulate emissions. As defined,
particulate matter is "any matter, other than uncombined water,
which exists in a finely divided form as a liquid or solid at
standard conditions (20°C, 760 mm Hg)." The promulgated standards
specify the use of Reference Method 5 to measure the amount of
particulate matter in stack emissions. In application, a known
volume of stack gas is withdrawn isokinetically from the source,
and particulate material is collected in the sampling train by
out-of-stack filtration. The particulate collection filter is
maintained at a temperature above the dewpoint of water to prevent
condensation and filter plugging. Particulate weight is deter-
mined gravimetrically after removal of uncombined water.
The objective of stack sampling is to provide a measure of
the particulate burden to the atmosphere as it would exist in the
dispersed plume. Each industrial category exhibits variations in
the properties of the emission, including composition, concentra-
tion, gas temperature, residence time, and moisture content. Thus,
it is necessary to determine if the assay procedure is influenced
by variations in stack conditions. Also, it must be established
if the formation or loss of particulate occurs due to the nature
of the collection process or the configuration of the sampling
train. Mechanisms for particulate formation during sampling
include chemical reaction of stack gas components with collected
particulate or sampling train components, catalytic conversion
and condensation of stack gases, and compound hydration. Such^
particulate that would not have formed in the atmosphere at 20°C
and 760 mm Hg is called false particulate.
With respect to these issues, the EPA has sponsored several
studies to evaluate Method 5 particulate measurement as applied
to specific industrial categories. The present report describes
180
-------
a laboratory evaluation of the interaction of sulfur oxides with
particulate collection filters.
Interaction of Sulfur Oxides and Particulate Collection Filters
The extraction of a representative sample from an elevated
temperature gas stream containing flyash or other process particu-
late, sulfur oxide constituents, water vapor, and many other chemical
species has posed a burdensome problem to chemists and engineers
since the inception of standard sampling methodology. Uncertainties
regarding the chemistry of the sample within the sampling train,
particularly along the length of the probe and across the filter,
have often resulted in doubt being placed upon the results obtained
from these analyses, leading to the following kinds of questions:
• Are chemical interactions involving transformation from
gas phase to particulate taking place on the filter that
might influence the measured particulate concentration?
• Does the moisture content, temperature, or chemical
composition of the sampled gases influence the results
or sampling procedure in any way?
The possibility of forming false particulate (particulate
matter formed as a result of the collection process) is of specific
concern, as this would result in an overestimate of the true
particulate emissions for a source. Such an overestimate could
cause an undue burden upon that source in terms of satisfying
compliance requirements.
Of special interest are sources containing considerable amounts
of sulfur oxides in the stack gas. These sources include fossil
fuel combustion and product recovery processes, such as petroleum
refinery catalyst regeneration and Kraft pulp mill black liquor
recovery. Up to 5% or more of the sulfur in these emissions can
be present as S03 which, in equilibrium with water vapor, exists
as sulfuric acid vapor. The H2S04 - H20 dewpoint chart is given
in Figure 1. As temperature is reduced from stack conditions,
generally in excess of 150°C (300°F) for these sources, to 120°C
(250°F), the temperature maintained in the --ollection train,
essentially all sulfuric acid in the flue gas is condensed and
collected as particulate. No information is awilable, however,
relating to the reaction of sulfuric acid mist vapor with glass
fiber filters, as could occur with a particulate sampling train
using an in-stack filter. Further, there are no data relating to
the oxidation of S02 to S03 during sampling with collection as
181
-------
220
240
260
280
300
320
Dew Point, F
Figure 1. H2S04 dewpoint for typical flue gas moisture
concentrations.
182
-------
sulfate through direct reaction with the filter media or catalytic
conversion by components in the collected particulate.
To experimentally determine the presence and extent of these
types of false particulate formation, a two-part laboratory investi-
gation was carried out. Part one concentrated on evaluating
stream parameters, including sulfur oxide components, their
concentrations, moisture content, and temperature. These streams
were passed through both clean filters and filters treated with a
metal salt to simulate a deposit with cationic properties. Subse-
quently, part two was carried out to evaluate a variety of flat
glass fiber filters under worst-case stream conditions to determine
the relationship between glass chemistry and formation of false
particulate.
PROGRAM PLAN
The objective of the study was to develop the background and
understanding necessary to evaluate the relationship between
filter/gas stream chemistry and false particulate formation.
Specifically, it was hoped to establish the conditions under which
the filter media, gas stream conditions, or combinations of both
exhibit reactive or catalytic properties contributing to the
collection of sulfur oxide or sulfuric acid vapor as measurable
particulate.
To provide best control of experimental parameters, it was
decided to carry out filter challenge studies in the laboratory.
The experimental plan required the generation of a dust free, SOx
enriched gas stream and the extraction of a known volume of this
stream through test filters with subsequent analyzers of the filters
to determine the presence and amount of collectable sulfate. The
experimental parameters to be tested were selected on the basis of
conditions typically encountered in source sampling. Parameters
of specific interest and selected for evaluation were:
• Gas stream/filter temperature - 120°C (250°F), typical
collection train temperature, and 205°C (400°F), typical
stack gas temperature.
• Gas stream moisture content - dry and 25%.
• Sulfur dioxide concentration - 500 ppm and 2000 ppm,
typical for low and high sulfur fossil fuel combustion,
respectively.
183
-------
• Sulfuric acid content - 10 ppra and 40 ppm, representing
about 2% of the SO2 concentration.
• Presence of a filter deposit - Does the presence of
certain metal ions in the deposit result in the catalytic
conversion of S02 to collectable sulfate? Filters were
immersed in a solution of MnCl2 at a concentration
designed to provide loadings of 30 /ig Mn+2 per filter.
• Filter media - Preliminary studies considered several
general classes of filters, including flat glass fiber
commercial filters (MSA 1106BH and Gelman A), an
experimental flat quartz filter (ADL/Balston Microquartz),
and a glass fiber thimble suitable for in-stack collection
(Svenska Flakt Jabrikan thimble). Later (Part 2) studies
utilized a variety of commercial flat glass filters and
both commercial and experimental flat quartz filters.
The initial (Part 1) set of experiments was carried out in
three phases, involving the following types of stream challenges:
Phase 1 - Sulfur dioxide - clean filter
Phase 2 - Sulfuric acid vapor (alone or with sulfur dioxide) -
clean filter
Phase 3 - Sulfur dioxide (alone or with sulfuric acid vapor) -
soiled filter containing Mn"1"2.
These phases were carried out to address the following types
of questions:
Phase 1 - Is there a reaction between sulfur dioxide and a clean
filter resulting in collectable particulate and for what stream
conditions? The experimental design is shown in Table 1 which
also indicates some additional evaluations carried out at a
subsequent time or with a different filter media.
Phase 2 - Is sulfuric acid collected under stream where it is in
the vapor state and does the co-existence of sulfur dioxide have
an influence? The experimental plan for these evaluations is
given in Table 2.
After this series of runs, a reaction between the gas stream
components and the stainless steel parts of the stream generation
apparatus resulted in system pluggage. Therefore, all stainless
was replaced by Teflon or glass fittings.
184
-------
Table 1. Phase I Test Grid: Sulfur Dioxide - Clean
Glass Fiber Filter Interaction
A. Statistical Experiment Grid
Sulfur Dioxide
Content
500 ppm
2000 ppm
Moisture
Content
0
25%
0
25%
Gas Stream/Fi
120°C
X X
X X
X X
X X
Iter Temperature
205°C
X X
X X
X X
X X
x - Simultaneous exposure of MSA1106BH and SFJ Thimble filters
B. Additional Evaluations (Gas Stream/Filter, Temperature = 205°C)
Gas Concentration
Filter 500 ppm S022000 ppm SO22000 ppm S02
25% H20 0% H20 25% H20
SFJ Thimble x x x x
MSA 1106BH y x x
ADL/Balston MQ x x x
x - Evaluations carried out in stainless/glass apparatus
y - Evaluations carried out in Teflon/glass apparatus
185
-------
Table 2. Phase 2 Test Grid: Sulfuric Acid Vapor - Clean Glass
Fiber Filter Interaction with and without Sulfur Dioxide
(Gas Stream/Filter Temperature - 205°C, Moisture Content = 25%)
Gas Concentration
Filter
500 ppm S02
10 ppm H2SO4
2000 ppm S02
40 ppm H2S04
40 ppm H2S04
SFJ Thimble
MSA H06BH y y y
Gelman A
ADL/Balston MQ
X X X X
xxx x, yyyy
y y
y y
X
X
y
X
X
y
X
X
y
x - Evaluations carried out in stainless/glass apparatus
y - Evaluations carried out in Teflon/glass apparatus
Phase 3 - Does the presence of a metal catalyst on the filter
result in the conversion of sulfur dioxide (alone or in the
presence of H2SO4 vapor) to collectable sulfate?
For this purpose, Mn+2 was selected as being representative
of active metal catalysts for S02 conversion to SO3 (and, in the
presence of moisture, to H2S04). The experimental plan for
evaluating the interaction of sulfur oxide-containing streams
with Mn+2 spiked filters is given in Table 3.
Subsequent to Part 1 experiments, a second series of
experiments was planned to determine the influence of glass fiber
composition and filter production variables on reactivity with
sulfur oxide-containing streams. A total of nine filter materials
was selected on the basis of glass composition (determined by X-ray
emission analysis), manufacturer, and usage for stack sampling.
The candidate filters, together with classification according
to composition, are identified in Table 4 which also presents
the relative x-ray interaction for major elements (excluding Si)
and sulfate levels. The experimental test grid for the Phase 4
studies are given in Table 5.
186
-------
Table 3. Phase 3 Test Grid: Sulfur Dioxide
and Sulfuric Acid Vapor - Mn+2 Spiked Glass Fiber
Filter Interaction (Gas Stream/Filter Temperature
= 205°C, Moisture Content = 25%)
Filter
SFJ Thimble
MSA 1106BH
Gelman A
ADL/Balston
Gas Concentration
500 ppm S02 2000 ppm SC>2 2000 ppm S02
10 ppm H2SO4 40 ppm H2SC>4
XXX XXX
yyy xxx, yyy xxx
y y y
MQ y y y y
x - Evaluations carried out in stainless/glass apparatus
y - Evaluations carried out in Teflon/glass apparatus
187
-------
Table 4. Identification of Filters: Qualitative Elemental Analysis and
Extractable Sulfate Content
Filter Type
Quartz
ADL/Balston Microquartz
Gelman Microquartz
High Titanium
Reeve Angel 934AH
Gelman A
High Barium
Watman A
SFJ Thimble
Borosilicate
Gelman Spectrograde
Schleicher & Schuell
810
Reeve Angel 900AF
MSA 1106BH
Gelman AE
Relative X-ray Intensity
Ca
300
1500
33,600
41,800
12,500
12,800
19,900
23,800
24,400
25,300
26,800
K
30
180
320
520
7100
7000
2600
3100
3300
3500
3600
Al
20
30
30
0
20
20
20
40
30
30
20
Ti
190
70
4200
2800
90
70
140
80
90
Fe
5
5
40
35
20
20
10
20
20
20
20
S
0
0
20
20
120
100
290
70
. 65
60
160
Cl
15
35
30
0
0
0
10
80
75
60
35
Amount Sulfate
(mg per filter)
0-0.2
0
0.6
0.2-0.35
0.15
0-0.6
NA
0.5
0.5
0.4-1.0
0.15
-------
Table 5. Phase 4 Test Grid: Filter Reaction with S02 + H2S04 Vapor
Containing Gas Stream (Gas Stream/Filter Temperature = 205°C,
Moisture Content = 10%, S02 = 2000 ppm, H2S04 = 10 ppm)
Run No.
102
103
104
105
106
107
108
109
110
1
A
RA
Q
M
ss
AE
GQ
K
GF
Filter Position
2
SS
AE
Q
RA
M
GQ
A
GF
K
3
AE
A
RA
Q
SS
GQ
M
GF
K
Q = ADL/Balston Microquartz
GQ = Gelman Microquartz
A = Gelman A
AE = Gelman AE
K = Reeve Angel 934AH
RA = Reeve Angel 900
M = MSA 1106BH
GF = Whatman GF/A
SS = Schleicher and Schuell 810
189
-------
EXPERIMENTAL METHODS
FilterExposure Apparatus
The designed program required the construction of a system for
generating sulfur oxide-containing streams, exposing test filters,
and recovering sulfur oxides from the filtered gas stream within
the following design requirements.
Gas stream and filter temperature 120° and 205°C
Flow rate ~2 m3/hr
Moisture content 0-25%
SO2 content 0, 500 ppm, 2000 ppm
H2S04 content 0, 10 ppm, 40 ppm
A schematic sketch of the exposure system is given in Figure 2.
Metered amounts of compressed air and water are passed through
an evaporation coil and into a furnace maintained at the desired
test temperature. After introducing S02, the stream is passed
through a mixing tank. Sulfuric acid mist is then injected into
the stream at a controlled rate by means of a syringe drive.
During stabilization of conditions, the entire stream is exhausted
with a portion of the stream passing through an NDIR gas analyzer
to monitor SO2 levels. For exposure, the pumps on the sampling
trains are activated, and a portion of the challenge gas stream is
pulled through the filters at a rate of about 0.05 m3/min., with
excess gas being exhausted. The sampling trains were commercial
versions of an EPA approved particulate sampling train with the
following modifications: 1) the heated sampling probe and nozzle
were excluded, and 2) a condenser was incorporated between the
filter holder and first impinger to recover the sulfuric acid
mist (and water) from the stream. The sampling train impingers
were charged with 10% hydrogen peroxide to collect S02.
The original stream generation apparatus was constructed with
stainless steel tubing and glass. However, during the Phase 2 and 3
evaluations involving the introduction of sulfuric acid mist, very
poor sulfuric acid mist recoveries from the system were noted.
Inspection of the system revealed substantial corrosion of the
stainless in the vicinity of sulfuric acid injection. Therefore,
the system was reconstructed with glass tubing and Teflon fittings.
The SFJ thimble filter could not be tested during this series of
evaluations, as it required the use of a stainless steel holder.
The apparatus was reconstructed for the Phase 4 challenge experiments
190
-------
-D
Heater
^ twft 1^
«Q-p4 — •—
laust
inge Driv
»^
-v
a
e
1 r
>
\
\
(x
s
V,
\
V
\
s.
s,
\NNN\\NNNN\-
TCI
1
1
L_
r
Mixing
Tank
V V N.
Fl
.F2
—
h><-
1
—V*rm"
T
j
C2
1
M~»
h^l
U. T~,—.
' U"M
\\\\\\x
^'
\
\ \>
\
\
S
\
\
N
\
\
\
\
\
N
\
\
\
\
\
\|
— ^ Trains
^Wll 1
Exhaust
Furnace
Figure 2. Schematic sketch of filter exposure apparatus.
-------
to achieve a gas stream flow of about 0.15 ra3/min. enabling the
simultaneous exposure of three filters.
