&EPA
               United States
               Environmental Protection
               Agency
             Environmental Sciences
             Research Laboratory
             Research Triangle Park
             NC 27711
EPA-600 9-78-020a
August 1978
               Research and Development
Workshop Proceedings on
Primary Sulfate Emissions
from Combustion Sources
               Volume 1
               Measurement Technology
  
  ,  -.-:
.
   
       m
<
                     ...(,>.

                                           n

-------
                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 arid Development
     8.  "Special" Reports
     9.  Miscellaneous Reports
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia  22161.

-------
                            EPA-600/9-78-020a
                            August 1978
Workshop Proceedings on
Primary Sulfate Emissions
from Combustion Sources

Volume 1
Measurement Technology
Sponsorship by
U.S. pnvirpnrrrrtal Protection Agency
April 24-p6,1i7&
Southern Pines, North Gmrolirm 28287

Coordination and Editing by
Kappa Systems, Inc.
         inia
Workshop Chairman
John S. Nader
Emission Measurements and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 2771 1
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
publication.  Approval does not signify that the contents necessari-
ly reflect the views and policies of the U.S. Environmental Protec-
tion Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.

In general, the technical content of the papers included in this
report have been reproduced in the form submitted by the authors.

Any papers included in the Program and not included herein were not
submitted for publication.

-------
Preface
     This volume contains the technical papers presented at the
Workshop on Measurement Technology and Characterization of Primary
Sulfur Oxides Emission from Combustion Sources held in Southern
Pines, North Carolina, April 24-26, 1978.  In addition, reports on
deliberations and recommendations of four Working Groups (corres-
ponding to the four sessions of technical presentations) are in-
cluded.

     A Working Group was formed for each of the four sessions of
technical presentations and consisted of all the speakers of that
session.  Each Working Group met in a Working Group Session, re-
viewed and critiqued the session presentations, and made its
initial report to all attendees in a summary session.  The report
summarized what  is known and deemed acceptable and what further
research activity needs to be pursued to provide desirable data and
information.  At the summary session, the initial report of each
Working Group was discussed and further modified to reflect the
comments and interaction between the Working Groups.  The report of
each Working Group resulting from this summary session is presented
in  this volume and follows the set of papers presented at the ses-
sion it addresses.

     The focus of sulfur pollutants impacting on ambient air quality
has been the criteria pollutant, sulfur dioxide and its oxidation
products, sulfuric acid and sulfate salts.  Considerable attention
has been directed to the sulfuric acid and sulfate salts resulting
from the chemical transformation of S02 both temporally and spatial-
ly  in  the atmosphere.  These are referred to as secondary sulfates.
Sulfuric acid and sulfate  salts emitted directly as emissions from
combustion sources also impact on  the ambient levels of sulfate.

-------
These direct emissions of sulfuric acid and sulfate salts are
referred to as primary sulfates.

     Since sulfates at this time are not criteria pollutants and
emission standards are not prescribed, no reference method is
established for their measurement in combustion source emissions.
With the current and ongoing concern about sulfur in fuels, there
is increasing effort in measuring and characterizing sulfur-contain-
ing emissions from combustion sources.  There is a need to identify
valid measurement techniques for primary sulfates, specifically
sulfuric acid, and to provide an accurate and consistent base of
characterization data on primary sulfate emissions from the
various combustion processes.  There is also the need to determine
what emission data on primary sulfates are available, their ac-
ceptability as valid measurements, and what further research effort
needs to be conducted to provide a good data base for a good under-
standing of the contribution of primary sulfate emissions to ambient
sulfate levels.

     The purpose of this Workshop was to help meet these needs.

