United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
EPA-600-2-78-193
August 1978
Research and Development
Chemical Analysis of
Stationary Source
Particulate
Pollutants by
Micro-Raman
Spectroscopy
Interim Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-78-193
August 1978
CHEMICAL ANALYSIS OF STATIONARY SOURCE PARTICIPATE POLLUTANTS
BY MICRO-RAMAN SPECTROSCOPY
Interim Report
by
Edgar S. Etz, Gregory J. Rosasco, and Kurt F. J. Heinrich
Analytical Chemistry Division
National Bureau of Standards
Washington, D. C. 20234
EPA-IAG-D6-F012
Project Officer
John Nader
Emission Measurements and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for pub-
lication. Approval does not signify that the contents necessarily re-
flect the views and policies of the U.S. Environmental Protection Agency,
endorsement or recommendation for use.
ii
-------
PREFACE
The ability to determine the chemical species composition of individual
microparticles is of major importance to the study of the origins and
transformations of particulate matter. This is especially true for the
analysis of particles emitted by stationary sources, such as oil-fired and
coal-fired power plants. The NBS micro-Raman spectrometer is a unique
instrument that can be applied to the solution of problems dealing with the
chemical species characterization of stationary source particulates. This
report gives the results of a preliminary study of the application of the
micro-Raman technique to power plant particulate emissions.
C. C. Gravatt
Deputy Chief
Analytical Chemistry Division
iii
-------
ABSTRACT
The application of laser-Raman spectroscopy to the molecular characteri-
zation of individual particulates from stationary sources is described. This
research was performed using the NBS-developed Raman microprobe. Analytical
capability to identify the principal molecular species present in micro-
particles is demonstrated on the basis of Raman spectra of selected compounds
and materials. Among the inorganic species studied, are sulfates, nitrates,
carbonates and oxides, for which Raman spectra are discussed for single, solid
particles of size down to 1 micrometer. Preliminary results on liquid sulfate
particles generated from sulfuric acid aerosol are presented. The method of
micro-Raman analysis is applied to the characterization of microparticles from
power plant emissions. Raman spectra have been obtained from microparticles
of oil-fired power plant emissions collected by the EPA with cascade impaction
samplers .
Vanadium pentoxide, V20s, has been identified as a major component of
microparticles present in such samples. The presence of certain other vanadium
containing species such as vanadyl, VO2 , and ortho-vanadate, V0$ ) is not
indicated from the results of these measurements. Other Raman spectra show
evidence of crystalline sulfate, SO2, , as a species present in major propor-
tions. However, the exact nature of the associated cation specie (s) has not
been determined. Many of the spectra obtained from fly ash particles show
Raman bands characteristic of polycrystalline graphite. These carbon bands
appear to derive, in the majority of cases, from the presence of carbonaceous
material associated with the particles. The need for further work is indicated
from these exploratory measurements. Recommendations are made as to the scope
and direction for this work.
This report was submitted in fulfillment of Contract No. EPA-IAG-D6-F012
by the Analytical Chemistry Division, National Bureau of Standards, under the
sponsorship of the U.S. Environmental Protection Agency. This report covers
the period April 1, 1976, to March 31, 1977, and work was completed as of
March 31, 1977.
iv
-------
CONTENTS
Preface
Abstract iv
Figures vi
Tables vii
1. Introduction ..... 1
2. Conclusions and Recommendations 2
3. Particle Analysis by Raman Spectroscopy 4
Requirements for Micro-Raman Analysis 4
4. Computerization of the Micro-Raman Spectrometer System 6
Modifications of Instrument Design 6
5. Reference Spectra for Micro-Particle Raman Analysis 8
Inorganic Compounds and Minerals 8
Organic Compounds and Polymers 17
Liquid Sulfate Aerosol 19
6. Characterization of Unknown Particles from Power Plant
Emissions 23
Fly Ash from Coal-Fired Power Plants 23
Particulate Emissions from Oil-Fired Power Plants 25
References. 36
-------
FIGURES
Number Page
1 Raman spectrum of a particle of anhydrite .............. 9
2 Raman spectrum of a particle of sodium sulfate ........... 10
3 Raman spectrum of a particle of calcium fluorapatite ........ 12
4 Raman spectrum of a particle of calcite ............... 13
5 Raman spectrum of a particle of sodium nitrate ........... 14
6 Raman spectrum of a particle of crystalline quartz (a-Sit^) in
urban air particulate dust .................... 15
7 Raman spectrum of a particle of sodium oxalate ........... 17
8 Raman spectrum of a particle of benzoic acid ............ 18
9 Raman spectrum of a microdroplet of cone. I^SOij supported by a
Teflon-coated sapphire substrate ................. 21
10 Raman spectra showing the transformation (from top to bottom) from
a microdroplet of cone. I^SOij to a microparticle of (NHi+^SOit on
reaction with ammonia ....................... 22
11 Optical micrographs of Site A power plant particulate emissions
collected on stages 1 and 2 of Battelle impaction sampler ..... 27
12 Optical micrographs of Site A power plant particulate emissions
collected on stage 3 of Battelle impaction sampler ........ 28
13 Optical micrographs of Site A power plant particulate emissions
collected on stage 4 of Battelle impaction sampler ........ 29
14 Optical micrographs of Site A power plant particulate emissions
collected on stage 5 of Battelle impaction sampler ........ 30
15 Micro-Raman spectrum of a large microcrystal (shown at 312X in
Figure 13) found on stage 4 of the sampled power plant
emissions ............................. 31
16 Raman spectrum of a particle of vanadium pentoxide ......... 32
vi
-------
Number Page
17 Raman spectrum of a particle of vanadyl sulfate 33
18 Raman spectrum of a globular particle found on stage 4 of the
sampled power plant emissions 34
19 Raman spectrum of a particle of calcite (CaCOj) in urban air
particulate dust 35
TABLES
Internal Vibrational Modes of the Sulfate Ion in Four Common
Crystalline Sulfates Measured in the Raman Microprobe 11
vii
-------
SECTION 1
INTRODUCTION
This report under EPA-IAG-D6-F012 agreement summarizes the work performed
for the period April 1976 through March 1977. The purpose of the project is
to investigate the application of Raman spectroscopy to the analysis of the
molecular composition of single, micrometer-size particles in particulate
matter from stationary sources. The scope of this work is to entail activ-
ities in the following principal areas:
1. The partial automation of the NBS micro-Raman spectrometer system to
permit the rapid sequential analysis of multi-particle samples.
2. The acquisition of particle reference spectra of various major
types of stationary source particulates of interest to the EPA, including
sulfates, nitrates and carbonates.
3. The Raman spectroscopic analysis of several representative particu-
late samples provided by the EPA. These samples should have been previously
characterized for elemental composition by the EPA to provide a basis for
comparison with the micro-Raman results.
4. Cooperation with the EPA in the development of stack sampling
methods compatible with the sample requirements of micro-Raman analysis and
consultation in the area of Raman and fluorescence spectroscopy.
, The research conducted under this work agreement addresses itself to
questions concerning the chemical species and the crystalline or glassy
state of particles. A major goal of the project is to demonstrate the
potential of the micro-Raman spectroscopy technique for the molecular
analysis of major constituents in particles from stationary sources, with
emphasis on those from the combustion of fossil fuels.
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
The results presented in this report demonstrate the potential of micro-
Raman analysis for the chemical characterization of fine participates. The
study of single microparticles of well-characterized materials has established
the capabilities and limitations of the present configuration of the Raman
microprobe. Individual particles down to 1 ym in size can routinely be analyzed
with the instrument. In these measurements good detectability is achieved for
the major types of inorganic species suspected to occur in the environment in
microparticulate form. These include the common oxides, carbonates, nitrates,
sulfates, and phosphates as well as a variety of the terrestrial minerals. In
addition, several classes of organic compounds and polymers have been char-
acterized by the micro-Raman technique.
The analysis of small particles is most straightforward when they are
transparent to the exciting radiation. Radiation absorbing, colored particles
may heat as a result of the interaction with the focused beam and give rise to
irreversible modification of the sample.
Important to the success of the Raman characterization of unknown parti-
cles is the availability of particle reference spectra obtained from the meas-
urement of well-characterized source materials. A major effort that has been
expended as part of this work and which will extend into the future does center
on the acquisition of a reference spectra file for microparticle analysis.
Vanadium pentoxide, V20s, has been identified as a major component of
microparticles present in oil-fired power plant samples. The presence of other
vanadium containing species (e.g., vanadyl, VO2 , and ortho-vanadate, V0$~) is
not indicated from the results of these measurements. Other Raman spectra show
evidence of crystalline sulfate, SO2. , as a species present in major propor-
tions. However, the exact nature of the associated cation specie(s) has not
been determined. Many of the spectra obtained from fly ash particles show
Raman bands characteristic of polycrystalline graphite. These carbon bands
appear to derive, in the majority of cases, from the presence of carbonaceous
material associated with the particles.