Characterization and Preparation of Test Filters
All candidate filter materials were analyzed by x-ray fluore-
scence to determine the relative amounts of major elements, permitting
a classification of filters according to glass composition. In
addition, x-ray fluorescence was used to make a qualitative estimate
of the presence and amount of sulfur and chlorine associated with
the filters. Subsequently replicate samples of all filters
were analyzed for background sulfate levels, using the barium
chloranilate colorimetric procedure.
Prior to gas stream exposure, all filters were pretreated by
heating to 500°C for one hour. Filters selected for spiking with
Mn"4"2 were soaked in a solution of MnClj and oven dried. Severj.1
spiked filters of each type were analyzed for Mn"1"2 content by atomic
absorption to establish that doping levels were within the desired
range of ~30 **g per filter.
Sample Analysis
At the completion of each filter exposure experiment, the
following individual samples were obtained from each of the collec-
tion trains:
• Filter
• Condenser solutions and rinsings (for experiments with
H2SO4 injection)
• Impinger solutions and rinsings
• Silica gel
During clean-up, the water gain, attributable to moisture
condensation, was measured and recorded in accordance with Method 5
protocol for moisture content determination. The filters were
removed from their holders and stored in individually sealed
petri dishes until such time as sulfate determinations could be
performed. After recovery, all train solutions were immediately
analyzed by wet chemical techniques to quantify the level of SO2
and/or H2SO4 and variable amounts of SC^ entrained within the
dropout liquor. To determine this entrained level, an aliquot of
the condenser solution was analyzed as sulfite by an iodimetric
192
-------
titration procedure. A second aliquot was treated with peroxide
to oxidize the sulfurous acid to sulfuric acid. This aliquot was
then titrated with standard base to determine total acid content.
The value obtained by the iodimetric procedure was then subtracted
from the total acid value to determine the amount of H2S04 collec-
ted in the condenser.
The impinger solutions, which initially had been charged with
7%-10% hydrogen peroxide solutions, were titrated directly with
a standardized base for quantitative determination of 862. This
value was added to the sulfite value previously discussed to cal-
culate the total S(>2 content of the sampled gas stream.
Soluble sulfate was extracted from the collection filters
by extraction in boiling water. The amount of sulfate was then
determined by the barium chloranilate method.
RESULTS AND DISCUSSION
Phase 1 - Sulfur Dioxide - Clean Filter Evaluations
The exposure of clean filters to streams containing sulfur
dioxide did not result in the collection of significant amounts
of sulfate. For the statistical grid of experiments considering
variation in SC>2 content, moisture content and gas temperature,
the measured sulfate values (given in Table 6) averaged 1.3 mg/m
for the MSA 1106BH filters and 0.3 for the SFJ thimbles. These
values are generally in the range observed for the sulfate con-
tent on clean filters.
In summary, the Phase 1 filter challenge experiments involv-
ing variation in S02 concentration, moisture content, and filter
temperature resulted in very low sulfate recoveries from the
tested filters. These recoveries were not substantially different
than blank sulfate levels for SFJ thimble and AD/Balston MO
filters and were less than one mg/m3 higher than the blank for the
MSA 1106BH filter.
193
-------
Table 6. Results of Phase 1 Sulfur Dioxide Challenge
Studies: Statistical Experimental Grid
A. MSA 1106BH
Sulfur Dioxide
Content
500 ppm
2000 ppm
B. SFJ Thimble
Sulfur Dioxide
Content
500 ppm
2000 ppm
Moisture
Content
0
25%
0
25%
Moisture
Content
0
25%
0
25%
Sulfate Gain
Gas Stream/Filter
120°C
0.7, 0.9
1.1, 1.4
1.0, 0.9
1.8, 1.4
Gas Stream/Filter
120°C
0.6, 0.2
0.6, 0.5
0.3, 0.2
0.5, 0.7
- mg/m3
Temperature
205°C
0.8, 0.8
1.7, 1.0
1.3, 1.3
1.7, 2.3
Temperature
205°C
0.3, 0
0.2, 0.2
0.2, 0
0.4, 0.1
194
-------
Phase 2 - Sulfuric Acid Vapor - Clean Filter Evaluations
Filters exposed to gas streams containing sulfuric acid
vapor (in equilibrium with water), sometimes in coexistence with
S02, were found in some cases to collect appreciable amounts of
sulfate at a test temperature well above the acid dewpoint, i.e.,
as false particulate. The data for individual experiments are
given in Table 7.
For the Teflon/glass system exposures, measurable sulfate
is collected by all test filters after exposure to a stream con-
taining 40 ppm H2S04 vapor (corresponding to about 160 mg/m3)
and 2000 ppm S02. The level of sulfate is appreciable for MSA
1106BH, being about 40 mg/m3, with lower levels of 7 and 2 mg/m3
for Gelman A and ADL/Balston Microquartz, corresponding to 25%,
4%, and 1% of the H2S04 vapor in the stream, respectively. The
data for the ADL/Balston Microquartz show equivalent sulfate
gains for exposure with and without the coexistence of 2000 ppm
S02. Therefore, in agreement with the Phase 1 results, S02 is
not being converted to collectable sulfate. For the MSA 1106BH
filter, comparable levels of sulfate (about 40 mg/m3) are
collected from streams with 10 ppm H2S04 and 40 ppm H2S04. This
suggests an upper limit to H2SC>4 vapor collection which, for MSA
1106BH, is about 40 mg/m3 corresponding to a stream concentration
of 10 ppm H2S04.
Phase 3 - Sulfur Dioxide and Sulfuric Acid Vapor -
Mn+2 Spiked Filter Evaluations
The major purpose of this series of experiments was to deter-
mine whether the presence of an appropriate catalyst for S02
oxidation on the filter results in sulfate collection. The
experimental results for individual experiments are presented
in Table 8. Experiments performed with the stainless/glass
system yield sulfate values that indicate an influence from
system memory. No conclusions can be drawn from these data.
For the case of evaluations in the Teflon/glass system, sulfate
recoveries are generally comparable to the results for Phase 2
studies for the same conditions of gas composition. It is con-
cluded that the presence of Mn"1"2 on the filter does not result
in the conversion of SO2 to collected sulfate.
195
-------
Table 7. Results of Phase 2 Sulfuric Acid Vapor -
Clean Filter Challenge Studies (Gas Stream/Filter
Temperature = 205°C, Moisture Content = 25 volume percent)
Gas Concentration
-°
Stainless/Glass
System
SFJ Thimble
MSA 1106BH
ppm SO2 2000 ppm SO2
ppm H?SO4 40 ppm HjSC^ 40 ppm H2S04
Sulfate Gain - mg/m
2.4, 8.2, 2.4, 3.3 20.5, 12.6, 44.7
6.1, 6.8, 4.2, 7.0 21.4, 15.8, 40.8
Teflon/Glass
System
MSA 1106BH
Gelman A
ADL/Balston MQ
45.7, 47.2, 47.3 54.0, 47.1, NA, 17.4
6.5, 7.7
2.2, 1.7
Stainless/Glass
System
SFJ Thimble
MSA 1106BH
1.4, 2.0, 2.1
Average Sulfate - mg/m (Std. Dev. - mg/m )
4.1 (2.8)
6.0 (1.6)
25.9 (16.7)
26.0 (13.1)
Teflon/Glass
System
MSA 1106BH
Gelman A
ADL/Balston
46.7 (0.8)
40.2 (19.9)
7.0 (0.9)
2.0 (0.4)
2.3 (0.5)
196
-------
Table 8. Results of Phase 3 Sulfur Dioxide and Sulfuric Acid
Vapor - Mn+2 Spiked Filter Challenge Studies (Gas Stream/Filter
Temperature = 205°C, Moisture Content = 25 volume percent)
Filter
Stainless/Glass
System
SFJ Thimble
MSA 1106BH
Gas Concentration
500 ppm S02 2000 ppm S02
10 ppm H2SO4 40 ppm H2S04
Sulfate Gain - mg/m3
12.0, 7.3, 5.8
13.7, 15.2, 11.7
2000
ppm SO 2
14.2, 10.0, 5.4
12.3, 11.9, 15.9
Teflon/Glass
System
MSA 1106BH
Gelman A
ADL/Balston MQ
35.8 15.8, 40.7 48.0, 69.0, 38.6
3.7, 3.6, 3.7
0, 0.8 2.9, 2.7
Stainless/Glass
System
SFJ Thimble
MSA 1106BH
Average Sulfate - mg/m3 (Std. Dev. - mg/m3J
8.4 (3.2)
13.5 (1.8)
9.9 (4.4)
13.4 (2.2)
Teflon/Glass
System
MSA 1106BH
Gelman A
ADL/Balston MQ
30.8 (13.2)
3.7 (0.1)
0.4 (0.6)
51.9 (15.6)
2.8 (0.1)
197
-------
Phase 4 - Sulfur Dioxide and Sulfuric Acid Vapor -
Clean Filter Evaluations
Phase 1, 2, and 3 studies gave evidence that varying amounts
of sulfuric acid vapor were collected on filters as false parti-
culate. The amount collected was apparently associated with the
glass composition of these filters and was highest for MSA 1106BH,
a borosilicate glass. As a result, a variety of filter materials
was evaluated to identify their composition and to determine
their susceptibility to sulfuric acid vapor collection. The
filters included all commercial grades used for stack sampling
and a high purity quartz filter (ADL/Balston Microquartz)
developed under EPA Contract No. 68-02-0585. These filters are
identified in Table 4.
The ADL/Balston Microquartz material is from a pilot scale
run using high purity quartz fibers. Produced on special stain-
less steel papermaking equipment, it represents the highest
purity filter material available. The Gelman Microquartz includes
5% glass fiber (to permit a lower annealing temperature), result-
ing in higher levels of Ca, K, and Cl. All of the other filters
are prepared from glass fibers. The Reeve Angel 934AH and Gelman
A represent high titania glasses of somewhat different composition.
The Reeve Angel 934AH should be more refractory (i.e., better
thermal and chemical stability at elevated temperature than the
Gelman A as a result of a higher Ti and lower Ca and K content).
The Whatman A and SFJ thimble filters are prepared from the same
composition glass, rich in Ba, Zn, and K. This composition is less
refractory than the titania glasses or quartz but more refractory
than the borosilicate composition found for the other five filters.
The results of triplicate exposures of these filters (ex-
cluding the SFJ thimble) to a gas stream at 200°C and containing
2000 ppm S02, 10 ppm H2S04 vapor, and 10%-15% moisture are presented
in Table 9. As expected from the chemical and thermal inertness
of the classes of filters, inferred from the glass compositions,
the ADL/Balston Microquartz shows no effect from stream exposure.
The Gelman Microquartz and Reeve Angel 934AH materials are very
inert, with collected sulfate levels of 0.5 mg/m3, perhaps slightly
above background levels. The Whatman GF/A and Gelman A filters
exhibit a moderate amount of sulfate collection, corresponding
to 7-8 mg/m3, and the four borosilicate composition filters
tested collect appreciable levels of sulfate, corresponding to
13-14 mg/m3.
The collected sulfate on these filters is false particulate,
for the challenge gas stream was maintained at 200°C, well above
198
-------
Table 9. Results of Phase 4 Sulfur Dioxide -
Sulfuric Acid Mist Challenge Studies (Gas
Stream/Filter Temperature = 205°C, Moisture
Content = 10%, S02 = 2000 ppm, H2S04 = 10 ppm)
Sulfate Gain - mg/m
Filter
Ave. Std. Dev.
ADL/Balston Microquartz
Gelman Microquartz
Reeve Angel 934AH
Whatman GF/A
Gelman A
Gelman AE
MSA 1106BH
Reeve Angel 900
Schleicher and Schuell 810
0
0.6
0.5
6.3
7.0
14.4
12.7
13.3
15.2
0.2
0.5
0.5
6.8
7.8
9.6
13.3
12.8
13.7
0.2
0.5
0.5
7.3
8.5
15.0
11.9
15.6
14.3
0.1
0.5
0.5
6.8
7.8
13.0
12.6
13.9
14.4
0.1
0.1
0
0.5
0.8
3.0
0.7
1.5
0.8
the sulfuric acid-water vapor mixture dewpoint of 135°C. As the
sulfur oxide content of the test gas stream is not unlike stack
emission streams for a number of stationary source categories, the
choice of the filter material used in stack sampling can have a
profound effect on the apparent level of particulate.
The nature of the reaction between sulfuric acid vapor and the
components of borosilicate and other glasses is not known. It is
known that components of the glass will partially hydrolyze in the
presence of water vapor, yielding acidic boron oxide and alkaline
compounds of sodium and potassium. It is probable that these
alkaline products react with the acid gas to form collectable
compounds. The amount of collected sulfate, therefore, depends
both on stream conditions (temperature, moisture content, H2S04
content) and on the presence, form, and amount of alkaline metals
in the glass.
Further tests of this nature should be conducted to determine
more completely the conditions under which false particulate can
occur and the amounts to ':>e expected for industrial source cate-
gories. Such information would provide a data base for determining
the advisability of excluding certain classes of filters for use
in stack sampling. In addition, other glass components of the
sampling system (probe liner, filter holder, and connectors), which
199
-------
are usually borosilicate, should be similarly evaluated to deter-
mine if they contribute to false particulate formation.
CONCLUSIONS
Particulate collection filters were exposed to gas streams
containing combinations of sulfur dioxide, sulfuric acid vapor,
and water at temperatures well above the acid-water dewpoint to
identify conditions leading to the collection of sulfate on the
filters. The presence of such false particulate would result in
an overestimate in mass for particulate stack sampling. From this
work, the following conclusions are drawn:
• Streams containing S02 at concentrations of 500 ppm and
2000 ppm did not result in collected sulfate on either
clean filters or on filters spiked with Mn+2.
• Streams containing sulfuric acid vapor (with or without
the coexistence of S02) resulted in the collection of
varying amounts of sulfate depending upon the filter
type.
• Evaluation of commercial filters commonly used for
stack sampling showed that the susceptibility to sul-
fate collection as false particulate is directly re-
lated to the glass composition of the filter. Typical
levels of sulfate collected by exposure to a gas stream
at 200°C (2000 ppm S02, 10 ppm H2S04, 10% moisture)
are:
Sulfate Grade Commercial Filters
(mg/m )
<1 Quartz ADL/Balston Microquartz,
Gelman Microquartz
High titania Reeve Angel 934AH
6-7 Medium titania Gelman A
High barium Whatman GF/A
13-14 Borosilicate Gelman AE, MSA 1106BH,
Reeve Angel 900, Schleicher
and Shuell 810
200
-------
RECOMMENDATIONS
The magnitude of false particulate collected as sulfate by
some commercial filters indicates the need for a better under-
standing of the mechanism for sulfate collection. Further study
should be conducted to address the following kinds of questions.
• What is the range of conditions (gas temperature,
sulfuric acid concentration, moisture content) for
which sulfate will be collected as false particulate?
• Is there a threshold level, after which no further
sulfate will be collected?
• Do other borosilicate glass components of a sampling
train (probe liner, filter holder, connectors) similarly
involve collection of sulfate as false particulate?
• Can the level of sulfate collected be altered by filter
pretreatment? For example, a high temperature exposure
of borosilicate glass filters to steam may remove the
reactive components.