     I am grateful lor the active participation of the Workshop
attendees who were invited to present and discuss their activities
and studies in the area of primary sulfate emissions and for their
contributions which made the Workshop an interesting and signifi-
cant accomplishment.  In particular, I want to thank the Session
Chairmen (James Dorsey, Kenneth Knapp, James Homolya, and John
Bachmann) and the Working Group Chairmen (Paul Urone, Dale Lundgren,
James Howes, and David Natusch) for their assistance in implement-
ing the Workshop agenda so effectively.  I also want to include
my appreciation for the efforts and cooperation of Ann Mitchell
and Wendy Martin of Kappa Systems in coordinating the Workshop and
in editing the Proceedings.
                                   John S. Nader
                                   Workshop Chairman
                                IV

-------
Contents
                                VOLUME 1


SECTION 1  Gas Sampling and Analysis

An Evaluation of a Modified Method 6 Flue Gas Sampling Procedure           3

      Russell N. Dietz
      Robert F. Wieser
      Leonard Newman

Measurements of Sulfur Trioxlde at Tennessee Valley Authority             27
Coal-Fired Power Plants Using the Condenser Method

      Elizabeth M. Bailey
      H. A.  Ruddock

Measurement of SO3/H2SO4  Concentration in Kraft Recovery Furnace        41
Stack Gas Using Controlled Condensation

      Ashok K. Jain
      R. O.  Blosser
      Howard S. Oglesby

Characterization of Combustion Source Sulfate Emissions with a             53
Selective Condensation Sampling System

      James L. Cheney
      James B. Homolya

A Specific Method for the Determination of Sulfuric Acid Emissions          63
from Combustion Sources

      Paul t/rone
      Robert A. Lucas

Measurements of Sulfuric Acid Vapor by Infrared Spectroscopy             79

      Roosevelt Rollins

Chemical Speciation and Concentration Monitoring of Sulfur Oxides         97
by Laser-Raman Scattering

      Richard K. Chang
      Robert E. Benner

-------
Report of the Working Croup on Measurement of Gaseous Sulfur            137
Oxides Emissions

      Russell N. Dietz, Reporter


SECTION 2  Particulate Sampling and Analysis

Collection Methods for the Determination of Stationary Source              145
Particulate Sulfur and Other Elements

      Kenneth T. Knapp
      Roy L. Bennett
      Robert J. Griffin
      Raymond C. Steward

A Stack Gas  Sulfate Aerosol Measurement Problem                        161

      Dale A. Lundgren
      Paul Urone
      Thomas Gunderson

Sulfur Oxide Interaction with Filters Used for Method 5 Stack Sampling     179

      Edward T.  Peters
      Jeffrey W. Adams

Particulate Sampling in Process Streams in the Presence of Sulfur          203
Oxides
      Kenneth M. Gushing

Primary Aerosol Sulfur Size Distribution Measurements Using a Low        227
Pressure Impactor

      Richard C. Flagran

Use of a High-Flow Stack Sampler for Determination of Particulate          241
Sulfate Emissions
      A. Jack O'Neal, Jr.
      Harold Cowherd

Inorganic Compound Identification by Fourier Transform Infrared          253
Spectroscopy
      Robert J. Jakob sen
      R. M. Gendreau
       William M.  Henry
       Kenneth  T. Knapp

Report of the Working Group on Measurement of Particulate Sulfur         275
Oxides Emissions

       Richard C. Flagon, Reporter

Appendix -  Participants and Observers                                  277
                                     vi

-------
                                 VOLUME 2
SECTION 1  Gas Emissions
An Assessment of Sulfuric Acid and Sulfate Emissions from the               3
Combustion of Fossil Fuels
      James B.  Homolya
      James L.  Cheney
Sulfur Oxides Emissions from Boilers, Turbines, and Industrial             13
Combustion Equipment
      Skillman C. Hunter
      Paul K. Engel
Some Recent Data on SO3 and SO4 Levels in Utility Boilers                  53
      Brian W.  Doyle
      Richard C. Booth
Measurement of Sulfur Oxides from Coal-Fired Utility and                  67
Industrial Boilers
      William R. McCurley
      Daryl G.  DeAngelis
Sulfur Oxide Measurements of Utility Power Plant Emissions                87
      James E.  Howes, Jr.
Effects of Combustion Modification on S03 Formation in Combustion          99
      Arthur Levy
      John F. Kircher
      Earl L. Merryman
Impact of Sulfuric Acid Emissions on Plume Opacity                        121
      John S. Nader
      William D. Conner
Query:  Is There a Connection between the Expansion of Areas of           137
Acid Rain and a Shift from Coal to Oil for Small-Scale Heat Needs?
      Arthur M. Squires
Report of the Working Croup on Characterization of Gaseous                143
Sulfur Oxides Emissions
      Arthur M. Squires, Reporter
                                     VII