As has been demonstrated in these measurements, spectra are obtained that
can be correlated with the presence of major constituents when appropriate ref-
erence spectra are available. In cases where particle identification based on
the Raman spectrum is not possible, the application of other microanalytical
methods must be sought to achieve the desired result. To date we have not ex-
plored the full analytical potential of combining the micro-Raman technique with
other methods of particle microanalysis. Identification of an unfamiliar Raman
spectrum is made easier if the elemental composition of the sample is known.
-------
The preliminary results obtained from the micro-Raman analysis of
particles in power plant emissions indicate that further research is required
to characterize such samples. Several problems are perceived that need to be
examined in future work. Among these is the collection of such samples for
micro-Raman analysis. Bulk particulate samples collected on filters or other
supports are least suited for single particle characterization. They require
the isolation of particles by methods that do not alter the integrity of the
particle. The measurement is frequently difficult due to the presence of
contaminants introduced by the method of sampling and through inter-particle
contact. The bulk sampling of particulates provides ample opportunity for
reactions to take place that change the morphology and chemical nature of
micropartides. These events tend to complicate the spectrochemical informa-
tion derived from such samples and frequently place into question the original
identity of the particle under study. The collection of suspended particles
by impaction on the Raman substrate offers the opportunity for unambiguous
sampling of the aerosol. When on-site sampling conditions are optimized, the
sample consists of a deposit of particles of sufficiently low particle
density to exclude the possibility of significant particle modification due to
inter-particle reactions. It is also clear from our analyses of unknown
particles that greater emphasis must be placed on the application of analytical
methods that combine the micro-Raman technique with other micro-analytical
probe techniques (e.g. electron probe and ion probe micro-analysis).
We propose a continuation of the project which should have as its
objective the detailed Raman investigation of power plant particulate emissions
collected on micro-Raman substrates. These studies should seek to optimize the
sample collection from these sources and include in-stack and out-of-stack
sampling of the aerosol. For these investigations to be of greatest utility,
it is suggested that any further work in this area on the part of NBS be
closely coordinated with the parallel efforts underway at the EPA in the
characterization of such samples by other methods.
-------
SECTION 3
PARTICLE ANALYSIS BY RAMAN SPECTROSCOPY
The application of Raman spectroscopy to the analysis of small, single
particles yields chemical information on such samples which cannot be obtained
by other microanalytical techniques (e.g., electron probe and ion probe micro-
analysis) . For a broad range of both inorganic and organic compounds the
technique has the potential of furnishing not only the structural formula of
the molecular species contained in the particle, but in addition may yield
information on the long-range molecular order (i.e., crystalline or glassy
state) of the material. The Raman spectrum that is obtained from such micro-
scopic samples will therefore, in many cases, represent a unique "fingerprint"
of the constituent chemical species and their structural coordination. The
technique can readily distinguish, for example, between sulfate, SO2. , sulfite,
SO?. , either of the corresponding protonated forms, HSO^ or HSOs , and
sulfide, S2 , in a respirable-size particle when these species are a major
constituent. Moreover, the Raman spectrum is sufficiently different for
distinguishing among the various crystalline forms of calcium sulfate, i.e.
gypsum (CaSOit 2H20) , hemihydrate (CaSOi^ %H20) , and soluble and natural
anhydrite
The analysis of molecular ions (such as SO2. , other oxyanions of sulfur,
, etc.) in the atmospheric environment is an area of intense interest and
the focus of many current investigations. Of primary importance is an
increased understanding of the atmospheric chemistry leading to the formation
of "acid aerosol" from sulfur dioxide, S02- This requires determining the
specific chemical nature of atmospheric aerosol and frequently a correlation
of particle composition with particle size. These species, including atmo-
spheric E^SOii present as droplets of the free acid and the half-neutralized
acid, NHifHSO^, have to this date proved difficult to identify by traditional
air sampling methods followed by ionic analyses (i.e. wet chemical methods) of
bulk samples or used in conjunction with microchemical tests for individual
species contained in single, microscopic particles.
REQUIREMENTS FOR MICRO-RAMAN ANALYSIS
The development of the new Raman microprobe was preceded by earlier work
[1] at NBS in which it was demonstrated that the Raman spectrum could be
obtained from single, micrometer-size particles. These results showed that the
Raman frequency and line shape are not affected (from analytical considerations)
by the fact that the molecules of the sample make up a small particle. It was
concluded that the spectra are essentially the same as those observed from bulk
quantities of the material and could provide a basis for the chemical identifi-
cation of small particles present in many forms of particulate matter.
-------
We have discussed in earlier published work [1,2] the experimental
requirements that must be met to permit the recording of the Raman spectrum
from microparticles of size down to 1 y. These criteria are, high spectral
sensitivity to extremely low signal levels, effective rejection of optical
interferences, and appropriate choice of irradiance (power per unit area)
levels placed in a focused laser beam that will not bring about modification or
destruction of the sample by heating or photodecomposition.
The new Raman microprobe developed at the NBS has been constructed to meet
these requirements for single particle analysis. A description of the design
of the instrument and a discussion of its optical and mechanical performance
has been presented in a recent publication [2]. This paper also presents the
results of measurements performed to demonstrate the broad capabilities of the
micro-Raman spectroscopy technique in several areas of application.
-------
SECTION 4
COMPUTERIZATION OF THE MICRO-RAMAN SPECTROMETER SYSTEM
MODIFICATIONS OF INSTRUMENT DESIGN
A complete description of the overall design and configuration of the
Raman microprobe is given in the recent literature [2]. Some modifications
have been made in the design of the instrument to optimize its performance in
the application to the analysis of environmental particulates contained in
multi-particle samples.
Fore Optical System
Initially, a reflecting objective (Beck, 15X, N.A. 0.28) was chosen to
focus the exciting laser beam to a diffraction-limited beam spot of approximate
diameter 2 ym. Because of the design of the Beck objective (which yields a
classical, non-Gaussian beam) the irradiance placed on the sample could only
be reduced below its maximum value by approximately a factor of 20, - the limit
which is imposed by the stable operating range of the laser. In order to
measure absorbing samples, it is frequently necessary to vary the irradiance by
an additional order of magnitude (i.e., a total range factor of approximately
200). This is accomplished most effectively by varying the spot size of the
focused beam, which is not possible with the Beck objective. To extend the
application of the instrument to the routine analysis of radiation absorbing,
temperature sensitive materials, the Beck objective was replaced with a refrac-
tive objective lens (Leitz, 5.6X, N.A. ^0.15) which allows measurements at
reduced irradiance levels. The power throughput for this lens is approximately
2.5 times that of the Beck and it furnishes a minimum spot diameter of
-------
They are replacing the more conventional differential screw micrometers used
to translate the sample stage in the earlier configuration of the system.
The new inchworm translators are driven by an external controller
(Burleigh, Model PZ-502) which is interfaced by the computer allowing for
pre-programmed stepping of the stage.
Spectrometer Computer Interface and Software Development
The micro-Raman spectrometer system has been designed to be interfaced to
a dedicated mini-computer to permit automation of the measurements and
optimized, automatic data acquisition.
The interface of the Nova 1200 computer to the spectrometer system is a
modified Digilab system. Each of its functions have been tested and are now
operative. The interface provides the capability to read the digital output of
the two-channel photon counter into the computer and will replace the present
analog, strip-chart recording system. This allows background-corrected and
intensity-normalized spectra to be obtained. In addition, the interface
permits computer control of the photon counter count time, the setting of the
spectrometer slits, stepping of the wavelength scan, and translation of the
sample stage. Full integration and testing of the operating system is
continuing. Significant improvement in the acquisition and manipulation of
spectral data will result from the computerization of the system.
-------
SECTION 5
REFERENCE SPECTRA FOR MICRO-PARTICLE RAMAN ANALYSIS
In the following are discussed selected examples of both inorganic and
organic compounds characterized as single particles by application of the
Raman microprobe. These materials are of interest since they are expected to
be present in stationary source emissions from fossil fuel power plants.
INORGANIC COMPOUNDS AND MINERALS
Sulfates/Sulfites
Simple (e.g., NaaSOiJ and complex (e.g., Fe(NHit)2(80^)2 ' 6H20) sulfate
salts have been studied as model compounds for the Raman characterization of
crystalline sulfate in airborne particles. Typical of these measurements are
the spectra shown-in Figure 1 and Figure 2. These and all other spectra discus-
sed exhibit approximately the same format. The frequency shifts in wavenumber
units for the Stokes-Raman scattering are displayed on the horizontal axis.