• How does the presence of other particulate influence
sulfate collection?
Such further work is needed to provide a rational basis for
deciding if certain classes of filters should be excluded for
purposes of stack sampling.
ACKNOWLEDGMENTS
The authors acknowledge the help of Larry Damokosh and Dr.
Judy Harris for carrying out the sulfate analyses. Appreciation
is extended to Dr. Kenneth Knapp of the EPA for his suggestions
and comments on experiment planning and review.
201
-------
Particulate Sampling in Process Streams in the
Presence of Sulfur Oxides
Kenneth M. Gushing
Southern Research Institute
ABSTRACT
Particulate sampling methods are performed on industrial
process streams to determine total mass concentrations
and particle size distributions. In some instances, the
amount of sulfur collected may be of primary interest;
however, sulfur oxide environments can also mask and in-
terfere with the determination of total particulate con-
centrations and size distributions.
Most sampling techniques incorporate some type of filter
material on which the particles are collected, either by
filtration or impaction. Recent laboratory and field
data concerning interference and weight gains by filter
materials in sulfur oxide environments will be presented,
including data showing a correlation between type of
filter material, total filter weight gains, and the
specific laboratory or field environment (concentration
of sulfur oxides, temperature, etc.). A brief discussion
will focus on a passivation technique for glass fiber
filters.
Sampling of ultrafine particles is usually performed with
some type of extraction-dilution system. It has been ob-
203
-------
served that under certain combinations of SOX concentra-
tion and dewpoint, fine sulfuric acid mists can be form-
ed. Data related to this phenomenon will be reported.
INTRODUCTION
During the past six years Southern Research Institute personnel
have performed many particle sizing research tests on control de-
vices treating industrial process streams. In many cases these
process streams have had high sulfur dioxide fractions, and this has
led to difficulties in obtaining accurate and reliable data. In
this paper two specific problems are addressed: SO2 uptake by glass
fiber filter media and SO3/H2SO4 masking of ultrafine particle measure-
ments .
SO2 UPTAKE BY GLASS FIBER FILTER MEDIA
In order to determine the particle size distribution of material
entering or exiting a control device, cascade impactors have been
widely employed because of their simplicity of operation, size dis-
crimination range (0.5-20 /am), and ability to operate in situ. In
order to reduce particle bounce and reentrainment in cascade impactors,
as well as to provide a. lightweight media on which to collect milli-
gram quantities of dust, glass fiber substrates are frequently used
in most commercial devices. Unfortunately, these glass fiber sub-
strates are susceptible to weight gains due to reaction of the sulfur
dioxide with basic sites in the glass fibers.
Prior to Southern Research Institute's investigations into this
problem, two previous experimental programs dealt with SO2 uptake on
glass fiber materials.
The work of Charles Gelman and J. C. Marshall (1) of the Gelman
Instrument Company, makers of various filter media and equipment, in-
dicates that SO2 absorption is the cause of the anomalous mass gains.
They acknowledge that a high pH glass fiber can absorb sulfur dioxide
and thus cause erroneously high particulate weights.
According to Gelman and Marshall, the SO2 reaction on glass fi-
bers could cause "a 30% error in the measurement of total suspended
particulate matter" in an urban atmosphere. It is possible that flue
gases would give even higher errors, especially if the gases have a
high moisture content, because the reactivity of S02 appears to in-
crease at higher humidities.
204
-------
Both quartz and glass fiber filter material was tested by
Gelman. The quartz was found to be non-reactive with S02. The
glass fiber materials, Gelman Type II and a newly developed Spectro-
Grade prepared with H2S04, were low in S02 pickup; however, the
SpectroGrade glass, prepared with HC1, picked up significant amounts
of S02 . (See Table 1.) Their explanation is that glass fibers pre-
pared with H2S04 reacted to form CaS04 which prevents further reac-
tion with S02 to form sulfates. The test used for S02 reactivity
was to expose the filters to a water saturated atmosphere of S02 for
20 hours. Mass change and initial pH of each filter were measured.
Gelman does not elaborate on the meaning of "prepared with H2S04."
Another type of SpectroGrade coated with an organic silicone
resin showed low S02 pickup. This type of SpectroGrade with the
silicone treatment is now a standard type supplied by Gelman. Use
of the siliconized SpectroGrade at elevated temperatures, however,
can result in the disappearance of the coating and S02 absorption
by the filter medium since the filter is not prepared with H2S04.
Table 1. Sulfur Dioxide Pickup
(After Gelman and Marshall) (1)
mg/Sheet - 20 Hour Exposure
SpectroGrade-HCl
Siliconized
SpectroGrade HC1
SpectroGrade
H2S04
Type II Fiber
H2S04
Quartz
Quartz
Alkali Strengthened
mg
3
17
3
0
23 (est.)
Initial
pH
7.1
9.4
6.8
6.8
7.0
9.5
205
-------
Although the work by Gelman and Marshall showed quartz fiber
media to be non-reactive with SO2 t the material has been found to
be too fragile to be used successfully as an impactor substrate ma-
terial.
Whereas the Gelman and Marshall program was aimed at reducing
the interference to particulate mass gains by filter media, the
work of Barton and McAdie (2) in 1970 was designed to make filter
media acceptable for sulfuric acid aerosol collection. They
cite a study by Lee and Wagman (3) in 1966 which reported that
atmospheric sulfur dioxide could be oxidized catalytically on
the glass surface and thus seriously interfere with the determi-
nation of actual H2S04 aerosol levels. Barton and McAdie developed
a technique to reduce these blank effects. The filter material
was soaked for two or three days in a 20% solution of H2S04
followed by a thorough washing in distilled water, 80% isopropanol,
and acetone, respectively, in order to deactivate any surface
contaminants which could be responsible for the apparent irreversible
absorption of sulfuric acid by the glass fibers. These test
results are shown in Table 2. Also reported in this table are
the*results of Scaringelli, Boone, and Jutze (4) who used an
acetic acid wash. Their method allowed only a 50% recovery of
H2S04. It can be seen, however, that the method of Barton and
McAdie allowed full recovery of the H2SO4 solution.
The purpose of the Southern Research screening tests sponsored
by the EPA was to gain an understanding of the SO2 induced mass changes
that occur and to facilitate the selection of glass fiber filter ma-
terials suitable for use as impactor substrate media. A suitable
Table 2. Absorption of H2S04 by Glass Fiber Filters
(After Barton and McAdie) (2)
Filter Treatment
No filter
H2SO4 treatment
No filter
HOAc treatment
HOAc treatment with
refluxing
H2SO4 treatment
No filter
H2S04 treatment
Mg per 10 ml, 80% isopropanol
Added
6.0
6.0
14.8
14.8
14.8
14.8
18.0
18.0
Found
6.3 +
7.0 +
14.8 +
7.2 +
3.0 +
15.2 +
17.9 +
18.5 +
0.4
0.4
0.3
0.4
0.5
0.3
0.5
0.5
206
-------
substrate material would be one which has stable low mass character-
istics and is mechanically strong to resist cutting, tearing, and loss
of material. Since the mass changes are apparently a result of
chemical reactions involving the production of sulfates, the labora-
tory work was principally concerned with exposure of the substrate ma-
terials to sulfuric acid and/or a wet SC>2 gas. The stability of mass
changes over long time periods was investigated in order to evaluate
the prospects for preconditioning techniques as a means for control-
ling mass changes. Two laboratory test methods were employed. One
approach used a flow of saturated gaseous SC>2 through the filter ma-
terial, and the second involved soaking the material in hot sulfuric
acid solution.
In this laboratory study, glass fiber substrate materials were
exposed to air, SOa, and water vapor at an elevated temperature.
Figure 1 shows a diagram of the conditioning apparatus. Dry air was
preheated in the conditioning oven and then bubbled through a heated
water container at 60°C (140°F). Next, S02 was introduced to the
heated and humidified air stream. All lines carrying 862 laden air
were then passed through a chamber containing the filter media being
tested. The chamber was designed so that conditioning gases flowed
through the filter stack being conditioned.
Both gravimetric and pH determinations were used to investigate
the rate of SO2 uptake by the sample material. The procedure used to
determine the filter pH was a modification of Gelman's method for 8"
x 10" filter sheets (1). Two 47 mm filters were used for each pH deter-
mination .
In one series of tests, four different kinds of glass fiber sub-
strate materials were treated in the laboratory conditioning chamber
on an hour-by-hour basis. After each hour of conditioning, the
substrates were weighed, desiccated, reweighed, and the weights were
recorded. It was found that desiccation resulted in no change in the
weights, so this practice was discontinued. The four substrate ma-
terials tested were Reeve Angel 934AH, Gelman AE, Gelman SpectroGrade,
and Whatman GF/A. All filters were 47 mm in diameter. Eight groups
of twenty filters each were prepared and conditioned in the following
order:
1. Reeve Angel 934AH
2. Gelman AE
3. Gelman SpectroGrade
4. Whatman GF/A
5. Reeve An^e' 934AH
6. Gelman AE
7. Gelman SpectroGrade
8. Whatman GF/A
207
-------
HEATER
TAPE
FLOWMETER
^ WATER HEATER^
f AIR-SO2 EXHAUST
SAMPLE CONDITIONING
CHAMBER
SO2 AIR
-O AIR FLOW DIRECTION
-» SO2 FLOW DIRECTION
-». A!R-SO2 MIXTURE FLOW
DIRECTION
Figure 1. Diagram of experimental set-up for filter
substrate conditioning experiment.
208
-------
Gas flow was such that the Reeve Angel material was exposed
first. Figure 2 shows the results for the first nine hours of condi-
tioning. The conditioning temperature was 220°C (428°F). Water sa-
turated air with 5% S02 was pumped through the chamber at a rate of
2.1 1pm. Note that after nine hours of conditioning the Reeve Angel
material had not gained but lost weight. However, the weight loss was
miniscule and probably due to handling. All others had gained signi-
ficant amounts.
Sulfuric Acid Wash Treatment of Filter Media
Another approach to passivating impactor substrates was also
investigated. Bundles of Reeve Angel 934AH and Gelman AE 47 mm fil-
ters were soaked in hot concentrated sulfuric acid-water (50/50) mix-
tures for 90 minutes. These filters were then washed in distilled
water, washed again in ethanol (ETOH) or isopropanol (IPA), dried,
baked, and desiccated. Upon conditioning for one hour under the con-
ditions described above (220°C [438°F], air-water gas mixture with 5%
S02), eighteen Gelman AE 47 mm filters gained 11.9 mg or 0.66 rag/fil-
ter. Twenty untreated Gelman AE 47 mm filters gained 67.7 mg or 3.39
mg/filter with the same conditioning. Therefore, the sulfuric acid-
wash can make a difference when the filters are known to gain weight.
The Reeve Angel material again showed no weight gains.
The hour-by-hour conditioning of the four different types of
glass fiber filter substrate materials was continued, and mass
gains were monitored for a total of 26 hours of conditioning. In
addition, the sulfuric acid washed Gelman AE and Reeve Angel
934AH materials were laboratory conditioned on an hour-by-hour
basis for a total of 18 hours. Figure 3 shows the mass gain per
47 mm filter versus laboratory conditioning time. Data for
Gelman AE, AE acid washed, Gelman SpectroGrade, and Whatman GF/A
are presented. Reeve Angel 934AH plain and acid washed filter
materials were also conditioned, but since mass increase in this
material was negligible, these data were not graphed. The Gelman
AE acid washed material gained approximately one third as much
mass as the plain Gelman AE. Figure 3 also shows that even after
26 hours of laboratory conditioning ma.~;s gains may be expected
with further conditioning.
Gas analyses were conducted on the conditioning gas at the inlet
to the conditioning chamber: S02 and S03 concentrations were measured
at approximately 10,000 ppm, and 3 ppm to 5 ppm, respectively. Iron
is a catalyst for the conversion of SC>2 to SO3 at the conditioning
temperature (220°C, 428°F) . The conversion efficiency is small, less
209
-------
SYMBOL FILTER TYPE
INITIAL FILTER
MASS, mg
7.0
6.0
5.0
§
f 4.0
uf
3
UJ
a: 3.0
u
2.0
1.0
0.0
GELMAN AE
GELMAN SPECTROGRADE
WHATMAN GF/A
REEVE ANGEL 934AH
132
133
88
110
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
CONDITIONING TIME, hours
8.0
-•
9.0
Figure 2. Mass gains of four types of glass fiber filter materials
versus exposure time to a water-air-5% SCL gas mixture
at 260°C (500°F).
210
-------
E
DC
LU
I'-
LL'
E
E
**
DC
LU
Q.
Z
-------
than 1% but still enough S03 is produced to be detected. Since
all the S02 carrying lines and conditioning chamber are stainless
steel, we should expect that the filters which have been S02
conditioned have also been exposed to S03.
Table 3 summarizes the end-point results presented in Figure
3. These data are presented in the order in which the 47 mm
filters were conditioned in the stainless steel conditioning
chamber (alundum filter holder). Results are presented on a
mass gain per filter and percent mass gain basis.
In another series of tests, chemical analyses were made
on the laboratory conditioned and unconditioned filters. Table
4 shows the barium, calcium, and soluble sulfate concentration
in two types of glass fiber filter material conditioned at Southern
Research: Reeve Angel 934AH and Gelman AE. These 47 mm filters
were analyzed when received, after being baked-out and desiccated,
and after being conditioned. The Reeve Angel material shows
large amounts of calcium and miniscule amounts of barium and
soluble sulfates, even after 12 hours of conditioning. The Gel-
man AE materials show large amounts of calcium as well, but after
conditioning there is a great gain in soluble sulfates. This
is reflected in the mass gains for this material. Each 47 mm
filter gained on an average 2.93 mg. The initial pH of the Gelman
AE material after baking was 9.8. With two hours of conditioning
the pH dropped to 8.8. This is in contrast to the behavior
of the Reeve Angel material. The pH of this substrate material
stayed rather constant at about 5.9 to 6.7 before and after con-
ditioning. Mass gains on conditioning for any length of time
were very small.
These results indicate that a laboratory induced sulfate
mass gain can be made to occur in glass fiber filter materials.
Whether or not this mass gain or "conditioning" lasts is another
question. To determine if the conditioning is a temporary effect,
samples of these filters (16 to 20 filters per sample) were con-
ditioned for 2 to 12 hours. Some were exposed to ambient air
after conditioning, while others were desiccated.