-------
SECTION 2  Particulate Emissions

Characterization of Fly Ash from Coal Combustion                         149

      David F.  S. Natusch

Sulfur and Trace Metal Particulate Emissions from Combustion              165
Sources
      Roy L. Bennett
      Kenneth T. Knapp

Inorganic Compounds Present in Fossil Fuel Fly Ash Emissions             185
      William M. Henry
      Ralph I. Mitchell
      Kenneth T. Knapp

Investigation of Particulate Sulfur by ESC A                               209

      Arthurs. Werner

Sulfur Emissions Sampling and Analysis                                  219

      Ray F. Maddalone

Operating Parameters Affecting Sutfate Emissions from an Oil-Fired         239
Power Unit
      Russell N. Dietz
      Robert F.  Wieser
      Leonard Newman

Report of the Working  Croup on Characterization of Particulate Sulfur       271
Oxides Emissions
      Roy F. Maddalone, Reporter

Appendix - Participants and Observers                                   275
                                     VIII

-------
Section 1
Gas Sampling and Analysis

-------
An Evaluation of a Modified Method 6 Flue Gas
Sampling Procedure
Russell N. Dietz
Robert F. Wieser
Leonard Newman
Brookhaven National Laboratory
     ABSTRACT

     The character of primary sulfate emissions  (i.e., H2S04
     and water-soluble sulfate salts) must  be  determined using
     flue gas sampling methods that differentiate between the
     acid form and the less nocuous sulfates.  These methods
     must be used in considering potential  health effects and
     in determining the mechanisms and parameters that affect
     the magnitude and distribution of such emissions.

     As recommended, the Modified EPA Method 6 does  not pro-
     vide for specific determination of sulfate  particulates
     and H2S04 .  A Brookhaven modification, based on an iso-
     kinetically-sized nozzle and quartz fiber filter assembly,
     has been shown to collect quantitatively, in situ, all
     the particulate sulfate.  Sulfuric acid was passed through
     the filter for subsequent quantitative collection, and
     no spurious formation of acid by oxidation  of  862 was
     found on the filter.  Problems associated with  sub-iso-
     kinetic particulate sampling, leading  to significant
     positive sulfate measurement errors,  will be discussed.

     In many cases, the isopropyl alcohol  (IPA)  used to col-
     lect the H2S04 vapor contained sufficient oxidant to pro-
     duce positive errors 10-fold or more  when actual flue  gas
     acid levels were less than 1 ppm.  In  addition, stripping
     experiments demonstrated the presence  of dissolved S02
     equivalent to 1 ppm to 3 ppm of flue  gas acid  even after
     16- to 32-minute purging periods.  The significance  of
     these errors and simple corrective procedures  will be
     presented.

-------
INTRODUCTION

     The character of primary sulfate emissions  (i.e., H2S04 and
water-soluble metal sulfates) must be determined with  flue  gas sam-
pling methods that differentiate between  the acid  form and  the less
nocuous metal sulfates, both for consideration of  potential health
effects and for determination of the mechanisms  and parameters
which affect the magnitude and distribution of those emissions.

     As presently recommended, the modified version (1) of  the EPA
Method 6 (2) does not provide for specific determination of par-
ticulate metal sulfates and H2S04.  A Brookhaven modification,
based on an isokinetically-sized nozzle and quartz fiber filter
assembly (3), has been shown to quantitatively collect, in  situ,
all the particulate metal sulfates.  Sulfuric acid was quantita-
tively passed through the filter for subsequent  collection  in the
probe and the final collection stageseither an isopropyl  alcohol
(IPA) midget bubbler and filter (EPA Method 6) or  a multi-turn con-
densation coil and filter (controlled condensation system).  The
latter is described in detail elsewhere (3) and, for the purposes
of this paper, will be considered to be an absolute method  to which
other techniques will be compared.