Plotted along the vertical axis is the value of the scattered light intensity
in photon counts observed in a measurement time referred to as the time
constant. The zero of light intensity is indicated by the solid horizontal
base line. In some examples, the Raman shifts characteristic of the particle
are indicated by markers labeled "P" and the bands associated with the sapphire
substrate are marked "S". All spectra have been recorded at room temperature
with an effective resolution of approximately 3 cm 1. Frequency calibrations
were obtained by recording the neon and argon emission lines, providing an
accuracy of ±3 cm l. The spectra of the sulfates are discussed in some detail
to indicate the specificity of the Raman method for qualitative identification.
In each case, the band assignments made are those reported in the literature.
As in the bulk (i.e., as single crystals), microparticles of sulfate salta
and sulfate minerals give very strong and sharp Raman spectra. These have been
reviewed, along with the vibrational spectra (IR data included) of other
classes of inorganic compounds and minerals, in the recent literature [3,4].
Theory predicts four fundamental Raman-active modes of the undistorted, free
SOff ion of tetrahedral symmetry. In the crystalline state, the number and
frequency positions of the vibrational modes associated with the SOff ion will
depend on the number of ions in the unit cell (primitive) and the local
symmetry about the So£ ion. Most characteristic of the solid sulfates is the
strong, symmetric stretch (vj) near 1000 cm l. In the sulfates we have
examined, this is a single, sharp line. The other fundamental modes are of
weaker intensity and are resolved as either doublets or triplets in the spectra
of the sulfate microparticles. The spectrum in Figure 1 of a microparticle
-------
SampleCaS04 Size l.0um
Substrate A
(°t sample)
diameter
Laser X0 514.5 nm
power 40 mw
beam « | 3
Spectral slit width 3
Time constant |
PINHOLE 140pm
Scan rate j gy
Full scale
cm '
s
cm"' /s
counts
1000
CM
-i
5OO
0
Figure 1. Raman spectrum of a particle of anhydrite.
of anhydrite (CaSOi^) shows many of the expected modes - marked "P" -
characteristic of the internal vibrations of the sulfate group in the crystal.
The bands marked "S" are most of the expected modes of sapphire and appear in
the spectrum at 378, 418, 432, 451, 578, 645 and 751 cm l. Analysis shows
there are nine Raman-active internal modes of the SC>£ ion in CaSC^.
%
Spectra of similar quality have been obtained from microparticles of
Na2SOif in sizes down to 2 ym. The spectrum in Figure 2 is representative of
these results. In this measurement, the hygroscopic particle is encapsulated
by a thin film of low fluorescence immersion oil to prevent its modification
by moisture in the air. Raman data have been reported for this salt and the
-------
4₯»**w^w
Sample Na2S04Size3x2um
Substrate AI203-Oil
Laser X0 514.5 nm
power 63 mw (a* sample)
beam *- | 3 Vm diameter
Spectral slit width 5 cm~'
Time constant | s
PINHOLE 140 pm
Scan rate |.67 crrTvs
Full scale _3 counts
Substrate
only
A
»A<»A««V*y \
. _^v
I I I I I I I I I I I I I
1000 CM_, 500 0
Figure 2. Raman spectrum of a particle of sodium sulfate.
bands observed in the spectrum of the particle (c.f. Table 1) are consistent
with the literature values. Included also in Figure 2 is the spectrum of
the substrate (i.e., sapphire), recorded by simply moving the particle out
of the focal spot of the beam. A third example is the spectrum of a small
particle of (NHi^SOij (c.f. Figure 4 of Ref. 2.) The internal modes of the
NH^ ion in this salt have been reported to be centered around 3124 cm *
(vi), 1669 cm~1(v2), 3137 cm'1 (v3), and 1429 cm"1 (v^). Of these the
bending modes appear in the spectrum of the particle as broad features around
1670 cm~2 and 1430 cm"1. Each mode of the NH^ ion is broader than the
corresponding one of the S0§~ £0n. The various bands observed below 400 cm"1
arise from external modes pf the crystal. The strong band at V73 cm"1 has
been assigned to a translatory mode of the SOg ion. The weaker bands in the
10
-------
region 150-210 cm 1 are attributed to translatory modes of the NHi,. ion. The
spectra presented in these figures show good signal-to-noise for the fundamen-
tal sulfate modes in these microcrystals. High irradiance levels were placed
on the microsample in each measurement without adverse effects on the particle.
Spectra of similar quality were recorded from microparticles of other common
sulfates (e.g., Na^Oij, PbSOi^ and double salts containing sulfate (e.g.,
Fe(NHit)2(SOit)2 ' 6H20).
The Raman shifts observed from microparticles of several common crystalline
sulfates are given in Table 1. Comparison of the frequencies for the sulfate
ion in anhydrite and in gypsum, CaSOi^ 2H20, shows that the hydrated and non-
hydrated forms can readily be distinguished.
The selected data presented in the table indicate the ability to spectro-
scopically discriminate among the various crystalline sulfates in micro-
particles. Raman data of this kind are available in the literature for many
inorganic compounds and minerals. The Raman shifts given for the free SOfj
ion correspond to the frequencies of the fundamental modes of the ion in
solution. The half-widths of Raman lines of dissolved species are generally
broader, and the bands are less intense (concentration effect) than in the
solid phase. These observations are relevant to the micro-Raman characteriza-
tion of liquid microparticles, as will be shown.
TABLE 1. INTERNAL VIBRATIONAL MODES OF THE SULFATE ION IN
FOUR COMMON CRYSTALLINE SULFATES MEASURED IN THE RAMAN MICROPROBE
Compound SOff Frequency Shift,
cm
vj v2 V3
4- cw ^s
symm. stretch symm. bend asymm. stretch asymm. bend
Na2SOit
(NHtl)2SOit
CaSOit
CaSOi,. 2H20
Free S0f]~ ion
997
976
1018
1006
981
472,455
452,447
497,418
493,413
451
1162,1135,1110
1062,1075,1089
1060,1128,1108
1142,1135,1116
1104
661,640,628
617,612
674,628,608
670,623
613
Sulfamic acid (amidosulfonic acid) H2NS03H, has been studied in the Raman
microprobe. With this material the primary interest lies in the spectroscopy
of the S0§~ grouping as compared to that of the SOf] grouping in solid
sulfates. The acid is suspected to be an important constituent of continental
aerosols, or precursor to atmospheric ammonium sulfate particles. Parallel
interests center on the spectroscopic_characterization of other crystalline.
materials containing the sulfite, S0§ and bisulfite, HS03 , species; this in
11
-------
view of the toxicological importance of tetravalent sulfur in respirable
particulates and the significance of heavy metal-sulfite complexes in the
atmospheric environment. Preliminary results indicate good detectability for
the common (i.e., Na , NHi* salts) sulfites down to particle sizes of a few
micrometers.
Phosphates
Phosphates give very strong, sharp Raman sgectra. Similar spectroscopic
arguments exist for the phosphate grouping, P0$ , as have been presented for
the so£ anion in crystalline solids. The Vi symmetric stretch of the phosphate
ion is normally the strongest and sharpest band in the spectrum and often a
good guide to identification.
Micro-Raman measurements have been made on a well-characterized sample of
calcium fluorapatite, CasCPOi^sF. Excellent Raman spectra have been obtained
for single particles down to 2-3 ym in size. These spectra show one-to-one
correspondence with the Raman data reported for bulk samples.
Figure 3 shows a typical spectrum, obtained from a small particle of
calcium fluorapatite. Since there are two P0^~ groups in the unit cell, theory
predicts 15 Raman-active internal modes, most of which are resolved in the
1*.
Ccu(POAF 5x5
Sample ° * rSize ym
Substrate LI F
Laser 5t05l4.5nm
power 50 rnW (at sample)
beam~ JO Mm diameter
Spectral slit width 3 cm"1
Time constant O.I s
Pinhole 140
Scan rate J.67
Full scale ' »
I03
pm
cnrT'/s
counts
I
I
f
1000
j/L
cm
-i
0
Figure 3. Raman spectrum of a particle of calcium fluorapatite.
12
-------
spectrum of the microparticle. The multi-component bands arising from
these modes are of much weaker intensity than the v\ symmetric stretch observed
at 965_cm *. These split_fundamentals are centered around 430 cm 1 (V2> ,
600 cm 1 (vij and 1050 cm l (vs). These results indicate spectroscopic
detectability for mineralogical phosphates in particles of size down to one
micrometer.
Carbonates
Microparticles of single-crystal calcite, CaCQ$, have been studied in the
Raman microprobe in order to establish detectability of microcrystalline
calcite. Good spectra were obtained for single particles approaching 1 ym in
linear dimensions. A representative result is the particle spectrum of
calcite shown in Figure 4. Parallel measurements performed on micrometer-size
particles of chicken eggshell (i.e., CaCOs) have yielded spectra consistent
with that shown in Figure 4.