Figures 4 and 5 show the results of these tests for the Reeve
Angel 934AH material. Figure 4 shows the percent weight change ver-
sus days after conditioning for groups of filter conditioned for 2
hours and 12 hours. One group was exposed to ambient air after con-
ditioning, and another group was desiccated after conditioning. In
both cases minute mass gains were seen for 12 hour conditioning, and
minute mass losses were seen for 2 hour conditioning. In either case
there appears to be no reaction after conditioning resulting in an
appreciable mass gain or loss. Figure 5 shows the pH of single fil-
ter samples measured after conditioning for 2 and 12 hours. As in
212
-------
Table 3. Mass Gains of 47 mm Glass Fiber Filter Substrate
Materials from Laboratory Conditioning
ro
00
Batch Number of 47 mm
Material
Reeve Angel
934AH
Gelman AE
Gelman Spec-
troGrade
Whatman GF/A
Reeve Angel
934AH
Gelman AE
Gelman Spec-
troGrade
Whatman GF/A
Reeve Angel
934AH
(Acid Washed)
Gelman AE
(Acid Washed)
Mass Before
Conditioning Conditioning
Number Filters Conditioned Time (hours) (grams)
3307
8204
8192-
20232
3563
3307
8204
8192-
20232
3563
4292
8206
20
20
20
20
20
20
20
20
20
20
26
2G
26
26
26
26
26
26
18
18
2
2
2
1
2
2
2
1
2
2
.1888
.6644
.6717
.7695
.2149
.6266
.6522
.7361
.0968
.6939
Mass After
Conditioning
(grams)
2
2
2
1
2
2
2
1
2
2
.1881
.8735
.8160
.8349
.2166
.8375
.8051
.8272
.0975
.7699
Mass Gain
per Filter
(
-------
Table 4. Barium, Calcium, and Soluble Sulfate Content
in Two Glass Fiber Substrate Materials
Original
After Bakeout
After Conditioning
(2 Hours)
After Conditioning
(12 Hours)
Reeve Angel 934AH
Barium
Ba++
Mass (M2) %BaO
63 0.06
54 0.05
135 0.13
<10 <0.01
Substrate Material
Calcium
Ca+H-
Mass (Mg) %C&0
14202 17.9
14055 17.8
14531 18.2
13820 17.6
Soluble
Sulfate
S04-
Mass (Mg) %S04
3.5
3.5
4.5
92
Gelman AE Substrate Material
Original
After Bakeout
After Conditioning
Barium
Ba+ +
Mass (Mg) %BaO
<10 <0.01
<10 <0.01
<10 <0.01
Calcium
Ca+ +
Mass (Mg) %CaO
6094 6.5
5470 6.1
5758 5.9
S04
Mass
<10
<10
3013
0.003
0.003
0.004
0.08
Soluble
Sulfate
—
(Mg) %S04
<0.01
<0.01
2.24
(2 Hours)
-------
+0.2
+0.1
0.0
i
-0.1
-0.2
O CONDITIONED FOR 2 HOURS
' CONDITIONED FOR 12 HOURS
SAMPLE EXPOSED TO AMBIENT AIR
J. AFTER CONDITIONING
+O.2
+O.K
On
t
-0.1
-0?
O CONDITIONED
I a • CONDITIONED
* ""•^^^" • — _.
• ' ' • •
.
SAMPLE DESSICATED AFTER
i i i i i i i i i
FOR 2 HOURS
FOR 12 HOURS
CONDITIONING
1
4 5 6 7 8 910
20
30
Figure 4. Percent weight change for Reeve Angel 934AH
glass fiber filter substrate material as a
function of time after conditionirg.
215
-------
pH
O pH BEFORE CONDITIONING
O CONDITIONED FOR 2 HOURS
• CONDITIONED FOR 12 HOURS
SAMPLE EXPOSED TO AMBIENT AIR
AFTER CONDITIONING
0 PH BEFORE CONDITIONING
O CONDITIONED FOR 2 HOURS
• CONDITIONED FOR 12 HOURS
SAMPLE DESSICATED AFTER CONDITIONING
6 7 8910
DAYS AFTER CONDITIONING
(conditioning occurs on day one)
Figure 5. pH of Reeve Angel 934AH glass fiber filter
substrate material as a function of time
after conditioning.
216
-------
Figure 4, one group was exposed to ambient laboratory conditions
while another group was desiccated. This substrate material appears
to have essentially no change of pH upon conditioning, and it is pos-
sible that since the material was only slightly acidic, changes in
the pH of water used in the pH determination could have caused the
changes shown in Figure 5. Whenever pH of a filter sample was
measured, the pH of the water used was also measured. We believe
this to be the case for the low pH recorded on day 2 of the desic-
cated sample and day 3 of the exposed sample. In this case the
raw distilled water used in the pH determination had a measured pH
of 4.32.
From these tests it would appear that, if pH is a good moni-
tor, the conditioning has a lasting effect. Samples of this material,
conditioned for 12 hours, which were stored under desiccation for
as long as 77 days, show no mass change and small change in pH
(6.10 before, 6.77 after).
A more detailed discussion of this work has been published in
a report entitled "Inertial Cascade Impactor Substrate Media for
Flue Gas Sampling" (5).
S03/H2S04 MASKING OF ULTRAFINE PARTICLE MEASUREMENTS
Southern Research Institute has for the past several years
been involved in determining particulate concentrations and size dis-
tributions for particles in the submicrometer size range. Much
of this work has been directed toward characterizing the emissions
from particular industrial sources and toward determining the ef-
ficiency at which industrial gas cleaning equipment removed parti-
cles in the 0.01 to 1 /urn f^ia-meter size range.
Typical sample gas conditions are shown in Table 5. At any
one source, the gas conditions may be any mixture within the ranges
shown in the table, and, in fact, these may not represent the ac-
tual extremes. Because of the high particulate concentrations, both
by mass and by number, the presence of large quantities of condens-
able vapors (i.e., water and H2S04) and high concentrations of cor-
rosive gases (H2SO4, S02 and others) exiensive sample conditioning
and dilution become mandatory. The required dilutions approach
5000:1 in some instances, and dilutions by factors of about 500:1 are
quite commonly needed. In many instances these dilutions are set,
not by requirements for diluting condensable vapors sufficiently
to insure that sample gas stream to the instrumentation is not
saturated, but by limiting factors in obtaining reliable data from
217
-------
Table 5. Typical Flue Gas Sample Conditions
^^^^^^^^^^^^^g.^—U^^^^jjg^g^^^f^f^^^^^fgfff^lf^^^^^fi^B^^i^^^^^^BR^BRnfB^^^^^^^BnBHH^^^^^^^n^nHBi^BHHHHml^^^tBi^^f^BIBfHUni^^m***^***^^^*^^.
Range Typical Value
Temperature: Ambient to 800°C 150°C
Absolute Pressure: 490 to 850 mm Hg 750 mm
Pressure Differential
to Ambient: + 75 mm Hg -25 mm
Moisture (Vol. Percentage) 1 - 40% 10%
CO2 (Vol. Percentage) 0 - 15% 12%
SO2 (Vol. Percentage) 0-1.4% 0.2%
SO3/H2S04 (Vol. Percentage) 0 - 0.25% 0.0005%
Particulate Mass Inlets 5 g/m
Concentrations to 20 g/m Outlets 0.05 g/m
Concentration by 7
Number of Particles 2 x 10 inlet
Larger Than 0.01 Mm to 3 x lO'/ml 2 x 10 outlet
218
-------
the available detectors. Reliable concentration data for the pur-
poses of diffusional analysis can be obtained only within the linear
response concentration limits of the condensation nuclei counters
used as detectors. These limits are 105 particles/ml for the in-
struments which we normally use (GE Laboratory Model CMC's and
Environment One Model Rich 100 devices). In almost all the cases,
dilutions by at least 100:1 have been necessary simply to reduce
the concentrations to the linear range limits of the condensation
nuclei counters.
On several field tests an interference phenomenon has been
observed that appears to be an acid condensation fume. Figure 6
is a sample data set which illustrates this phenomenon. As the sam-
ple gas flowrate to the diluter was increased (thus reducing the
dilution factor), the measured diluted concentrations increased li-
nearly with the sample flowrate until a critical value was reached.
At this point slight increases in sample flowrates led to very large
nonlinear increases in apparent aerosol concentrations.
It would appear that when S03/H2S04 (existing in the flue as
a vapor) is cooled below a characteristic dewpoint, it condenses
to form an acid fume and gives rise to the phenomenon shown in Figure
6. The condensation aerosol thus produced has a very small mean
size because a diffusion battery having 50% penetration by 0.016
Mm diameter particles will pass only a few percent of the condensa-
tion aerosol particles. The dewpoint is sensitive to the concentra-
tions of both H20 and 863. Thus, the condensation fume can be
avoided by either using sufficiently high dilutions or by removing
the S03/H2S04 while in a gaseous state. (Once .H2S04 has condensed
very high tempeiatures are required for re-evaporation.) Dilution
is effective in avoiding the condensation fume. It is for this
reason that the phenomenon is not normally encountered at inlet sam-
p"1 • ng ports to control devices where high dilutions are used. How-
ever, dilution alone may be inadequate because frequently the mini-
,dm dilution at which the onset of the condensation fume occurs can
be so high tuat the diluted particle concentrations are below the
minimum detection limits of the sizing instruments.
Gilmore Sera (6) has also reported experiencing this condensation
fume when using an extraction dilution system similar to the one de-
veloped by Southern Research. His data are shown in Figure 7. These
data were taken at the inlet to a baghouse on a western coal-fired
utility boiler. The dilution factor of 21 used at the inlet was
apparently not sufficient to provent formation of these particles.
Sera calculated that the undiluted number concentration, if they
existed in the stack, would be 23 x 10 particles/cm . Coagula-
219
-------
100
g •
cc
I-
| 60
40
O
S
- 20
i
> 104
o
o
2468
SAMPLE FLOW RATE
10
Figure 6. Behavior of diluted sample concentration at
onset of sulfuric acid condensation.
220
-------
CO
,0
CO
Ul
0.005 0.01 0.02 0.05 0.1 0.2
PARTICLE DIAMETER, Dp,
0.5
Figure 7. Typical size distribution at the inlet to the
baghouse collector on a coal-fired stea... boiler,
measured and shown diluted 21x. The large
number of very small particles may be formed
by SOX condensation in the diluter. After
Sem (6).
221
-------
tion rates at this concentration would be very rapid, causing such
particles to grow in size by collision with others, resulting in a
lower number concentration. Thus, it does not appear that these
particles could have been older than several seconds, pointing to
the sampling and conditioning system as the probable source.
To cope with this condensation problem, an oven containing S03
absorbers was incorporated into the Southern Research Sample Ex- -
traction and Dilution System shown in Figure 8. In this system a
0.5 ACFM sample flow is removed from the process exhaust stream and
is pulled through a rigid probe, a flexible connector hose, and a
cyclone into a "T" where the flow splits. The excess flow (needed
to maintain a constant 0.5 ACFM flow through the cyclone) is dumped
to ambient, and the desired sample flow (0.003 to 0.5 ACFM) goes
into the diluter via a calibrated orifice and the SOX absorber
bank. The cyclone, orifice, and SOX absorber bank are housed in a
heated o'ven so that all components of the system, except the di-
luter, can be maintained at 400°F to prevent condensation. Pressure
taps for the cyclone and the orifice allow continuous monitoring of
the cyclone flowrate and the orifice flowrate by reading the pres-
sure drop across the respective component. The oven was designed
to house an adequate number of heated SOX absorbers for reducing
the vaporous SO3/H2S04 concentrations by diffusion to an ab-
sorber reagent. The number of absorbers required depends on the
sample flowrate through the absorbers (residence time), the reagent
used in the absorbers, and the initial levels of the SO3 in the
stack gas. Several reagents were considered (barium oxide, calcium
oxide, PbO, granulated copper, calcium carbonate, silica gel, and
activated charcoal), but only three were readily available in a
granulated form: granulated copper, PbO, and activated charcoal.
These three were used in a field test at a copper smelter operation
where the 803 content of the stack gas ranged up to 0.25%. PbO and
copper were found to be inadequate at this high S03 level even when
six absorbers were used. Activated charcoal, however, was success-
ful when six absorbers were used. In the configuration shown, acti-
vated charcoal allowed continuous running times of about six hours,
although there were intermittent concentration increases which ex-
ceeded the removal capability of the absorbers and allowed a fume
to form momentarily. Activated charcoal was, therefore, selected
as the standard reagent. At many locations, S03 levels in the stack
gas are so low that an acid fume is not encountered, and at others
the concentrations can be adequately reduced by using only a few ab-
sorbers.
It should be noted that iron can act as a catalyst in the con-
version of S02 to S03. For this reason, all parts of the SEDS which
are made of stainless steel and come into direct contact with the
222
-------
CO
TIME
AVERAGING
CHAMBER
SIZING
INSTRUMENT
BLEED DILUTION DEVICE
CHARGE NEUTRALIZER
DIFFUSIONAL
DRYER
SOX ABSORBERS (OPTIONAL)
PROCESS EXHAUST LINE / /
CHARGE NEUTRALIZER
CYCLONE
ORIFICE WITH BALL AND SOCKET
JOINTS FOR QUICK RELEASE
HEATED INSULATED BOX
RECIRCULATED CLEAN. DRY. DILUTION AIR
FILTER BLEED NO. 2
MANOMETER
COOLING COIL
3630-036
PRESSURE
BALANCING
LINE
DRYER
BLEED NO. 1
Figure 8. Sample extraction-dilution system (SEDS).
-------
hot undiluted sample gas were passivated (removal of the iron from
the exposed surfaces of the stainless steel) by immersion in hot
nitric acid so as to avoid the generation of S03/H2S04.
SUMMARY
Sampling for total mass particulate or particle number concen-
trations in gas streams containing sulfur oxide fractions can result
in inaccurate data unless precautions are taken to passivate glass
fiber media or properly condition the sample streams. Techniques
have been mentioned by which glass fiber materials can be conditioned
against the uptake of S02. SOX absorbers are also useful for the
removal of S03/H2S04 prior to sample stream cooling and dilution in
an ultrafine particle measuring apparatus.
224
-------
REFERENCES
1. Gilman, C., and J. C. Marshall. High Purity Fibrous Air
Sampling Media. American Industrial Hygiene Association
Journal, 36(NA):512-517 , 1975.
2. Barton, S. C., and H. G. McAdie. Preparation of Glass Fiber
Filters for Sulfuric Acid Aerosol Collection. Environmental
Science and Technology, 4(9):769-770, 1970.
3. Lee, R. E., and J. Wagman. A Sampling Anomaly in the Determi-
nation of Atmospheric Sulfate Concentration. American Indus-
trial Hygiene Association Journal, 27(3 ):266-271, 1966.
4. Scaringelli, F. P., R. E. Boone, and G. A. Jutze. Journal of
the Air Pollution Control Association, 16(6):310, 1966.
5. Felix, L. G., G. I. Clinard, G. E. Lacey, and J. D. McCain.
Inertial Cascade Impactor Substrate Media for Flue Gas Sam-
pling. EPA-600/7-77-060, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, 1977. 89 pp.
6. Sem, G. Submicron Particle Size Measurement of Stack Emissions
Using the Electrical Mobility Technique. In: Proceedings of
the Workshop on Sampling, Analysis, and Monitoring of Stack
Emissions, Electric Power Research Institute Report SR-41,
Palo Alto, California, April 1976.
225
-------
Primary Aerosol Sulfur Size Distribution
Measurements Using a Low Pressure Impactor
Richard C. Flagan
California Institute of Technology
ABSTRACT
The use of a reduced pressure cascade impactor to
measure aerosol sulfur size distributions in flue
gases from fossil fuel combustion will be examined.