Somple CdCO, Size '?m
Substrate ALOj
Loser V5tC5nm
power 40 mw (at sample)
beam ~ 2.0um diameter
Spectral slit width 3 cm"'
Time constant 0,4 s
Pinhole I40um
Scan rote | g-^ c
' j
Full scale
2000
3000
-i r
cm
1000
Figure 4. Raman spectrum of a particle of calcite.
Carbonates in general are recognized by the appearance of the strong
symmetric stretch vj near 1100 cm 1. For calcite, this internal vibrational
mode has shift 1088 cm 1. The asymmetric stretch V3 of the CO^ anion is
predicted at 1432 cm * and is barely resolved in the spectrum of the micro-
particle, in part due to the moderately high background signal level in this
spectral region. The asymmetric bend vii appears with Raman shift_714 cm *,
and the expected lattice (external) vibrations are seen at 283 cm 1 and
156 cm *. The symmetric bend V£ of the carbonate ion is Raman-inactive for
solids of the calcite structure.
Nitrates
Several nitrates have been characterized as microparticles. The spectrum,
shown in Figure 5, of a small particle of single crystal NaNOs, is typical of
these results. Sodium nitrate has the calcite (CaCQ%) structure and five
13
-------
0 ~7y-2 A
SampleNaN03 Size * ym
Substrate Li F
Laser X. 514.5 nm
power 50 mW (at sample)
beam ~ 11 urn diameter
Spectral slit width 3 crrH
Time constant 0.4 s
Pinhole 140 prn
Scan rate \,QQ cm~'/s
Full scale ' 3 counts
10
1X2.3
.d L
1000
cm"
0
Figure 5. Raman spectrum of a particle of sodium nitrate.
Raman-active vibrations are expected for this crystalline solid. The frequency
shifts corresponding to these normal modes are seen in the spectrum of the
microparticle. Among the three allowed internal modes, the nitrate symmetric
stretch vj gives rise to the strong, sharp band at 1068 cm 1. Other character-
istic lines have shifts 1385, 724, 185 and 98 cm 1, where the latter two
represent the expected lattice vibrations. These values are consistent with
those given in the literature. The position of these bands in the alkali-metal
nitrates is sensitive to the associated cation and the crystalline structure of
the solid. In the case of KNOs, for example because of the higher mass of
the cation the two corresponding lattice modes are shifted appreciably (by
^55 cm *) toward lower frequencies. The nitrate symmetric stretch v\ in
crystalline KNOs falls at 1048 cm"1 and is equally strong in intensity. The
NaNOs particle in the spectrum of Figure 5, it is noted, is supported by a LiF
substrate. There is no first-order scattering from this material and back-
ground levels are seen to be extremely low in the absence of any broad band
fluorescence from these substrates.
Microparticles of NH^NOs and Pb(NC>3)2 have also been measured in the
microprobe furnishing good spectra useful for reference purposes.
These nitrates are very strong Raman scatterers and can routinely be
detected and identified in particles of size down to 1 ym. Since some of these
salts are hygroscopic, measurement difficulties have been encountered when
ambient humidities are high and as a result the nitrate microparticles are
14
-------
modified due to adsorption of atmospheric moisture. Under these conditions,
spectral lines tend to broaden and Raman intensities fall off considerably.
This may be associated with changes in particle morphology or surface related
effects.
Oxides and Hydroxides
Particle Raman spectra have been obtained for several selected oxides and
hydroxides. Thorium oxide, Th02, has been of particular interest on account of
its refractory properties (m.p. 3050 °C) and the exceptional simplicity of its
vibrational Raman spectrum. The spectrum of a 0.8 ym particle of Th02 is shown
in Figure 3 of Reference 2.
Other oxides characterized include crystalline quartz (a-Si02) and alumina
(a-Al203) , i.e., sapphire, the latter because of its use as a substrate.
Micr opart icles prepared from crushed, single-crystal, natural quartz have been
measured to locate the expected Raman frequencies in the spectrum. Figure 6
of this report shows the spectrum of an "unknown" particle of respirable size
found in a bulk sample of urban air particulate dust. The particle is
identified as crystalline quartz.
URBAN
Sample DUST Size
in urban air particulate dust.
15
-------
Compared to quartz, sapphire or the mineral corundum (a-Al203) , is a
comparatively weak Raman scatterer, and for this reason serves as a low-
interference substrate material. The particle spectrum of Na2SOi| (see figure
2 of this report) includes the background spectrum of sapphire, excited with
a beam spot of ^2 ym.
The hydroxides of Ca2 and Mg2 were characterized in microparticulate
form. Ca(OH)2 was of interest in conjunction with our studies of CaCOs single
particles. The predicted Raman frequencies were observed in the particle
spectrum of each compound. Spectroscopic sensitivity for these solid
hydroxides extends to particles a few micrometers in size.
Glasses
The matrix of coal fly ash consists primarily of insoluble alumino-
silicate glasses, with the surface layers of ash particles generally showing a
predominance of certain trace elements.
We have characterized by Raman spectroscopy two types of synthetic glasses
which are representative of a series of NBS "standard glasses" of known elemen-
tal composition. These glasses have been developed at NBS to serve as elemental
standards for quantitative electron probe microanalysis and x-ray fluorescence
analysis. They consist in each case of a fused, solid mixture of several
oxides. Raman spectra have been obtained from bulk samples (small chips) of
two of these glasses. These are identified by their respective composition, in
weight percent:
- Glass K-309 (40%, Si02, 15% BaO, 15% A1203, 15% CaO,
15% Fe203). This material is, in the bulk, an opaque,
black glass. In microparticulate form, individual
particles are still nearly opaque to transmitted light.
- Glass K-240 (40%, Si02, 30% BaO, 5% ZnO, 5% MgO, 10%
Zr02). This is a clear, honey-colored glass in the
bulk. Small particles of it appear totally transparent
(colorless) in transmitted light.
In some respects, small particles of these glasses may model the composi-
tion and structure (or lack of it) of the solid phase characterizing the
"glassy", frequently hollow, spheres found in fly ashes from coal-fired power
plants.
The spectroscopic results obtained from measurements on these samples
verify the disordered "glassy" structure of these solids.
Our conclusions drawn from a study of these spectra are in agreement with
those drawn by others from spectroscopic studies of bulk glasses and various
mineral silicates of glass-like structure [4,5,6]. It is a general observa-
tion from these studies that the loss of long-range crystalline order in the
glass results in very extensive broadening of most spectral features. The
vibrational spectrum then is usually indicative of the glassy nature of the
sample and often indicates the family of glass (e.g., high or low silica content)
16
-------
but is much less definitive with respect to the precise composition of the
material. Thus, certain of the mineral glasses and glassy solids produced by
high temperature reactions may yield well-defined and reasonably intense Raman
spectra, whereas others give extremely poor Raman spectra because of the low
degree of orderliness in the glass network. It can be generally concluded that
the more complex the composition of a glass becomes, the weaker and the more
diffuse the Raman bands become. Spectra are then usually flat and featureless;
sometimes a sharp band may be identified with included bits of a crystalline
phase (e.g., silicate).
These observations have significant implications in regard to the Raman
characterization of particulates produced in high temperature zones, such as
emissions from coal-fired power plants.
ORGANIC COMPOUNDS AND POLYMERS
It is felt that organic surface layers may exist on power plant emissions.
Therefore, the development of techniques for organic micro-Raman analysis is of
importance. Spectroscopic measurements have been performed on a variety of
solid organic microparticles as well as on polymers, including the following
materials:
- Sodium Oxalate,
Obtained excellent Raman spectra for single particles down to 2
in size. An example is shown in Figure 7 of this report.
.JL
M_ r o 2.3x2.6
Somplel>la2Vj2v-'4Size urn
Substrate AI203
Laser X05l4.5nm
power 43 mw (at sample)
beam~2.0 urn diameter
Spectral slit width 3 cm'1
Time constant 0.2 s
Pinhole 140 \im .
Scan rate 3 33 cm '/s
Full scale
10°
counts
M
ll
2000
1000
cm"
0
Figure 7. Raman spectrum of a particle of sodium oxalate.
17
-------
Urea (carbamide), H2NCONH2
Spectra of excellent slgnal-to-noise and with good resolution of
multiple bands were obtained for particles of size 2-3 ym. These
are identical to the reported Raman spectrum of a macro-sample of
urea.
Benzoic Acid, CgHsCOOH
Spectra obtained for particles <10 urn in size, requiring low
irradiance levels. Typical of these measurements is the spectrum
shown in Figure 8, obtained with 10 mW of laser power concentrated
in a beam spot of ^14 ym diameter. Higher irradiance levels
(40 mW) have caused slow destruction of the sample as mounted on
the LiF substrate. Similar observations have been made for other
radiation-sensitive organic particles. Heating of the sample in
these cases may be due to the presence of absorbing impurities
present in or on the particle.