The impactor was developed to determine the contribution
of sulfates to atmospheric submicron aerosols. A
minimum cutoff diameter of 0.05 ^m was achieved by
operating four stages of the impactor well below atmos-
pheric pressure. The small deposits on each impaction
stage are ideal for sulfur analysis using a sensitive
(1 ng sulfur detection limit) flash vaporization/flame
photometric detection system. The impactor was designed
and calibrated to size segregate particles smaller than
about 8 MM diameter from a room temperature aerosol.
Re-entrainment and bounce-off are small as long as the
total loading on each stage is small and there are few
par '-ides larger than about 10 ^m diameter.
Several changes in the operation of the reduced pressure
are necessary if a hot gas stream containing high aerosol
concentrations is sampled. Large particles should be
removed upstream of the impactor. Dilution of the hot
gases is necessary to provide reasonably long sample
times before re-entrainment becomes a problem and to
reduce the temperature without significantly altering the
particle size. Rc-deoign of the impacted for operation
at elevated temperatures will be considered.
227
-------
INTRODUCTION
For pollutant aerosols, the distribution of chemical species
with respect to particle size is important in the evaluation of
health effects and transport behavior. Cascade impactors commonly
used for the size segregation of aerosols in chemical analysis can
collect particles as small as 0.3 to 0.5 /u,m. Smaller particles
which pass through the impactor are collected on the after-filter.
Recent studies of fine particles in the flue gases of coal-
fired power plants have shown that large numbers of submicron par-
ticles are produced (1)(2). Moreover, the fine particles are
found to be enriched with volatile species including a number of
heavy metals and sulfur (3). In order to determine the impor-
tance of these emissions and to understand their formation, it
will be necessary to determine their chemical composition. Fine
ash and soot particles may be formed by homogeneous nucleation in
the high temperature combustion zone. It has been suggested that
0.3%-3% of the ash in coal may be vaporized during pulverized coal
combustion (4). This volatile ash may then condense and coagu-
late, forming a narrow size distribution at about 0.1 um mean dia-
meter according to recent calculation. Since this is near the min-
imum in the collection efficiency of most gas cleaning devices,
much of that fine particulate matter will be emitted into the
atmosphere. Disproportionate quantities of the heavy metals that
are volatile at combustion temperatures may, thus, be emitted into
the atmosphere.
Sulfate aerosols are also probably formed by condensation.
Most of the sulfur in coal is oxidized to form S02, but only small
amounts form S03 in the flame. Additional SO3 may be formed by
heterogeneous reactions on particles or heat transfer surfaces.
Only at quite low temperatures will the 803 react with water or
other species forming condensed phase sulfates or sulfites. Sul-
fate aerosol may be formed either upstream or downstream of the gas
cleaning equipment, depending upon temperature, ash composition,
and other possible variables. Thus, sulfate aerosols may not be
removed by the gas cleaning equipment.
Sulfur represents an additional problem, since it may condense
in sample lines even though it is a vapor in the flue gases. For
this reason it is desirable to operate gas sampling and analysis
equipment at flue gas temperatures. If this is not possible, dilu-
tion systems may be used, but very large dilution ratios will be
required to prevent condensation (5).
228
-------
Reduced pressure cascade impactors have been developed for the
study of submicron particles in the atmosphere (6) (7). At low pres-
sures the mean free path of the gas molecules is comparable to the
diameter of the particles. The aerodynamic drag on the particles
is reduced, making it possible to collect the fine particles by
inertial impaction. Particles as small as 0.05 /j.m diameter have
been collected with low pressure impactors.
Roberts and Friedlander (8) have developed a sensitive method
for the analysis of aerosol sulfur. The aerosol is collected on
small (0.1" x 0.8" x 0.001") stainless steel strips mounted on the
collection surface of a single jet cascade impactor. The sulfur
deposited on the strip is pyrolyzed by rapidly heating the strip to
about 1200°C and the sulfur is then detected using a flame photo-
metric detector. The detection limit of this method is about 2
ng sulfur. The total sulfur content is measured by this method,
but the chemical species are not identified. Roberts' method for
aerosol sulfur detection has been used with the low pressure
cascade impactor to determine the sulfur size distribution in
urban aerosols.
This combination should be useful for the measurement of sul-
fur in combustion generated aerosols. The low pressure impactor
has been developed and calibrated for operation at ambient tempera-
ture and pressure. The present study examines the possibility of
design modifications for elevated temperature operation.
INSTRUMENT DESCRIPTION
To make use of Roberts' technique, a low pressure impactor
was developed with one circular jet per stage. The impactor has
eight stages, and samples at a rate of one liter per minute. The
aerodynamic cutoff diameters, corresponding to particles collected
with an efficiency of 50%, are 0.05, 0.075, 0.11, 0.26, 0.50, 1.0,
2.0, and 4.0 ftm. Particles larger than 0.5 jtim are sampled at atmos-
pheric pressure using the first four stages of the Battelle impac-
tor (Delron #DC15, Powell, Ohio). fo^r additional stages, operat-
ing at pressure of 8-150 mm Hg, absolut3 size segregate the smaller
particles. These pressures (Table 1) refer to the stagnation pres-
sure below the jet. A critical orifice, 0.036 cm in diameter,
separates the atmospheric and low pressure stages and determines
the sample flow rate. The impactor is cylindrically shaped, with
a diameter of 2.5", standing 18" high. For operation it requires
a vacuum pump with a displacement of at least 100 Lpm. In this
229
-------
lab, either the Leybold Hereas Trivac S4A or Sargent Welch 1403
was used.
Each stage of an impactor is characterized by its collection
efficiency as a function of particle size. The diameter of a par-
ticle of unit density which is collected with a 50% efficiency is
referred to as the aerodynamic cutoff diameter. In the design of
the impactor, two things are considered: (1) the choice of jet
diameters at the specified flow rate to obtain the desired cutoffs,
and (2) impactor geometry and jet Reynolds number to minimize the
cross sensitivity between stages.
Table 1. Low Pressure Impactor Design and Operational Parameters
Stage Number
a
b
c
d
1
2
3
4
Orifice
5
6
7
8
D - (cm) Pa (mm Hg) V . (m/sec) Re .
J «J \J
.249
.140
.099
.064
.036
.110
.099
.099
.140
1 atm = 745 mm Hg
Calibration data from
This work
Method of calibration
744
743
740
720
150
140
106
50
8
3.5
11
22
54
-
93
150
300
300
Delron Research Products
described in Part II
560
990
1400
2170
-
1270
1430
1480
1050
(Powell,
Cutoff (/im)
4.0b
2.0b
1.0b
0.50b'C
-
0.26C
O.llc'd
0.075d
0.05d
Ohio)
For all stages, the jet to plate spacing is one half the
jet diameter and the length of the jet throat is 6 mm.
230
-------
For this impactor it was desired to build four low pressure
stages to give approximately equal logarithmic intervals in the
size range between 0.05 and 0.5 jum. To attain the reduced pres-
sures, these four stages are preceded by a critical orifice which
limits the flow rate to 1 Lpm; the orifice and these four stages
follow the four atmospheric pressure stages of the Battelle
impactor.
The jet diameters and operating pressures of the low pressure
stages are chosen on the basis of the cutoff diameter calculated
from the Stokes number, defined as
St =
d
P
V
D .
J
C
D d2VC
P P
18 D .
"i
.j
particle density
particle diameter
jet velocity
jet diameter
slip factor
air viscosity
The collection efficiency of each impaction stage exhibits the same
functional dependence on the Stokes number, provided the geometry
and flow regime of the stages are similar. Of particular interest
is the value of the Stokes number at a 50% collection efficiency,
St50. From this number the aerodynamic cutoff diameter is calcula-
ted for a specified jet diameter, pressure, and mass flow rate.
For the impactor, St50 for the atmospheric pressure stages is
r. .09 + 0.01, and this value was used to design the low pressure
stages. Although the low pressure stages have geometry and jet
Reynolds numbers similar to the atmospheric pressure stages, the
Mach numbers are significantly higher, and this may change the
value of St
stages.
50
Thus it is necessary to calibrate the low pressure
Inspection of Eq. 1 shows that for a fixed mass flow rate,
smaller particle cutoff diameters ;aay be achieved by either de-
creasing the jet diameter or decreasing the pressure. The reduc-
tion in the drag on a particle at the lower pressure is reflected
in che increased value of tne slip factor, C, which is a function
of the ratio of particle diameter to the mean free path of the air
molecules. The raultijet low pressure impactors of McFarland et
al. (9) and Buchholz (10) achieve successively smaller size cuts
231
-------
by decreasing the jet diameters while maintaining a constant
pressure. All of the low pressure stages of the McFarland
impactor operate at 24.3 mm Hg, and those of the Buchholz
impactor operate in the range of 32-39 mm Hg. By contrast,
for this impactor the low pressure stages operate at pressures
from 8 to 140 mm Hg. Here the successively smaller size
cuts are achieved not by decreasing the jet diameter but by
decreasing the pressure. This variation in the operating
pressure of each stage arises from the compressibility of
the flow at the higher jet velocities used here.
The particles are classifed according to their aerodynamic
drag at this reduced pressure. The aerodynamic diameter of the
particle is
P C
-.-V^-S:-. <2)
where ds is the Stokes diameter and Cs and Ca are the slip correc-
tions to the Stokes drag for particles of diameter ds and da,
respectively. The slip correction is given by
C = 1 + d~ [1.257 + 0.04 exp (-.55d /X)J
where A, is the mean free path. For particles much larger than
the mean free path, C = 1 and da = V7Tds. In the free molecular
limit, da = pds.
Stages 4, 5, and 6 were calibrated using monodisperse poly-
styrene latex spheres (Dow Chemical) (6). A ThermoSystems, Inc.,
Model 3071 Electrostatic Classifier was used to produce a near
monodisperse aerosol of sodium fluorescein (uranine) for cali-
bration of stages 4, 5, 6, 7, and 8 (7). The collection effi-
ciency curves from these studies are shown in Figure 1. Stages
4, 5, and 6 have relatively sharp efficiency curves. Some par-
ticles much smaller than the cutoff diameter, d5Q, are collected
on the eighth impactor stage.
The combination of the impactor stages and the after-filter
collected 97%-100% of all the particles in the size range 0.057
to 0.39 film (7). The collection efficiency for smaller particles
was lower, 82% and 75% for 0.034 /Am and 0.0078 nm particles,
respectively. Diffusional losses are expected to become significant
232
-------
0.01 0.10 1.0
AERODYNAMIC DIAMETER (/im)
Figure 1. Low pressure impactor calibrations at c^ibient
pressure and temperature (7).
233
-------
for small particles in the low pressure stages of the impactor.
The calculated wall losses in the free molecular regime are 7% for
the 0.034 jum particles and 35% for the 0.0078 /xm particles. Most
of these losses occur between the eighth stage and the after-filter
Particle rebound was not found to be a serious problem when
coated impaction surfaces are used. The large particles which
would rebound from the lower stages are collected with high effi-
ciency on the uj'per stages. The problem could be more severe when
the gas stream contains substantial quantities of particles much
larger than the cutoff diameter of the first impactor stage. When
a pulverized coal aerosol was sampled, 10 to 50 /xm particles were
observed on the seventh and eighth stages. It may be necessary to
remove large particles from the sample upstream of the impactor in
order to prevent the lower stages from being contaminated.
IMPACTOR PERFORMANCE PREDICTIONS
A stage of the low pressure impactor is sketched in Figure 2.
The pressure, temperature, and velocity at point 2 determine the
cutoff diameter of the stage. These conditions determined by the
mass flow rate, the stagnation pressure and temperature, P°
and T° ,
and the jet diameters d
The flow from 1 to 2 can be treated as
a nozzle flow using the Bernoulli equation for compressible flow
I fv 2 _ 2
2 CV2 Vl
Y-
P
_
[3]
The density at 2 can be related to the velocity by continuing, i.e.,
,2, [4]
The pressure ratio for a non-ideal nozzle is Crocco (11)
[5]
where the reduced velocity is defined as
234
-------
2' 2' 2
z
I
Figure 2. Impactor geometry.
235
-------
W . V2/(2cpT°)
1/2
[6]
and n is the nozzle efficiency. Combining Equations 3-6, we find
V 2
r-i V2 \ y-i
1 9 Y
— V = '
2 V2 y-i
1 -
1 -
4 m
[7]
Equation 7 can be solved numerically to determine V2. P2 is
then evaluated using Eq. 5. The temperature is calculated
using Eq. 4 and the equation of state for an ideal gas.
The converging flow of a conical nozzle results in an
efficiency of
= (1 + cos 0) /4
[8]
Where 8 is the half-angle of the converging section (11). The
45° half-angle of the Battelle impactor gives TJC = 0.72. The
actual efficiency is lower. Impactor stage pressure drops
comparable to those observed for the subsonic stages, stages 1-6,
are predicted for a nozzle efficiency of T| = 0.5, see Table 2.
The calculated cutoff diameters are also compared with
the calibration data of Bering et al. (6)(7) in Table 2. The
predictions are very close to the observed values, even though
the Mach numbers of the jets are as high as 0.5 on stage 6.
The flow through stages 7 and 8 is choked. Calculated cutoff
diameters for the sonic stages do not agree with the measured
values. Calibration provides the only reliable means of
determining the cutoff diameters of these stages at the present
time.
Several factors will alter impactor performance at elevated
temperatures. The mass flow rate in the impactor, which is con-
trolled by choked flow through the critical orifice, varies
inversely with the square root of the stagnation temperature.
The viscosity increases with increasing temperature, as does the
mean free path. Thus, particle drag will increase in the con-
236
-------
tinuum regime and decrease in the free molecular regime when the
gas temperature is increased.
The predicted cutoff diameters of Hering's impactor operated
at elevated temperatures are summarized in Table 3. Stages 1 and
2 are expected to collect larger particles as the temperature is
increased. The performance of stage 3 (1 /am) is not expected to
change significantly. The cutoff diameters for stages 4-6 should
decrease at higher temperatures.
Table 2. Comparison of Impactor Calibration
with Predicted Performance
Stage
1
2
3
4
Orifice
5
6
7
8
D . (cm)
.249
.140
.099
.064
.036
.110
.099
.099
.140
P (mm Hg)
744
743
740
720
150
140
106
50
8
p
calc
745
744
740
715
-
134
100
-
-
V (cm/s)
344
1090
2190
5400
6620
15600
-
-
d5Q O.m)
4.0
2.0
1.0
.50
-
.26
.11
.075
.05
calc
4.4
1.8
1.0
.49
-
.23
.11
-
-
T° = 295K
P° = 745 mm Hg
237
-------
Table 3. Predicted Impactor Performance in
Elevated Temperature Operation
Stage
1
2
3
4
Orifice
5
6
7
8
D,i
.249
.14
.099
.064
.046
.110
.099
.099
.140
*Measured
T=295K
(cm) d5Q Gum)
4.4
1.8
1.0
.40
.23
.11
.075*
.05*
by Hering et al. (7)
T=400K
d50
4.6
1.9
1.1
.48
.18
.079
-
-
T=600K
d50
4.8
2.1
1.0
.44
.13
.053
-
-
In summary, it should be possible to use the low pressure
impactor for in situ size segregation of the submicron particles.