Substrot* LI F
Laser lu 514.5 nm
power |0 iW (ot sample)
beam* 14 um diamiler
Spectral si* width 3 cm'1
Time constant 0.5 >
Plnhoto 140 inn,
Scan rat* 083 cnr'A
Full scol* '3 counts
3000
L
2000
cm
'°°°
Figure 8. Raman spectrum of a particle of benzoic acid.
- Polyvinyl chloride (PVC)
Spectra have been obtained from single particles of size
2-7 ym. In these measurements up to 60 mW of laser power
(\ = 514.5 nm, Beck-focused beam spot of ^2 ym) have been
placed on the sample without any deleterious effects. The
demonstration spectrum of a small (^4 ym) particle of PVC
is shown in Figure 7 of Reference 2.
Several other polymeric materials have been studied in the Raman micro-
probe, primarily for the purpose of evaluating various polymeric membranes as
supporting films for particle collections. Several types of plastic films are
commonly used as filter media (e.g., Millipore, Nucleopore, Fluoropore, etc.)
for the collection of ambient aerosol. These thin, microporous membrane
filters (the nominal thickness of 0.8 ym pore size Nucleopore filter is
ym) are not suitable as supporting films for particles to be analyzed in
18
-------
the Raman microprobe. For example, membrane filters of the polycarbonate and
Teflon variety exhibit fairly complex Raman spectra with intense bands over the
entire spectral range of interest. These serious interferences preclude the
use of these membranes as substrates and require the removal of particle col-
lections from such filters.
The results obtained to date on organic compounds in microparticulate form
are very encouraging and indicate that many of the colorless organics are
identifiable in particles down to 1 pm. Limited measurement experience has
been gained in the detection and identification of contaminating organic layers
on inorganic- or mineral-core particulates. In some cases, organic or biologi-
cal contaminants present in or on particles have been decomposed at
apparently high induced temperatures to a carbonaceous residue, as evidenced
by the appearance of a pair of broad bands in the 1200-1700 cm"1 region of the
spectrum. These features, one mode centered near 1350 cm"1 and the other
around 1600 cm 1, are attributed to a form of activated carbon (or polycrystal-
line graphite) produced in the thermal decomposition of the organic material.
LIQUID SULFATE AEROSOL
A large number of studies have been conducted by various workers to
characterize the properties and identity of atmospheric acid aerosol. In this
area of environmental measurement, the emphasis has been on the detection,
monitoring and quantitation of several molecular forms of ambient acid aerosol,
principally I^SO^, NH^HSOtj and (NHi^SOif These species are thought to exist
as microdroplets in atmospheric aerosol. Major interest centers on their
mechanisms of formation, transformation, dispersal and removal in ambient air.
Free (i.e., molecular) E^SO^ and NH^HSOt* in aerosol form react in situ with
NHs to form (NHif)2SOtt. Both I^SO^ and NHi+HSOit are hygroscopic substances which
are aqueous solution droplets at all humidities from 30 to 100 percent. In
contrast, (NH^^SOt^ is a deliquescent salt which undergoes a transition from
the dry crystal to a solution droplet at the relative humidity corresponding
to that over the saturated solution of the salt (^80 percent). Thus under
normal atmospheric conditions, the latter two forms of acid sulfate may exist
in either the solid or the liquid phase.
Because of the significance of the sulfate aerosol system in power plant
emissions, we have in preliminary experiments applied the Raman microprobe
to the characterization of liquid sulfate aerosols. We have attempted to study
spectroscopically the transformation from liquid sulfuric acid aerosol to
solid, microcrystalline aerosol of ammonium sulfate by reaction with ammonia
vapor.
Our experiments have involved the generation of polydisperse sulfuric
acid aerosol by nebulization of concentrated sulfuric acid. For micro-Raman
investigation, the acid aerosol was collected on the surface of the standard
sapphire substrate which, for this application, had been coated with a thin
film of a Teflon-like polymer. Initial trials to maintain the spheroidal
shape of the aerosol droplets on the uncoated substrate surface failed due to
the unfavorable wetting properties of the sapphire. Application of the hydro-
phobic polymer fi?.m provided an impaction surface for the aerosol upon which
19
-------
droplet shape was maintained for extended periods of time. This also prevented
any extensive coalescence of microdroplets.
The transformation from the liquid to the solid aerosol was made by "gas
phase titration", exposing the sample to vapors of NH3 of known concentration
(below 100 ppm). Experimental conditions were adjusted to allow for the slow
growth of microcrystallites of (NHi+)2SOit from the liquid phase. Spectroscopic
measurements on single microdroplets were made with the new, expanded beam spot
of diameter approximately 7 um, at full laser power. The size range of the
sampled aerosol varied from <35 ym to 2-3 urn. These experiments allowed the
spectroscopic detection of undissociated sulfuric acid, E2SOi^, in microdroplets
(down to the size of 5 urn) of the concentrated_acid aerosol, the observation
of the characteristic Raman bands for the HSOit ion, and the monitoring of the
intense Raman line of the SOfp ion both in the liquid phase as well as in the
solid phase of
Representative results of these very preliminary experiments are shown in
Figures 9 and 10 of this report. Prior to reaction with NH3, the predominant
species are undissociated H2SOit and the bisulfate, HSO^ , ion. In the
spectrum (Fig. 9) of the microdroplet of cone. I^SOi^ several broad features of
low intensity are seen. The Raman shift around 903 cm 1 is assigned to the
symmetric vibration vj of the H2SO)4 molecule. In less concentrated solutions
(<80 wt%, as is the case here) there exist ionized species, so that the Raman
shift centered at 980 cm * can be assigned to the free SOfp ion in solution
(c.f^ Table 1). The remaining three frequency shifts are characteristic of the
HSO^ ion in solution, with the band around 1050_cm 1 being the strongest of
these. The two bands around 420 cm and 600 cm are primarily due to the
bending modes (at 424 and 592 cm"1) of the HSOii" ion but also contain the
contributions by the v2 and v^ modes of the free SO^ ion (c.f. Table 1).
The transformation from the 112801+ solution phase to the solid phase of
is demonstrated by the spectra shown in Figure 10. Measurement
parameters were the same as those used to record the spectrum of Figure 9,
except for laser power (150 mW) and time constant (0.8 sec). The spectrum
(top) of the non-reacted acid droplet displays the features seen in Figure 9.
Titration with NH3 furnishes the "neutralized" droplet, the spectrum (middle)
of which is dominated by the sharp, intense 980 cm"1 line of the so£~ ion in
solution. Because of the more complete dissociation of H2SO^, this species is
now present in much greater concentration, hence the large increase in the
strength of the 980 cm 1 signal. The broad feature centered around 1100 cm"1
arises from the v3 mode of the SOfp ion. As the transformation is completed
and the microparticle is formed, the lower spectrum results, characteristic of
crystalline (NHit)2SOtf. The sulfate symmetric stretch in the solid falls at
976 cm l and is increased in peak intensity by one order of magnitude. The
other bands appear at the expected frequencies (c.f. Table 1).
The significance of these results lies in the capability of the Raman
technique to extract species information from liquid microparticles. Consid-
erably more work will be required to assess the present limits of detection
(both in terms of droplet size and species concentration) for the various
20
-------
sulfate forms suspected to be present In atmospheric liquid aerosol. We
anticipate that this capability will also be useful to the characterization
of liquid aerosol emissions from stationary sources (e.g., oil-fired power
plants).
KSa-Droplet
Sample ^lze3Q ^m
SubstrateA|oO» NveDQT
*^*(B»^O * w»^%»"
Laser X0 514.5 nm
power 143 mw (Qt sample)
beam-x 7 Mm diameter
Spectral slit width 3 cm"'
Time constant 0.5 s
Pinhole 140 urn .
Scan rate | gy crrT'/s
Full scale " IA^ counts
1 - 1
1 - \ - 1 - 1 - 1 - 1 - 1 -
1000 cm-i 500
0
Figure 9. Raman spectrum of a microdroplet of cone.
Teflon-coated sapphire substrate.
supported by a
21
-------
DROPLET
D~8.5um
1.2 t
NHS-
"NEUTRALIZED"
DROPLET
CRYSTALLIZED
DROPLET
(NH«)eS04 11.4 t
I ' '
T
1500 . 1000
cm"1
Figure 10. Raman spectra showing the transformation (from top to bottom) from
a microdroplet of cone. I^SOit to a microparticle of (NHi^SOit on
reaction with ammonia.
22
-------
SECTION 6
CHARACTERIZATION OF UNKNOWN PARTICLES FROM POWER PLANT EMISSIONS
With the emphasis on the need to control emissions from stationary sources
there arises the need for more information on the composition of fly ash
particles emitted into the atmosphere. Coal- and oil-fired power plants are
among the largest anthropogenic point sources of particulate matter.
Limited knowledge is available from bulk samples on the relationship
between particle size distribution and composition in power plant emissions.
On a particle-to-particle basis there exists no information on the molecular
identity of these particulates.
The project objectives include feasibility studies on the micro-Raman
characterization of air pollution particles caused by the combustion of fossil
fuels, especially a study of the identity of combustion-produced (sulfate) acid
aerosol in emissions from oil-fired power plants. The primary interest centers
on the identity of the molecular states of oxygenated sulfur-containing species
that are believed to be closely associated with the particulate carbon (i.e.,
soot) in these samples. Preliminary studies have been undertaken in this area
of stationary source emissions characterization, and the results of these
measurements are summarized in the following.
FLY ASH FROM COAL-FIRED POWER PLANTS
Exploratory micro-Raman measurements were carried out on single fly-ash
particles contained in NBS Standard Reference Material, Coal Fly Ash, SRM 1633
("Trace Elements in Coal Fly Ash") . The bulk ash is a sieved (<88 um) and
blended material collected by electrostatic precipitators.
Fly ash particles generally in the size range 5-20 ym were examined in the
Raman microprobe, employing high irradiance levels placed on the sample with
the Beck-focused ^2 ym beam spot. The majority of the particles analyzed
exhibited weak and diffuse spectra, often with high background levels merging
into the broad wing of the Rayleigh line. In a small number of cases, that is
for particles so far unidentified, the spectrum shows a broad band at around
450 cm"1, accompanied by a second broad feature at around 950 cm *, indicative
of a glass-like structure and perhaps evidence of Si-0 stretching vibration.
For a number of large (10-20 ym) irregularly-shaped particles, a very strong,
sharp band has been observed at 470 cm -1, in addition to a broad band of
medium intensity at 205 cm"1 and a very sharp intense band at 128 cm l. Col-
lectively these features are indicative of crystalline quartz (a-Si02).
23
-------
The lack of strong Raman scattering observed for most of the particles
analyzed may also be explained by the fact that in these cases they may be
hollow spheres.
Coal fly ash sample #DPC ISA is a bulk sample of fly ash made available
for micro-Raman study by the EPA. The sample is an in-stack sample, collected
after precipitators; received in the form of a heavy, dense layer of particulate
deposit on a 80 mm diameter glass fiber filter. A micro-Raman sample (using a
LiF substrate) was prepared by transferring a light brushing of the ash deposit
onto the substrate. Examination of this preparaton in the polarizing light
microscope at magnifications from 50-625X showed that the ash consisted of a
distribution of very fine, spherical particles, all of which are <2-3 ym in
diameter. The majority of the particles in this ash proved to be around 1 urn
in size and smaller, appearing in the form of extensive clusters and chains in
the sample prepared for Raman analysis. Only a small fraction of the ash
particles existed in this preparation as discrete particles. Individual fly ash
spheres or clusters of particles showed little or no birefringence. They
appeared fairly transparent (although difficult to judge because of their small
size) in transmitted light and showed various hues of color (mainly shades of
brown and green). Spectroscopic measurements were made with 514.5 nm excita-
tion (i.e., green line of Ar /Kr laser) employing a range of irradiance levels.
The lowest practical irradiance levels (on the order of kilowatts/cm2) placed
on the sample produced heating in the particles probed by the beam, as was
evident from the spectral response. In these cases the spectra show high back-
ground signal levels (due to broad-band non-Raman emissions from the sample)
that usually completely swamp any Raman event. Frequently these high, inter-
fering light levels decay with time and continued irradiation of the particle.
Moderate to high irradiance levels placed on these ash particles have had the
effect of decomposing, melting, and vaporizing a fraction or all of the
irradiated sample. In cases where a larger beam spot (7-15 ym in diam.) was
placed on a cluster of particles, this interaction has brought about the
fusion of these particle aggregates into large spheres or globs of solid
material. Residue spheres of this type have been produced with diameters
20-30 ym. We attribute the excessive heating of these particles to the
presence of constituents effectively absorbing the energy at the excitation
frequency- These may be various colored metal oxides in the matrix or on the
surfaces of the particles or, in fact, films of combustion-produced carbon.
Further research is required in the study of such samples. Certain
advantages might be gained, for example, from the use of other excitation
frequencies. These have not been explored in our work. Other output frequen-
cies (e.g., 488.0 nm blue and 676.4 nm red) are available from the laser at
adequate power levels but have not been employed in the routine measurements
described here. As is typical of all Raman Spectroscopic analyses of colored
samples, some advantages can be gained from the use of other frequencies. The
choice of a "best" excitation frequency is complicated by such factors as
heating from absorption of the exciting beam, fluorescence from the sample and
possible resonance effects. For unknown samples it is difficult to a priori
select the optimum excitation frequency, and only after experience is gained
for each measurement problem can the "best" excitation be chosen.
24
-------
PARTICULATE EMISSIONS FROM OIL-FIRED POWER PLANTS
Various EPA particle samples of oil-fired power plant emissions have been
received for micro-Raman analysis.
In the time available for the application of the technique to the charac-
terization of these samples, we have focused our efforts on a preliminary study
of a set of samples from an oil-fired power plant designated as Site A by EPA.
These samples were sent to NBS in the early part of September 1976.
This set of five samples was collected by the sampling staff of the
Stationary Source Emissions Research Branch of the EPA/RTP, at the Site A
power plant on August 19, 1976. Sample collection involved the use of a five-
stage Battelle cascade impactor in the out-stack sampling mode. Sampling was
conducted at the fourth port from the bottom, stack temperature was 325 °F
(163 °C). The sampler was positioned at the end of a 8ft.-long sampling train
(i.e. probe of %" I.D.), with the probe and sampler heated to 200 °F (94 °C) .
On each of the five stages the impactor employed a micro-Raman substrate
to collect the size-fractionated aerosol. Only on the third stage was a
substrate used bearing an aluminum film particle finder grid.
The power plant at Site A is operated without emission controls. During
the test, the unit was operated at an excess boiler oxygen level of about 0.2
percent, using a fuel of sulfur content 2.5 percent with concentrations of
vanadium about 400 ppm.
The particulate emissions from the plant at Site A have recently been
characterized by EPA investigations [7,8]. These studies have included the
determination of particulate mass, particle size distribution and trace element
composition from bulk collections obtained by in-stack and out-stack sampling
methods. Elemental analysis by x-ray fluorescence shows compounds of sulfur,
vanadium and nickel to be major components in these particulate emissions
samples. The molecular form of these compounds (e.g., oxides, sulfates,
vanadates, etc.) cannot be inferred from this data on the trace element composi-
tion. The carbon content of these samples is found to be typically 60 wt.
percent. Thus, these samples as collected in the bulk (e.g., as deposits on
filters) have the appearance of finely-divided black powder. Exposed to the
ambient air, such particulate samples are believed to take up enough moisture
to form appreciable amounts of "acid smut" (i.e., a mixture of hygroscopic acid
aerosol and carbon).
Characterization by Light Microscopy
Microscopic investigation of the set of five Site A emissions samples
was made with a research microscope with photomicrographic camera. All observa-
tions were made in transmitted light, at various magnifications from 50X to
625X. The object was to take note of the optical properties and morphology of
the particles collected on each stage and to record their location on the
substrate for subsequent micro-Raman analysis.
25
-------
In Figures 11 through 14 are given selected optical micrographs showing
representative fields of view of the particle deposits on each of the five
sample substrates. The substrate is in all cases the usual optical-quality
sapphire. On stage 3 the sapphire had deposited on it a thin aluminum film in
the design of a particle finder grid. Figure 11 shows the particulate material
collected on stages 1 and 2 of the sampler. On these stages, as on the others,
the sampled particulate material ranges from colorless to pale yellow, orange,
light brown, greenish brown, brownish red and black. Figure 12 shows repre-
sentative areas of the stage 3 sapphire substrate where the impacted aerosol
(apparently solid and liquid) has reacted with the thin aluminum film. What is
seen, therefore, in the aluminized areas are predominantly the reaction
products. Vast areas of the grid are destroyed and have been etched away as a
result of the chemical interaction that has taken place. Careful microscopic
examination of this stage shows liquidus or deliquescent microparticulates and
a multitude of microcrystallites of various colors, reminiscent of dendritic
growth. In Figure 13 are shown selected photomicrographs of the emissions
collected on stage 4. The low-magnification (SOX) photo shows the central area
of the substrate, with several large (>100 ym) dendrite type crystals in a
field of black, spongy particles or aggregates. The predominant, four-leaf
crystal shown at 125X and 312X magnification is about 200 ym in size. Very
interesting are also the many groupings of micro-dendrites on this stage, of
which a typical one is shown (at 312X) in the fourth micrograph of Figure 13.