The temperature will, however, affect the flow rate and cutoff
diameters. It will certainly be necessary to recalibrate
the instrument for elevated temperature operation. Finally, the
impactor calibration data of Hering et al. (7) show that particles
much larger than d50 are likely to bounce off the impactor stage.
In order to prevent contamination of the smaller particle size
fractions, it probably will be necessary to remove large particles
from the sample upstream of the impactor inlet.
238
-------
REFERENCES
1. McCain, J. D., J. P. Gooch, and W. B. Smith. J. Air Pollut.
Contr. Assoc., 25:117, 1975.
2. Flagan, R. C., and S. K. Friedlander. Particle Formation in
Pulverized Coal Combustion. In: Recent Developments in
Aerosol Science, Shaw, D., ed. (in press).
3. Davison, R. L., D. F. S. Natusch, and J. R. Wallace. Environ.
Sci. Techn. 8:1107, 1974.
4. Ulrich, G. D., J. W. Riehl, B. R. French, and R. Desrosiers.
Mechanism of Submicron Fly-Ash Formation in a Cyclone, Coal-
Fired Boiler. ASME International Symposium on Corrosion and
Deposits, Henniker, New Hampshire, 1977.
5. Dolan, D. F., D. B. Kittelson, and K. T. Whitby. ASME Paper
No. 75-WA/APC-5, 1975.
6. Hering, S. V., R. C. Flagan, and S. K. Friedlander. Design
and Evaluation of a New Low Pressure Impactor, I, 1978.
7. Hering, S. V., S. K. Friedlander, J. J. Collins, and L. W.
Richards. Design and Evaluation of a New Low Pressure
Impactor, II, 1978.
8. Roberts, P. T., and S. K. Friedlander. Atmos. Environ. 10:403,
1976.
9. McFarland, A. R., H. S. Nye, and C. H. Erickson. EPA Report
No. EPA-650/2-74-014, 1973.
10. Buchholz, H. Staub Reinholt der huf (English Translation)
30:15, 1970.
11. Crocco, L. Fundamentals of Gas Dynamics. H. W. Emmons, ed.
Princeton University Press, 1958.
239
-------
Use of a High-Flow Stack Sampler for
Determination of Particulate Sulfate Emissions
A. Jack O'Neal, Jr.
Harold Cowherd
Long Island Lighting Company
ABSTRACT
The Long Island Lighting Company developed a high-flow
stack gas sampler suitable for use by power station
technicians. It allows for the collection of large
quantities of particulate material (300 mg-500 mg) for
physical and chemical analysis in 10-15 minutes of
sampling time. The device is particularly well suited
for evaluating the relative effects of changes in
furnace conditions, and it also appears to have acceptable
accuracy on an absolute basis. Sulfate concentrations
are determined from the filter extract gravimetrically
by barium precipitation.
The sampler was qualified by evaluating both the filter
characteristics (standard high-volume ambient sampler
filters can be used) and non-isokinetic sampling. Even
though large gas flows at high SO2 concentrations
(~1500 ppm) are passed through the filter, artifact sul-
fate formation appears to be minimal. Filter retention
was evaluated by analyzing the content of a liquid im-
pinger that received a portion of the gas stream on the
downstream side of the filter.
Primary sulfate emissions were evaluated for both high
(2.8%) and low (0.3%) sulfur residual oil-fired boilers,
and were found to be about 0.5% of the total sulfur
emissions. This figure compares well with Brookhaven
National Laboratory data obtained under similar conditions,
but using the controlled condensation method.
241
-------
INTRODUCTION
Conventional stack gas sampling, at best, is tedious, ex-
pensive, and time-consuming and was initially designed to examine
efficiencies of dust collectors. Today, one needs to know not
only dust collector efficiency but also the very nature of com-
bustion products which escape collection and become stack
emissions.
Over the past 15 years a growing body of evidence tells us
that the nature of combustion products is largely dominated by
conditions of the fire, some 4-6 seconds before sampling at
stack entry. Believing this, LILCO decided to find the cause-
effect relationships between furnace conditions and the nature
of stack gases. However, the use of conventional equipment
would virtually prohibit such a study unless one had unlimited
time, money, and manpower, so it was necessary to design a stack
gas sampler which would meet the following criteria:
1. A large enough mass (300-600 mg) of particulates
must be collected to allow thorough and accurate
physical and chemical analyses;
2. Five to fifteen minutes of actual sampling should
produce 300-600 mg of material;
3. Results should be quite reproducible;
4. It should be a one-man operation; and
5. The sampling procedure should be easily learned by
the average technician.
SAMPLING EQUIPMENT
Figure 1 presents a schematic of the prototype high-flow
stack gas sampler that was designed and fabricated to conform to
the constraints of this project. The unit was constructed
essentially of 2" and 3" 316 stainless steel schedule 5 pipe,
except for the filter holder which was constructed of two
modified Hi-Vol stainless steel diffuser sections and was de-
signed to accept the standard 8 x 10" Hi-Vol glass fiber filter.
A 38" long and a 100" long sampler probe were designed for use
with the instrument. Both probes were of 2" nominal diameter.
The probe assembly was designed to pass through a 4" schedule 40
gas duct sample port.
242
-------
OJ
Primary Ejector
Air Inlet
Temperature Probe
Filter Holder
Filter
Air Flow Meter
Impinger
Outlet
Flow Meter
Manometer
Filter Manometer
Shut-Off Valve
Depth Plate
Pitot Tube
Sample Gas Inlet
u
Ejector Air
Modulating
Valve
Air Ejector
( \
Figure 1. Schematic: Hi-flo stack gas sampler.
-------
In operation, the sample probe was inserted into the gas
stream through the gas duct sample port. The depth of insertion
was controlled by setting the adjustable depth plate. After
opening the shut-off valve, the sample gas was induced to flow
into the sample probe, through the shut-off valve, across the
particulate filter, and through the measuring orifice plate, by
the pressure drop potential created by the compressed air operated
ejector. The pressure drop across the particulate filter, the
orifice plate, and the gas duct was measured by manometers and
the temperature of the gas in the duct and the sample gas enter-
ing the filter was measured using standard thermocouples. A
pitot tube was mounted at the bottom of the sample probe and was
used to measure gas duct velocity head and to determine when the
probe inlet was facing directly into the gas stream.
Interchangeable air flow meter orifice plates of 1.00, 1.25
and 1.50" diameter and ejector primary air nozzles of .25, .375
and .50" diameter were fabricated to provide flexibility to the
design.
CALIBRATION
The airflow meter was flow calibrated for all three orifice
plates while installed in the sampler in the normal operating
configuration. The exit of the sampler ejector was connected to
the inlet of a Rootsmaster, Model ALP 125 gas meter calibrator.
Gas sampler orifice plate pressure drops were recorded for values
of calibrator volume flow rate. After the calibration of the
assembly was completed, the air flow meter section was removed
from the assembly, and its exit was connected to the calibrator.
This configuration was calibrated using the 1.25" orifice plate
only. Except for the very lowest air flow, the calibration re-
sults were identical to those previously obtained for this
orifice plate.
PRELIMINARY RESULTS
Preliminary testing of the prototype gas sampler (under
various furnace conditions) has been completed. The tests have
established the following:
1. The concept of high volume flow sampling is a
feasible concept.
244
-------
2. The repeatability of sample collection is well within
the acceptability range. Standard deviation for mul-
tiple samples is equal to 10% of actual values.
3. A sample of sufficient amount can be collected within
less than fifteen minutes, depending upon unit loading
and fuel sulfur content.
4. Isokinetic sampling is not as critical, at least for
this instrument, as was initially thought. (See Figure
2.)
5. The filters maintian their integrity after exposure
to the 300°F stack gas and can be used for further
analysis after weighing to determine the collected
sample weight.
6. Filter retention can be examined by pulling filtered
gas through a glass orifice immersed in a column of
demineralized water. (See schematic in Figure 1 for
sample source, marked Impinger Outlet.) The water was
analyzed for flue gas particulate constituents, and only
barely discernible traces were found. Thus, we believe
the glass fiber filter pad has excellent retention with
a labeled porosity of 0.3 microns.
During preliminary testing, condensation of the stack gas
in the sampler did not appear; however, it did appear during later
stack sampling when experimental incineration of boiler cleaning
solvent was being conducted. Condensation appeared in both the
sample collection section and the pitot tube section of the probe.
This problem is common with stack gas sampling equipment; however,
for this instrument, it is not as prevalent in the summer as in
the other seasons. We are in the process of devising a means of
alleviating this problem at the present time.
SOME RESULTS OF ACTUAL TESTING
Data have been obtained on two LILCO units to date, both oil-
fired, the same design, and the same size at 185 MW, normal load.
Barrett No. 2 burned a 0.3% sulfur oil containing about 0.01% to
0.02% ash; magnesium oxide additive is fed equal to oil ash on
a pound-per-pound basis. Thus, each 341 Ib. barrel of oil contained
about 0.07 Ibs. of oil ash plus 0.07 Ibs. of MgO. The furnace
normally operates between 0.8% and 1.1% excess oxygen at 185 MW.
There is an operating cyclone ash collector where all collections
245
-------
TEST DATE; e/31/76
LOCATION- GLENWOOD POWER STATION, UNIT 50
SOUTH AIR HEATER EXIT DUCT (GAS SIDE)
UNIT POWER OUTPUT 46.5MW
(9
hi
* .eH
kl
0»
CJ fi _j
— .v"
kl
Z
2
O
at
^ A-\
o
*
kl
0.
(A
.2 .4 .6 .8
SAMPLER INLET AIRFLOW / DUCT VELOCITY
I.O
Figure 2. Hi-flo stack gas sampler effect of non-iso-
kinetic sampling.
246
-------
are re-injected back into the furnace. Table 1 shows the data from
four tests on October 23 and four tests on October 24, 1976.
There was a time lapse between the Barrett tests and the ones
on Port Jefferson No. 4 unit in August through November of 1977.
The instrument was used in that interim to justify incineration of
boiler organic cleaning solvents as an environmentally acceptable
disposal technique (1).
Between August 18 and November 15, 1977 , 36 tests were per-
formed on Port Jefferson No. 4 unit. Some were performed when the
boiler was clean, others when it was dirty; some tests utilized
steady-state furnace conditions, others when furnace conditions were
being altered. Teflon-backed filters and glass fiber filters were
compared. During all tests the oil was essentially constant in
quality with sulfur averaging 2.34%, vanadium pentoxide 0.055%, and
ash (including MgO additive) 0.14%, by weight of the oil.
The sample site was at the outlet of the west induced draft
fan where duct velocity ranged between 19 and 41 m/sec, depending
on MW output and furnace conditions. The ratio of probe velocity/
duct velocity ranged widely, but most tests were performed between
0.7 and 1.1. The sample site was downstream of an always-operating
electrostatic precipitator which had a calculated collection effi-
ciency between 58% and 71%, depending on gas velocities and dust
burden in the ESP inlet gases. All ESP collections were re-injected
back into the radiant furnace, but not continuously. Some gas sam-
ples were taken during re-injection, but most were not.
Filters were tared quickly after 24 hours in a desiccated con-
tainer. After sampling, the filters were treated exactly as in the
pre-tare procedure, the increase in weight representing the particu-
lates collected during sampling. A sizable area was cut from the
filter, precisely measured in area, digested in mild HC1 and filtered.
The filtrate was made ammoniacal to precipitate the R203 group
and filtered. The filtrate was adjusted to the M.O. endpoint, treated
with bromine and boiled. Ten percent barium chloride was added drop-
wise to precipitate all sulfates as barium sulfate which was filtered,
ignited, and weighed.
Some of the more significant observations were as follows:
1. At constant 185 MW load and constant excess furnace
air, burner tilt depression from +15° to +5° caused
a reduction of 10% in particulate emissions. The
concentration declined from 53.03 to 47.87 mg/m3, and
the mass declined from 32.02 to 28.90 kg/hr.
247
-------
Table 1. Total Suspended Particulates and Sulfates
on 0.3% S Oil
Filter
#
468
469
471
472
473
474
475
476
TSP
521.3
392.0
567.5
575.1
811.6
769.6
645.5
732.7
Total
S04
mg
161.2
117.7
163.4
175.1
190.2
190.9
164.4
185.2
Sample
Volume
*m3
22.55
15.99
32.32
28.62
21.62
22.21
22.58
18.64
TSP
mg/m3
23.11
24.52
24.34
20.09
Average
37.54
34.65
28.59
39.31
Average
S04
mg/m3
7.15
7.36
7.01
6.12
6.91
8.80
8.60
7.28
9.94
8.66
*At 20°C and 76.0 cm Hg
248
-------
2. One of the causes of high particulate emissions was
a rapid increase in MW load during testing.
3. Degree of boiler dirtiness seemed to have an important
effect on both particulates and sulfates. See Figure
3 for the effect on concentration of these materials
in mg/m3. See Figure 4 for the effect on mass emitted
per unit time in terms of kg/hr.
4. Approximately 12% of suspended solids were deposited in
the probe before the filter with no apparent selectivity
as to chemical components. After five consecutive
samplings with a freshly-cleaned probe, the probe was
washed and found to contain 12% of the mass accumulated
on the five filter pads. The data in this paper have
not been adjusted for this factor.
5. Approximately 145,000 kg of total ash input were needed
to put the P.J. #4, 185 MW boiler into a dirty condition
which would require washing.
6. Only about 0.1% of the input carbon showed in stack
gases. Only 10% of input metals showed in stack gases
(Table 2).
7. We can, therefore, make a general observation on total
sulfate particulates. If we assume all the sulfur is
converted completely to S04 (sulfate) in 2.4% sulfur oil,
we find only about 0.3% of this in the stack gases. We
find twice this, or 0.6% conversion on 0.3% sulfur oil,
making the same assumption as on the 2.4% sulfur oil.
See BNL data on LILCO units in these proceedings for
agreement even though the methods are different.
8. So far, there is no significant difference in artifact
sulfate between Teflon-backed and glass fiber filters.
We will examine these materials further, but we prefer
the glass fiber filter at the moment because of its
low carbon content. The carbon content of particulates
is important to LILCO, and the very high carbon blank
in Teflon-backed filters tends to make our carbon
findings suspect. GMW 810 Glass Fiber Filters, General
Metal Works, Inc., Cleves, Ohio, is identification of
the LILCO glass fiber filters. Pallflex, Model
TX40HI-20 identifies the Teflon-backed filters used.
249
-------
50-|
* @ 20 C & 76cm Hg
40 60
% Dirty
80
100
Figure 3.
ESP outlet particulates vs. boiler dirtiness:
mg/m @ 185 MW, #6 oil containing 2.4% S &
0.14% ash.
250
-------
40-i
kg
per
hr
30-
20-
10-
Sulfates
i
20
I
40
t
60
I
80
100
% Dirty
Figure 4. ESP outlet particulates vs. boiler dirtiness:
kg/hr. @ 185 MW, #6 oil containing 2.4% S &
0.14% ash.
251
-------
9. It appears there is less-than-linear reduction in
sulfate particulate when fuel oil sulfur is reduced
substantially, say from 2.4% to 0.3%. Table 3 shows
the effect on two LILCO units of the same size, one on
2.4% sulfur and 0.14% ash, the other on 0.3% sulfur
and 0.04% ash. High sulfur was burned at furnace excess
O2 of about 0.7%, low sulfur at about 1.0% excess 02.
The high sulfur unit had an ESP, the low sulfur unit
a cyclone collector. Both units operated at 185 MW
gross.
Table 2. Percentage of Input Metals Retained in System
F 74% to 97%
Cu 81% to 90%
Ni 89% to 93%
V 84% to 96%
Mg 98% to 99%
Na 77% to 94%
o
Table 3. Sulfate Emissions in mg/m
Boiler Conditions
Clean
50% dirty
100% dirty
2.4% S Oil
10.0
18.0
28
0.3% S Oil
-
7.8
-
REFERENCE
1. O'Neal, A. J., Jr., H. Cowherd, and D. J. Hassebroek.
Experimental Incineration of Boiler Internal Cleaning
Solvent at Long Island Lighting Company. Combustion,
August 1977.
252
-------
Inorganic Compound Identification by Fourier
Transform Infrared Spectroscopy
Robert J. Jakobsen
R. M. Gendreau
William M. Henry
Battelle-Columbus Laboratories
Kenneth T. Knapp
U. S. Environmental Protection Agency
ABSTRACT
EPA-sponsored work at Battelle has led to a method which
permits the identification of inorganic compounds, even
in complex mixtures. Development of this needed capa-
bility has enabled us to identify specific sulfates and
oxides in both coal and oil fly ash emission samples.
This technique is based on both the sensitivity and the
data handling capabilities of Fourier Transform infrared
systems along with the proper preparation, handling, and
conditioning of samplies and reference standards. The
technique could aid in the development of compliance regu-
lations by providing the ability to identify and measure
specific compound emissions.
INTRODUCTION
Vast tonnages of particulates are emitted annually from
sources using or processing fossil fuels. These fossil fuels are
nearly totally inorganic species, and surprisingly little is known
as to their specific chemical nature. Intelligent health effects
testing and data interpretations depend on such knowledge, as do
studies of control process effects. Past chemical analyses of
fossil fuel emissions mainly have consisted of elemental determina-
tions of metals and anions with some compound Identification pro-
vided by use of the limited capabilities of x-ray diffraction.
While conventional dispersive infrared spectroscopy has been
widely used for the identification of organic compounds, its use
for inorganic identifications was mostly limited to the detection
253
-------
of certain anions. This limited use was especially true for highly
complex inorganic mixtures such as coal and oil fly ash emission
samples. Infrared spectroscopy, as well as most analytical tools,
was rarely used for inorganic compound speciation and never when
the speciation involved different cations with the same anion
(i.e., identification of individual compounds in a mix of inorganic
sulfates).
However, the advent of commercial Fourier Transform infrared
systems (FT-IR) provided analytical spectroscopy with both extra
sensitivity and extra data-handling capabilities. We, therefore,
began a program to investigate the use of FT-IR for inorganic
compound identification. This paper reports the first results of
this investigation which can be briefly summarized as follows:
(1) A method has been developed which permits the identifi-
cation of inorganic compounds even in complex mixtures.
(2) Using this method, we have been able to identify speci-
fic sulfates in both coal and oil fly ash emission
samples.
This technique is based on the use of FT-IR, the proper pre-
paration and handling of both samples and reference standards, and
the assistance of elemental chemical analyses.
An explanation of the differences between FT-IR and conven-
tional dispersive infrared spectroscopy is not germane to the pur-
poses of this paper. However, it is necessary to emphasize that
the interferometer of the FT-IR systems provides great sensitivity,
and the dedicated computers of FT-IR systems permit both storage
of spectra and the capability to subtract spectra. Thus, as will
be discussed later, reference standards can be prepared in a
variety of ways, and the spectrum of each preparation can not only
be run, but it can be stored for future uses (for comparison with
sample spectra or in subtraction routines). Likewise, samples can
be prepared in several ways with each sample spectrum being saved.
Equally important to a storage capability is the capability to
subtract spectra. Subtraction of spectra both enhances the
ability to detect small differences and can be used to remove
unwanted absorption bands. Such infrared absorption bands often
mask bands of other components and, when removed, permit additional
identifications.
Instrumental sensitivity is also important to inorganic com-
pound speciation. This is especially true for opaque samples such
as the coal fly ash emissions in which there are sizeable amounts
of elemental carbon. In addition, sensitivity is also needed for
the observation of small differences by subtraction of spectra.
254
-------
REFERENCE STANDARDS AND SAMPLES
As stated previously, a most important aspect of our method
development has been the preparation and/or handling of both
samples and reference standards. Such reference standards are a
necessary part of any method based on infrared identifications;
because the spectra of the reference standards or known compounds
are used to identify components of the sample either by direct
comparison of spectra or through the subtraction routines. Thus
it is important that the reference standards be in the same
physical and chemical condition as the sample. Figure 1 demon-
strates how easily a standard or sample can change and shows how
difficult it can be to have both the standard and the sample
in the same state or condition. This figure shows spectra of
f.igS04 „ 7H20 (.run as KBr pellets) v»hen the standard came from a
freshly opened bottle (Figure 1A) and after the sample stood
overnight in air (Figure 13). The spectral differences probably
reflect changes in hydration state, but they could also represent
changes in crystalline structure. Of most importance, however,
is the fact that these changes drastically alter the spectrum. If
a sample containing MgSC4 was in a condition such that it gave one
of the spectra of Figure ^ and the reference standard gave the
other spectrum, identification of MgS04 in the sample would be
difficult, if not impossible. Thus, it is essential to have
"in storage or file spectra of reference standards in as many
states or conditions as possible. The alternative to this is to
find a way of preparing both standards and samples in a i-epro-
ducible manner.
In order to evaluate whether such reproducibility could be
achieved and what conditions would bring this about, we prepared a
physical mixture of 13% NiS04 . GH2O, 41% MgS04 . 7H20, and 46%
YOSC4, Infrared spectra of each of these compounds are shown in
Figure LA, B, and C, respectively. These reference spectra were
run early in the program and stored in the memory of the FT-IR
system. Later when the physical mixture of these compounds was
prepared, the reference spectra were recalled from memory, and
the computer was used to generate a spectrum of the mixture.
This computer-generated spectrum is shown in Figure 3C. The
spectrum of the actual physical mixture can be seen in Figure 3B.
A comparison of the spectra of Figures 3B and 3C shows that the
computer-generated mixture spectrui; and the spectrum of the actual
physical mixture are similar but not identical. This again illus-
trates the difficulty in having even reference standards in the
same or reproducible state. As it turned out, the compounds used
to obtain the reference spectra for the computer-generated spec-
trum and the compounds used for the physical mixture came from
different bottles or different nanufacturers. Thus, even though
255
-------
1X3
Ul
O)
3000
2000
1000
Figure 1. Infrared spectra of NiS04 • 6H20 for freshly prepared
sample and one day old sample.
-------
300Q
2000
1000
Figure 2. Infrared spectra of (A) KiS04
7h20; (C) VOS04.
5H20; (B) MgS04
257
-------
3000
2000
1000
Figure 3. Infrared spectra of a mixture of NiS04 • 6H20
(13%), MgS04 • 7H20 (41%), and VOS04 (46%) (A)
after dissolving H20 and drying; (B) actual
physical mixture; (C) computer generated
spectrum of mixture (from components in
Figure 2).
258
-------
the spectra should have been identical, some of the sulfates
were in different physical or hydration states. However, it
should also be observed in Figures 3B and 3C that the computer
can be used to determine if the reference standards are in the
same physical state. This variation in physical state has been
the main problem in past work attempting to use infrared spec-
troscopy for inorganic compound identification and the ability
to detect such variations in the first step in successful
inorganic speciation.
The physical mixture, described above, was dissolved in water,
the water was evaporated, and the residue was dried. A spectrum
of this dried residue is shown in Figure 3A. Note that there are
significant changes in the spectrum of the physical mixture as a
result of being dissolved in water (compare Figure 3A and 3B).
Not only are there new infrared bands, but the strong S-0 vibra-
tion (1050-1200 cm~1) has broadened considerably. Thus, the
physical state and the infrared spectra of inorganic compounds can
be considerably altered as would be expected as a result of being
dissolved in water. This is not always the case; we have found
that some inorganic compounds remain unchanged after being
dissolved in water. In either case it is important to know if
being dissolved in water affects the compound for reasons to be
described in subsequent discussion. Therefore, it is necessary
to get spectra of the reference standards before and after being
dissolved in water. The same is true for fly ash samples, but
here we can use the computer to determine if dissolving in water
has altered the physical or chemical state of the samples.
Figure 4 shows spectra of the same physical mixture after
being dissolved in water and after baking in argon at various
temperatures for eight hours. There are some minor differences
in band intensity between the unbaked sample (Figure 3A) and
the baked samples, but there are virtually no differences be-
tween the samples baked at 80°, 120°, and 350°C. The major
spectral difference between the baked and unbaked samples ap-
pears to be a sharpening of the infrared bands in the baked
samples.
Thus, we establish a procedure for handling samples as
follows:
(1) Obtain infrared spectra of the fly ash samples
before and after heating in argon at 350°C for
eight hours.
(2) Do a water extraction of the sample, separate the
water soluble and the water insolubles, and dry
each fraction.
259
-------
3000
2000
1000
Figure 4. Infrared spectra of a mixture of NiS04 * 6H20,
MgS04 • 7H20 and VOS04 after heating for 8
hours in argon at (A) 350°C; (B) 120°C; (C) 80°C.
260
-------
(3) Obtain infrared spectra of the water solubles and
the water insolubles before and after heating in
argon at 350°C for eight hours.
Essentially the same procedure is followed for the refer-
ence standards, with the obvious exception that for a pure com-
pound there is only a water soluble or water insoluble fraction.
It is necessary to follow this procedure for the reference
standards and permanently store the spectra of each reference
standard under all the conditions of the procedure. However,
we have not been able to save all these spectra due to the limited
storage capacity of our current computer. This inability to save
all the needed reference spectra has been the major limitation to
our work to date, but this will be alleviated this summer when
we acquire a new FT-IR system with unlimited storage capacity.
The above-listed procedure was selected for the samples
because separating the water solubles (most sulfates) and the
water insolubles (oxides and silicates) aids in the interpretation
of the infrared spectra. This separation aids the spectral inter-
pretation by removing interfering absorption bands. Each sample
and each sample fraction are run both heated and unheated in order
to follow changes due to the heating and because the heating (or
baking) tends to put samples and reference standards in a repro-
ducible physical or hydration state.
Some of these techniques are shown in Figures 5 and 6. These
figures show spectra of a Picway coal fly ash and of the fractions
obtained from our fractionation procedure. Figure 5 shows the
fractions before heating or baking while Figure 6 shows the frac-
tions after baking in argon at 350°C. Note that the spectrum
(Figure 6C) of the water soluble baked fraction is considerably
different from the other spectra in Figure 6. This brings up
the question of whether dissolving the sample in water altered
the physical state of the fraction. In Figure 7A, the spectrum
of the water soluble fraction shown in Figure 6C is repeated,
while Figure 7B shows the subtraction of the total sample
(Figure 6A) and the water insoluble fraction (Figure 6B). This
subtracted spectrum represents a computer-generated spectrum of
the water soluble fraction. Comparison of Figures 7A and 7B show
that they are virtually identical, establishing that the water
soluble components were not altered by being dissolved in water.
Thus, the computer can be used to determine when there are changes
in the physical state of the samples due to the extraction
procedures.
261
-------
1900 1600 1300 1000 700 400
Frequency, cm
rl
Figure 5. Infrared spectra of Picway coal fly ash (no
heating) of (A) total sample; (B) water insol-
uble fraction; (C) water soluble fraction.
262
-------
1900 1600 1300 1000
Frequency, cm1
700 400
Figure 6. Infrared spectra of Picway coal fly ash
(heated at 35C°C for eight hours) of (A)
total sample; (B) water insoluble fraction;
(C) water soluble fraction.
263
-------
CM1 1600
1000
400
Figure 7. Infrared spectra of Picway coal fly ash
(heated at 350°C for eight hours) of (A)
water soluble fraction; (B) subtraction of
total sample minus water insoluble fraction
(i.e., computer generated spectrum of water
soluble fract ion).
264
-------
OIL FLY ASH RESULTS
Using the fractionation procedure described above, six oil-
fired fly ash samples were analyzed. The results of this analysis
are shown in Table 1. As can be seen in Table 1, a large percent-
age of each oil fly ash is water soluble, and this fraction is
mainly composed of sulfates. In Table 1, the percentage of water
soluble components, the percentage of NH4 , and the percentage
of the four most abundant elements are listed for each fly ash.
The infrared results for each fly ash are also listed in Table 1
with the most abundant sulfate (based estimations from infrared
band intensities) listed first, the second most abundant next,
etc. It can be seen in Table 1 that this infrared method identi-
fied several sulfates in each fly ash, and these identifications
were, in general, supported by the elemental analyses. Thus,
FT-IR coupled with elemental analyses (especially to guide the
initial subtractions) can identify individual sulfate compounds
in a mixture of sulfates.
COAL FLY ASH RESULTS
The water soluble content of the coal fly ash samples is, in
general, much lower than for the oil fly ash samples. Thus, while
the water solubles are still important, the water insolubles
assume a much greater importance than for the oil fly ash samples.
Figure 8 shows infrared spectra of the water soluble portion
of Millcreek (Figure 8A) and Picway (Figure 8B) coal fly ashes.
Note that the spectra are very similar and even though there are
several informative infrared bands in the 600-700 cm~~1 region,
it is difficult to identify individual sulfates from such spectra.
All that can be said is that there are large amounts of CaS04
and/or Fe2(804)3 present. However, having the capability to
subtract infrared spectra can be very useful, especially when the
two spectra are very similar. Figure 9 shows the subtracted
spectrum of the Millcreek and Picway fly ash samples (Figure 8B)
and a scale expanded version of this subtraction (Figure 8A). The
absorption band pointing downwards (1210 cm"1) clearly indicates
the presence of A12(S04)3 in the Picway sample. Since the spectra
(Figure 8) are so similar, A12(S04)3 must also be present in the
Millcreek sample, but there is more of it relative to the other
sulfates in the Picway fly ash. Also the ban^s in the 600-700 cm~1
region indicate the presence of both CaS04 and Fe2(S04)3 and this
indicates there is more of these components in the Millcreek
sample than in the Picway sample since the bands in the 600-700
cm region are pointing upwards. Thus, from the subtracted spec-
tra, A12(S04)3, CaS04, and Fe2(S04)3 can be identified in both
265
-------
Table 1. Infrared Results for Oil Fly Ashes
Oil
#1
#2
#3
#4
#5
#6
Fly Ash Sol. NH4
Ea 58 0 NH.