It appears that these microcrystals have grown on the substrate, presumably
from the reaction of various combustion products present in all phases of
matter. As of this time, micro-Raman measurements have only been made on
selected particles of stage 4. An overview of the sampled aerosol collected
on stage 5 is given in Figure 14. The SOX micrograph shows splays of fluid
collected at the periphery of the substrate, with ensembles of microcrystal-
lites entrapped in transparent, solid "skins". The center of the sample,
presumably the area of most direct, hard impaction, is shown at 125X featuring
carbonaceous, spongy material and films or sheets of solid, colored material.
The two higher magnification (312X) photos show extensive liquidus material, in
microdroplets of all sizes. Particularly interesting is the colony of micro-
crystals seen in the fourth micrograph. The origin and formation of these
features is unexplained.
Results of Micro-Raman Analysis
Exploratory spectroscopic measurements have been made on particles
collected on the Raman substrate of stage 4. In these studies spectra could be
obtained from particles that did not show any intense coloration and appeared
to be relatively free from fine particulate soot. Basically three representa-
tive types of particulates have been probed in this sample. Predominant on
this stage are several large and many smaller dendrite crystals grown on the
substrate (i.e., sapphire) surface. A number of these microcrystals (c.f.
Fig. 13) have been analyzed in the Raman microprobe. The second major type
consists of particles of varying size and shape close to the brown micro-
crystals described above. These particles are mostly globular, do not appear
to be porous, but seem to be a compact, fused mass of crystalline material,
with various degrees of transparency. Spectroscopic analysis indicates that
these particles are inhomogeneous in composition. They seem to be made up of
a colorless, transparent solid material interspersed with other solid material
26
-------
Vs**
5*. T"». i
125X
312X
125X
«,** d*"» *i ^FsspWi*'" J ,>«"
** V w iSwi^r'i j> ^v
jf^^Sw^i-i^
. * *».x <^,
.
*
312X
Figure 11.
Optical micrographs of Site A power plant particulate emissions
collected on stages 1 and 2 of Battelle impaction sampler.
' !
-------
SOX
125X
* ._4£7»*.~ .'.-XT
^ '^%V:
' :/f:
+
fyt ., *
Figure 12.
312X
Optical micrographs of Site A power plant particulate emissions
collected on stage 3 of Battelle impaction sampler.
28
-------
?«*.
SOX
125X
312X
312X
Figure 13- Optical micrographs of Site A power plant particulate emissions
collected on stage 4 of Battelle impaction sampler.
29
-------
SOX
125X
. 'fiS
' * », IS
-.
i
312X
312X
Figure 14.
Optical micrographs of Site A power plant particulate emissions
collected on stage 5 of Battelle impaction sampler.
-------
varying in color from shades of brown to green, with bits of black material
included. This second class of particles has also been analyzed spectroscopi-
cally, and Raman spectra have been obtained which are discussed below. A third
type appears as colorless, transparent strands, connecting the two types of
particulates described above. These thin strands, and frequently sheets, of
colorless, deposited material have also been analyzed.
Our measurements of several particulates on the fourth impactor stage have
furnished a number of essentially identical Raman spectra for the dendrite type
microcrystals. Representative of these spectra is that shown in Figure 15,
obtained from probing a small segment of the large, four-leaf crystal (c.f.
optical micrographs shown in Figure 13). The predominant features in these
spectra are bands with Raman shifts around 145, 280 and 1000 cm 1. We have
performed additional measurements on several other microcrystals of this type,
type, including small dendrites (c.f. 312X micrograph of Figure 13) of size
down to 5 urn. From these studies we have obtained spectra consistent with the
earlier ones. As a result of further characterization of known compounds, we
have since been successful in interpreting the observed Raman spectra. The
Sample
Sub
ANT
stage 4
Spe
i. 514 5 nm
wer 60 mw (at sample)
-7 u diameter
Time constant 0.8 s
Pinhole 140 (im ,
Scan rate Q83 cm /s
Full scale (Q counts
1 1 1 1 1 1 r
3000 2000
~\ r
-i r
cm
1000
0
Figure 15. Micro-Raman spectrum of a large microcrystal (shown at 312X in
Figure 13) found on stage 4 of the sampled power plant
emissions (see text).
major constituent present in these particles is crystalline vanadium pentoxide,
V205. This is indicated from a comparison of the reference spectrum of a small
(4x5 um) particle of V205, - shown in Figure 16 , with the spectrum of the
unknown microcrystal shown in Figure 15. The vibrational Raman spectrum of
crystalline V205 has been discussed in the literature [9]. The bands that are
nicely resolved in the spectrum of the microparticle have frequency shifts 104,
144, 285, 406, 701 and 995 cm l. These same bands are observed with equally
good resolution in the spectra of the unknown microcrystals. There exist,
however, additional bands of medium intensity, in the region 820 to 970 cm
- which are absent in the spectrum of pure V205. At this time we cannot
-1
31
-------
urn
Sample V205 Size4x5u
Substrate Li F
Laser \0 514.5 nm
power 3 mW (at sample)
beam~ 13 um_diameter
Spectral slit width
Time constant 1.6
Pinhole 140
Scan rate Q ^3
Full scale 3
cm
s
1000
cm
H
0
Figure 16. Raman spectrum of a particle of vanadium pentoxide.
account for the existence of these spectral features; these indicate the
presence of a second component which we have not detected in the spectra of the
other two types of particulates which we have analyzed. In parallel with our
study of V2C>5 we have also examined the spectra of microparticles of reagent-
grade, crystalline vanadyl sulfate, VOSOt^ZE^O, [10]. The spectrum of a small
particle of this salt just under 10 ym in size, is shown in Figure 17.
Infrared data for this salt have been reported [10], but Raman data do not
appear to exist. The most intense band in the spectrum is the vj symmetric
S-0 stretch of the sulfate ion falling at ^1010 cm 1. Two other sharp bands of
strong intensity appear at higher frequencies of the \)\ fundamental mode. In
addition^ several well-resolved, sharp bands are seen in the low frequency
(<400 cm *) region which may be attributed to various V-0 vibrations and
lattice (or external) modes of the crystal. None of the bands characteristic
of VOSOtj are present in the spectra of the unknown microcrystals (c.f. Figure
15). We therefore feel that the vanadium present in these particles does
not exist in the form of the vanadyl ion, VO2 , but appears to essentially
exist in the +5 oxidation state (as in ₯205). Other oxides of vanadium such
as V£03 and v^Ott are known to exist and are formed by oxidation of vanadium.
Since no spectral data exist for these species it is not clear whether the
32
-------
VOSCX-2H.O
Sample H * Size9x7um
Substrate L( p
Laser X05l4.5nm
power 30 mW (at sample)
bearrrv (3 pm_diameter
Spectral slit width 3
Time constant 3
Pinhole 140
Scan rate Q 33
Full scale " »
I03
cm"1
s
Mm
cm"1 /s
counts
1000
cm'1 0
Figure 17. Raman spectrum of a particle of vanadyl sulfate.
features in the 820 to 970 cm 1 region are due to these possible lower oxides.
We have also considered the vibrational spectroscopy of the (ortho-) vanadate
ion, V0§ , also containing pentavalent vanadium , and are inclined to
conclude that the existence of this species is not indicated in the spectra we
have recorded from these particles. The vanadate anion is tetrahedral as are
POtf and Soi| , therefore four fundamental modes are expected to be active in
the Raman. For the "free" vanadate ion, these internal vibrational modes are
observed [4] at Raman shifts 827 (vx), 341 (v2) , 780 (v3) and 341 cm'1 (v4).
Raman data on crystalline vanadates (including the meta-vanadate ion, VOj)
seem to be non-existent in the literature. For this reason, we have begun to
study the spectroscopy of some of these solids (e.g., KV03, NasVOi^) . Various
other microcrystals in the specimen have yielded Raman spectra in full agree-
ment with the spectrum obtained for ₯205 (c.f. Figure 16). In these cases,
the additional spectral features seen in the region 820 to 970 cm l are absent,
and it can be concluded that these crystals consist of the oxide essentially
free from any other second component.
In addition, we have obtained a second type of Raman spectrum from the
globular type of particulates described earlier which indicates the presence
of a crystalline sulfate. Representative of these results is the spectrum
33
-------
shown in Figure 18. It shows four distinct peaks, with the major band
exhibiting maximum intensity at 981 cm"1. This band appears split and has a
second, sharp component at 990 cm"1. Toward lower frequencies appear three
other bands, centered around 625, 460 and 150 cm"1. Taken together, these
spectral features are indicative of crystalline double sulfates, and in
particular hydrated metal ammonium sulfates , some examples of which have been
discussed in the literature [11,12]. We do not believe that ammonium, NH/+, is
present in crystalline particulates from oil-fired power plant emissions.