IR
E 230. 1NH4
IR
E
IR
E 72 0.8
IR
E 98 0.2
IR
E 83 7.3
I R NH H SO
(NH4)S04
% Indicated Elements
4.7Mg 3.9Na l.ONi O.GCa
MgS04 Na2SO4 ~ CaS04
2.2V 1.2Mg 0.6Ni O.SNa
VOSO4 MgSO4 NiS04
VOSO. Na0SO.. CaSO.
424 4
9.OV 5.0Mg l.INi 0.5Na
VS04b MgSO4 NiS04 VOS04
Other0
12.9V 2.7Mg 2.3Ni 2.0Na
VOSO4 MgS04 NiS04
2.4Mg 0.8V 0.4Fe O.SNi
MgSO4 — — NiSO4
CaS04
voso4.
= Elemental Analysis
is used to indicate a vanadium sulfate other than VOS04
For fly ash #4, a 5th unidentified sulfate has been detected
266
-------
2000
1600
CM1
1200
800
400
Figure 8. Infrared spectra of water soluble fraction of
coal fly ashes of (A) Millcreek; (E) Picway.
267
-------
1600
1200
CM
H
800
400
Figure 9. Subtracted infrared spectra (water soluble
fractions of Millcreek minus Picway) with
(A) scale expanded; (B) no scale expansion.
268
-------
samples; the amount of A12(S04)3 relative to the other sulfates
is higher in the Picway sample; and the amount of CaSCU and Fe2
(804)3 is higher in the Millcreek sample. This is substantiated
by the elemental analysis which is given in Table 2. Here it can
be seen that even though the total quantity of soluble material
is greater in the Millcreek sample, the total amount of possible
Al2(S04)3 relative to the other possible sulfates is greater in
the Picway fly ash.
Table 2. Elemental Analysis for Coal Fly Ash Samples
Coal
Fly
Ash
Picway
Millcreek
H2O
Sol.
14
35
Percent
0.6A1
1.9Fe
of Total
E
0.6Fe
1.6A1
0.2Ca
1.6Ca
The spectrum of the water insoluble fraction of the Picway
coal fly ash is shown in Figure 6B and will not be repeated here.
The spectrum of the Millcreek insoluble fraction is almost identical
to the Picway fraction except that the Millcreek spectrum shows a
greatly reduced 560 cm"1 (Fe304) infrared band. Thus, the spectra
of the insoluble fractions show large amounts of Si02 and Fe304
with more Fe304 in the Picway sample. No A1203 was detected in
either sample in spite of the fact that relatively large (about
10%) amounts of Al were found in each sample. This probably
indicates that the Si02 component contains an aluminum iron sili-
cate. We have not obtained as yet many reference spectra of sili-
cates; thus such identifications are not possible at this time.
GLASS MELTS
The spectra of the water insoluble fractions of the coal fly
ashes not only raise the questions of obtaining reference spectra
for the identification of silicates but also raise the question
of the amorphous (glass) versus crystalline content of these frac-
tions. This can be especially critical in the area of inorganic
compound identification because x-ray diffraction is dependent on
crystallinity to be effective. Thus a technique for compound
269
-------
identification in glassy samples is vitally needed. Our work in
this area is just beginning, so only preliminary data are listed
below.
Table 3. Composition of Glass Melts
Glass %
G-l
G-2
A1203
51
40
Fe203
20
15
SiO2
29
45
Two oxide samples, of the composition shown in Table 3, were
prepared. For each sample, the oxides were mixed thoroughly,
melted at high temperatures, quenched, and ground to less than 300
mesh. Infrared spectra of each sample were obtained both before
and after melting. The spectra before melting are shown in Figure
10, while the spectra after melting are shown in Figure 11. The
spectra of the two mixes before melting are quite similar (Figure
10) with the major_difference being the increased band intensities
r.t 790 and 1100 cm in G-2. These bands are both due to Si02. A
comparison of the height of the 790 cm-1 band (baseline corrected)
indicates 50%-75% more SiO2 in G-2 than in G-l. That this is true
can be seen in Table 3 which gives the actual composition of the
samples. A series of subtractions of G-2 minus G-l were performed
to determine which sample contained a higher Fe203 to A1203 ratio.
Very little difference was observed in these subtractions which
led to the conclusion that the ratio of Fe2Oa to A12O3 was about
equal in the two samples. Reference to Table 3 indicates that al-
though the absolute amounts of Fe2O3 and A12O3 vary between the
two samples, the Fe2C>3/Al203 ratio remains nearly constant.
Figure 11 shows the spectra of the two samples after melting.
Note that these spectra are completely different from the corre-
sponding unmelted spectra. This probably reflects the extreme
treatment the samples received. NBS standards of SiO2 which are
either totally crystalline or totally amorphous do not show the
drastic spectral changes seen between Figures 10 and 11. In spite
of the major differences upon melting, we can detect (Figure 11)
bands at 1110, 930, and 800 cm"1 in both melts. The 1110 cm'1
band is stronger in G-2 and a 560 cm"1 band appears in G-l. The
270
-------
1600
1200
800
400
CM
Figure 10. Infrared spectra of synthetic oxide mixtures
before melting for (A) C--1; (E) G-2.
271
-------
1600
1200
CM1
800
400
Figure 11. Infrared spectra of synthetic oxide mixture
after melting for (A) G-l; (B) G-2.
272
-------
1110 cm"1 band reflects the Si-0 content of the samples and shows
that G-2 has more Si02 than G-l (also note SiO2 content in Table
3). The 560 cm"1 likely means that we have more combined Fe203
and A1203 relative to Si02 in G-l than in G-2. Thus, even though
these samples were subjected to drastic conditions and the melt
spectra are different from most glass spectra, information about
the composition of the glass can still be obtained.
SUMMARY
The results obtained from the use of the methodology described
in this paper demonstrate that FT-IR, when coupled with careful
sample preparation and guided by the elemental analysis, can pro-
vide unique information on inorganic compound speciation. Pre-
liminary information indicates that this information can be ob-
tained on glassy as well as crystalline samples. Relative quanti-
fication of the compounds identified is already possible and being
done but absolute quantification will require getting accurate
extinction coefficients for all the reference standards and for
a variety of conditions. This is not possible at the present
time because of lack of computer storage space for spectra.
Because of the lack of computer storage for both qualitative
and quantitative work, we have not yet fully exploited the poten-
tial of this method for inorganic compound identifications. This
is especially true for silicates and for mixed salts. Until we
routinely begin to acquire far infrared (400-100 cm ) spectra,
our capability to identify many oxides and to identify halogen
salts is limited. In spite of these limitations the FT-IR tech-
nique has provided a unique method for inorganic speciation.
273
-------
Report of the Working Group on
Measurement of Particulate Sulfur
Oxides Emissions
Richard C. Flagan, Reporter
The objective of this working group was to make recommendations
for future research on sampling and analysis of primary sulfur oxide
aerosols. Primary sulfur oxide aerosols are included in particulate
matter present in the plume after initial dilution with ambient air
but prior to any secondary reactions. This includes aerosols pro-
duced by condensation of sulfur trioxide or sulfuric acid vapor.
RECOMMENDATIONS
Research is needed to relate in-stack measurements of gases
and aerosols to the total primary sulfur oxide aerosol which exists
in the plume several stack diameters from the stack exit. -The time
scale for any primary aerosol formation upon initial dilution and
cooling is short compared to the time required for any secondary
oxidation of sulfur compounds (secondary aerosols).
Measurements need to be made to determine the contribution to
particulate emissions resulting from transient phenomena including
start-up, boiler deposit build-up, soot blowing, precipitator
cleaning, shut-down, and malfunctions. Research needs to be done to
determine the adequacy of and, possibly, the development of measure-
ment methods for studying the transient aerosol emissions.
Research is needed to understand the physical and chemical
transformations of particulate matter during and after sampling
the flue gas. This would include studies of sample aging. Input
from the analyst is essential to the understanding of the signifi-
cance of any transformations.
275
-------
Detailed studies of the sulfur aerosol speciation are necessary
for an understanding of the effects of primary sulfur oxide aerosols.
Guidelines need to be developed for the consistent presentation
of aerosol data.
1. It should be recognized that "total" aerosol samples
usually exclude very large particles. The particle
size dependent collection characteristics of the
sampling system should be determined and reported.
2. Where particle size is determined, the definition of
particle size must be clearly stated.
3. Particle size distribution data should not be normalized
as percent versus size because of the inherent bias
introduced by the sampling system collection charac-
teristics. Data presentation such as AM/A log dp versus
log d should be employed.
276
-------
Appendix
PARTICIPANTS AND OBSERVERS
Jeffrey W. Adams
Arthur D. Little, Inc.
Acorn Park
Cambridge, Massachusetts
617/864-5770 x.3036
02140
Aubrey P. Altshuller
Director
Environmental Sciences
Research Laboratory
Environmental Protection Agency
Environmental Research Center
MD/59
Research Triangle Park
North Carolina 27711
919/541-2191
John Bachmann
Environmental Protection Agency
Environmental Research Center
MD/12
Research Triangle Park
North Carolina 27711
919/541-5231
Elizabeth M. Bailey
Division of Environmental
Planning
Tennessee Valley Authority
Muscle Shoals, Alabama 35660
205/383-4631 x.2788
Roy L. Bennett
Research Chemist
Environmental Sciences
Research Laboratory
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3173
Richard K. Chang
Department of Engineering and
Applied Science
Yale University
New Haven, Connecticut 06520
203/432-4470
James L. Cheney
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3172
Harold Cowherd
Environmental Engineering
Long Island Lighting Company
175 East Old Country Road
Hicksville, New York 11801
516/733-4700
Kenneth M. Gushing
Research Physicist
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
205/323-6592
Daryl DeAngelis
Research Engineer
Monsanto Research Corporation
1515 Nicholas Road
Dayton, Ohio 45407
513/268-3411
Russell N Dietz
Chemical Engineer
Brookhaven National Laboratory
Building 426
Upton, New York 11973
516/345-3059
277
-------
James Dorsey
Industrial Environmental
Research Laboratory
Environmental Protection Agency
Environmental Research Center
MD/62
Research Triangle Park
North Carolina 27711
919/541-2557
Brian Doyle
Principal Engineer
KVB, Inc.
246 North Central Avenue
Hartsdale, New York 10530
914/949-6200
Edgar S. Etz
Research Chemist
Center for Analytical Chemistry
National Bureau of Standards
Chemistry Building
Room A-121
Washington, D.C. 20234
301/921-2862
Richard C. Flagan
California Institute of
Technology
MS 138-78
Pasadena, California 91125
213/795-6811 x.1383
William M. Henry
Projects Manager
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
614/424-5210
James B. Homolya
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3085
James E. Howes, Jr.
Senior Researcher
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
614/424-5269
Skillman C. Hunter
KVB, Inc.
17332 Irvine Boulevard
Tustin, California 92680
714/832-9020
Peter Jackson
Central Electric Generating
Board
Marchwood Engineering
Laboratories
Marchwood Southampton
England SO44ZB
Ashok K. Jain
Research Engineer
NCASI
Box 14483
Gainesville, Florida 32604
904/377-4708
Robert J. Jakobsen
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
614/424-5617
Kenneth T. Knapp
Chief, Particulate Emissions
Research Section
Environmental Sciences
Research Laboratory
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3085
278
-------
Arthur Levy
Manager
Combustion Systems Technology
Battelle-Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
614/424-4827
Dale Lundgren
Environmental Engineering
Sciences
University of Florida
Gainesville, Florida 32611
904/392-0846
Ray F. Maddalone
Section Head
TRW Defense and Space
Systems Group
One Space Park 01/2020
Redondo Beach, California 90278
213/535-1458
Richard E. Marland
Office of the Assistant
Administrator for Research
and Development
Environmental Protection Agency
RD 672
401 M Street, S.W.
Washington, D.C. 20460
202/755-2532
William R. McCurley
Research Engineer
Monsanto Research Corporation
1515 Nicholas Road
Dayton, Ohio 45418
513/268-3411
John Nader
Chief
Stationary Source Emissions
Research Branch
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3085
David F. S. Natusch
Professor
Department of Chemistry
Colorado State University
Fort Collins, Colorado 80523
303/491-5391
A. Jack O'Neal, Jr.
Chief Chemist
Electric Production Department
Long Island Lighting Company
P.O. Box 426
Glenwood Landing, New York 11547
516/671-6783
Richard Rhudy
Project Manager
Electric Power Research Institute
Box 10412
Palo Alto, California 94303
415/855-2421
Roosevelt Rollins
Environmental Protection Agency
Environmental Research Center
MD/46
Research Triangle Park
North Carolina 27711
919/541-3171
Arthur M. Squires
Department of Chemical
Engineering
Virginia Polytechnic Institute
Blacksburg, Virginia 24061
703/951-5972
Paul Urone
National Environmental
Investigation Center
Denver Federal Center
Building 53 - Box 25227
Denver, Colorado 80225
303/234-4661
279
-------
Jack Wagman Arthur S. Werner
Environmental Protection Agency Manager
Environmental Research Center Analytical Laboratory
MD/46 GCA/Technology Division
Research Triangle Park Burlington Road
North Carolina 27711 Bedford, Massachusetts 01730
919/541-3009 617/275-9000
280
-------
TECHNICAL -XEFORT UA"; A
'
T NC.
AC
ION NO.
4. nn_i AND SUBTITLE
WORKSHOP PROCEEDINGS ON PRIMARY SULFATE EMISSIONS FROM
COMBUSTION SOURCES
Volume 1. Measurement Technology
7. .AUTHOR(S)
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Kappa Systems, Inc.
1501 Wilson Boulevard
Arlington, Virginia
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory -
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Technical papers on techniques for measuring
August 1978 '
;6. PERFORMING OR G AN i ZA Tl O\- CODE
RTP, NC
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT MO.
1AD712 BC-52 (FY-78)
11. CONTRACT/GRANT NO.
68-02-2435
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/09
primary sulfate emissions
from combustion
sources, presented at a workshop sponsored by the U.S. Environmental Protection Agency,
are compiled in Volume 1 of a proceedings.
The objectives of the workshop were to review and discuss
and problem areas for sulfur oxides emission
current measurement methods
with attention focused on
sulfates, and sulfur-bearing particulate matter; to review and discuss
sulfuric acid,
emission data
from various combustion sources operating under different conditions which include
various pollutant controls, fuel composition, excess boiler oxygen, etc.; and to
delineate and recommend areas in need of research and development effort.
Scientists were invited to present the result of their studies on primary sulfate
emissions. The 3-day workshop devoted one day to measurement technology, a second to
characterization, and a third to critical assessment of the presented papers and
development of summary working group reports
2 days. Thirty-one papers were presented by
on each half-day session of the initial
29 participants on measurements and
characterization. Four working group reports were developed and summarized in the
last day.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
* Air pollution
* Sulfates
* Emission
* Combustion products
* Measurement
* Collecting Methods
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
b.lDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report)
UNCLASSIFIED
20. SECURITY CLASS (This page)
UNCLASS]
FIED
c. COSATI Field/Group
13B
07B
21B
14B
21. NO. OF PAGES
289
22. PRICE
EPA Form 2220-] (Rev. 4-77) oREVious EDI TIC N i s o eso LF TE
-------