Rather, we suspect that particle transformations may have occurred as a result
of particle reactions with ambient (i.e., non-stack) concentration levels of
ammonia from a contaminating source. Spectra very similar to the one shown in
Fig. 18 have also been observed from portions of the colorless, transparent
strands. In some of our measurements of globules and strands, we were not
always successful in recording a spectrum that appeared to be of the type shown
in Figure 18 (from the existence of the strong ^980 cm"1 band). -The diffi-
culties encountered were due to heating of the sample on laser irradiation
which brought on its (slow) decomposition. In these cases we could attribute
the observed particle heating to carbonaceous material in the sample. A
number of these same spectra that could be obtained showed spectroscopic
POWER PLANT
EMISSIONS stage 4
Sample
Substrate AlgOj
Laser Ji« 514.5 nm
power 20 mW (at sample)
beam ~ 20 Mm diameter
Spectral slit width 3 cm"1
Time constant 1.0 s
Pinhole 140 \m
Scan rate 0.83 cm /s
Full scale 3 counts
2000
1000
cm
Figure 18. Raman spectrum of a globular particle found on stage 4 of the
sampled power plant emissions.
34
-------
evidence for the presence of a carbonaceous residue. We are familiar with the
appearance of "carbon bands" in the spectra of environmental particles from our
measurements on urban air particulate dusts. An example is shown in Figure 19.
The major constituent of this small particle is calcite, CaC03 [13]. Present
also is a small quantity of anhydrite, CaSOjj, evident from the medium
intensity band at VL020 cm'1 ; a pair of distinct, well-defined bands attrib-
utable to carbon is observed with the two maxima centered around 1350 cm"1 and
1580 cm . These same features have appeared superimposed upon the spectrum of
Figure 18 for several of the emissions particulates we have analyzed. Bands
very similar in shape and peak frequency have been observed from pure, poly-
crystalline graphite [14]. We can suggest two possible explanations for the
appearance of carbon bands in the spectra of urban dust particles and power
plant particulate emissions. One would be that they arise from combustion-
produced carbon (e.g., soot) associated with such samples. On the other hand,
we cannot exclude the possibility that these bands may derive from laser-induced
thermal decomposition of organic components (e.g., oil films). The first
rationale is in agreement with the observations made by other workers in the
study (by ESCA methods and laser Raman spectroscopy) of carbon associated with
bulk samples of atmospheric particulates [15]. We are presently conducting
experiments with the intent to study spectroscopically the creation or evolution
of these carbon bands in the analysis of microparticles intentionally contami-
nated with polycrystalline carbon films or with surface layers of polycyclic
organic compounds of the type suspected [8] to exist with oil-fired power plant
emissions.
We intend to continue our measurements of the Site A particle samples and
plan to also perform electron-probe analyses on particles measured in the micro-
Raman spectrometer. The characterization of known materials will continue as an
important activity so as to increase the reference data base for molecular
identification of unknowns. Based on our preliminary results obtained in the
study of liquid aerosol microdroplets, we hope to be able to extract useful
spectroscopic information from the liquid material collected on stage 5 of the
sampler (c.f. Figure 14).
URBAN 5x5
Sample DUST ,s^° "m
Substrate Li r
Laser V 514.5 nm
power 48 my/(at san-p!e)
beam* 12 um^jiiometer
Spectral slrt width C- cm-!
Time constant 0.5 s
Pinhole I'1-O urn
Scan rote 1.67 cnT'/s
Full scale 3 counts
10
3000
i r
2000
cm"
1000
0
Figure 19. Raman spectrum of a particle of calcite (CaC03) in urban air
particulate dust.
35
-------
REFERENCES
1. Rosasco, G. J., E. S. Etz, and W. A. Cassatt. The Analysis of Discrete
Fine Particles by Raman Spectroscopy. Appl. Spectrosc., 29:396-404,
1975.
2. Rosasco, G. J., and E. S. Etz. A New Microanalytical Tool: The Raman
Microprobe. Res. & Devel., 28:20-35, June 1977.
3. Ross, S. D. Inorganic Infrared and Raman Spectra. McGraw-Hill Book
Company (UK) Limited, Maidenhead, Berkshire, England, 1972, 414 pp.
4. Griffith, W. P. Raman Spectroscopy of Terrestrial Minerals. In: Infrared
and Raman Spectroscopy of Lunar and Terrestrial Minerals. C. Karr, Jr.,
ed., Academic Press, New York, 1975, pp. 299-323.
5. White, W. B. Structural Interpretation of Lunar and Terrestrial Minerals
by Raman Spectroscopy. In: Infrared and Raman Spectroscopy of Lunar and
Terrestrial Minerals, C. Karr, Jr., ed. , Academic Press, New York, 1975,
pp. 325-358.
6. Brawer, S. A., and W. B. White. Raman Spectroscopic Investigation of the
Structure of Silicate Glasses. I. The Binary Alkali Silicates. J. Chem.
Phys., 63:2421-2432, 1975.
7. Knapp, K. T. , W. D. Conner, and R. L. Bennett. Physical Characterization
of Particulate Emissions from Oil-Fired Power Plants. Paper presented at
the 4th National Conference on Energy and the Environment, Cincinnati, OH,
October 4-7, 1976.
8. Bennett, R. L. , and K. T. Knapp. Chemical Characterization of Particulate
Emissions from Oil-Fired Power Plants. Paper presented at the 4th
National Conference on Energy and the Environment. Cincinnati, OH,
October 4-7, 1976.
9. Gil son, T. R. , 0. F. Bizri, and N. Cheetham. Single-Crystal Raman and
Infrared Spectra of Vanadium (V) Oxide. J. Chem. Soc. (Dalton) , 291-
294, 1973.
10. Ladwig, G. Zur Bildung und Natur des a-VOSO^ und seines 1-Hydrates,
Z. anorg. allg. Chemie, 364 (No. 4/5) :225^-240, 1969.
11. Ananthanarayanan, V. Raman Spectra of Crystalline Double Sulfates.
Part II. Ammonium Double Sulfates. Z. Phys. 166:318-327, 1962.
36
-------
12. Fawcett, V., D. A. Long, and V. N. Sankaranarayanan. A Study of the
Internal Frequency Region of the Raman Spectrum of a Single Crystal of
Sodium Ammonium Sulfate Dihydrate. J. Raman Spectrosc. 3:217-228, 1975.
13. Porto, S. P. S., J. A. Giordmaine, and T. C. Damen. Depolarization of
Raman Scattering in Calcite. Phys. Rev. 147:608-611, 1966.
14. Tuinstra, F. and J. L. Koenig. Raman Spectrum of Graphite. J. Chem.
Phys. 53:1126-1130, 1970.
15. Rosen, H. and T. Novakov. Raman Scattering and the Characterization of
Atmospheric Aerosol Particles. Nature 266:708-710, 1977.
37
-------
TECHNICAL REPORT DATA
(Please read Instructions en the. reverse before completing)
1. Flj
4. TITLE AND SUBTITLE
CHEMICAL ANALYSIS OF STATIONARY SOURCE PARTICULATE
POLLUTANTS BY MICRO-RAMAN SPECTROSCOPY
Interim Report _ __
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
August 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
E. S. Etz, G. J. Rosasco, and K. F. J. Heinrich
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Gas and Particle Science Division
National Bureau of Standards
Washington, D. C. 20234
10. PROGRAM ELEMENT NO.
1AD712 BD-Q7 (FY-771
11. CONTRACT/GRANT NO.
EPA-IAG-D6-F012
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. N. C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Interim 4/76-3/77
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Analytical capability to identify the principal molecular species present in microparti-
cles is demonstrated on the basis of Raman spectra of selected compounds and materials
Among the inorganic species studied are sulfates, nitrates, carbonates and oxides, for
which Raman spectra are discussed for single, solid particles of size down to 1 micro-
meter. The method of micro-Raman analysis is applied to the molecular characterization
of individual microparticles from power plant emissions. Raman spectra have been ob-
tained from microparticles of oil-fired power plant emissions collected by the EPA with
cascade impaction samplers.
Vanadium pentoxide, V205, has been identified as a major component of microparticles
present in such samples. The presence of certain other vanadium containing species such
as vanadyl, V0^"% and ortho-vanadate, V043', is not indicated from the results of these
neasurements. Other Raman spectra show evidence of crystalline sulfate, S042", as a
species present in major proportions. However, the exact nature of the associated cat-
on specie(s) has not been determined. Many of the spectra obtained from fly ash par-
ticles show Raman bands characteristic of polycrystalline graphite apparently due to
the presence of carbonaceous material associated with the particles.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. cos AT I Field/Group
* Air pollution
* Particles
* Chemical analysis
* Raman spectroscopy
* Vanadium oxides
* Sulfates
13B
07D
14B
07B
8. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
46
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLET
E38
------- |