     Two problems have been encountered with the utilization of
the Method 6 IPA midget bubbler for collection of  H2S04.  In some
preliminary field experiments (3) employing the  IPA bubblers, we
noted both the presence of an oxidant in  the IPA solution which
converted dissolved S02 to HaSO4 and the  tendency  of the solutions
to retain dissolved S02 even after the recommended purging  period
with ambient air.  The presence of oxidant in the  IPA  solutions
has been noted by others (4)(5).

     This paper describes an improved version of the IPA bubbler
technique for collection of sulfuric acida version which  can re-
duce the errors associated with neglecting the correction for the
presence of an oxidant in the IPA and the inability to entirely
strip the dissolved S02 prior to determination of  H2S04.  Labora-
tory and field experimental results confirmed the  existence of
both phenomena as well as the apparent existence of a  volatile
oxidant in the flue gas which converted S02 to H2S04 in the IPA.
Because of the significant errors associated with  even this improved
version, especially at fossil-fueled combustion  sources utilizing
low sulfur and low vanadium fuels, it is  recommended that the Modi-
fied EPA Method 6 using IPA bubblers be replaced by a  miniaturized
version of a controlled condensation system.  Errors associated
with non-isokinetic sampling of the particulate  metal  sulfates
will be presented.

-------
THE SAMPLING APPARATUS

     Modifications to the EPA Method 6 were made  in order  to quan-
titatively separate the particulate metal sulfates from  the sulfuric
acid in the flue gas.  This section describes  the sampling apparatus
employed to implement the separation in the Brookhaven Method 6
(BM6).


Nozzle and Filter Assembly (Metal Sulfates Collection)

     The first significant modification, a miniaturized  version  of
the Brookhaven controlled condensation system  (CCS) in situ flue
gas nozzle and filter assembly  (NFA), was designed and fabricated
as shown in Figure 1.  The nozzle inside diameter was 1  mm for
isokinetic sampling (100 ft/sec velocity) at 1  f/min and 2 mm for
25 ft/sec.  In order to compare isokinetic sampling for  particulate
metal sulfates with the non-isokinetic approach sampling normal  to
the direction of flue gas flow, a tube and filter assembly was
also devised to approximate the dimension in the  EPA version  (1).
The efficiency of the acid-treated quartz fiber filter used in both
assemblies has been well documented  for nearly 100% collection of
the particulate metal sulfates  while passing generally 95% or more
of the flue gas H2S04 (3).  The trace amount of H2S04 retained by
the filter was recoverable by washing first with  100% IPA  before
washing with water for recovery of the water-soluble metal sulfates.


Probe and Bubbler Assembly  (H2S04 Collection)

     After passing through  the  NFA and heated  probe  (c~350 to
450F) , the particulate-free  flue gas then passed through  the
second major Brookhaven modificationtwo midget  bubblers  separated
by a quartz fiber filtershown schematically  in  Figure  2. The
first of the two midget bubblers, each containing 15 mf  of 80%  IPA
for collection of H2S04 aerosol, was followed  by  a heat  and acid-
treated quartz fiber  filter mounted  in an assembly shown in detail
in Figure 3.  As demonstrated  in laboratory studies,  the filter
would not allow any acid aerosol to  pass  into  the second bubbler.

     Sulfur dioxide in the  flue gas  was collected by  the two  midget
impingers each containing 15  mP of 3% H202 witi, a third  impinger
remaining empty to collect  any  entrained  solution.   The  dried  gas
was then pumped through the dry test meter  for an accurate measure
of the sampled volume with  a  small portion  (~0.4  H)  collected  in
the pre-evacuated gas bottle  sampler at a constant  rate  for  about
20 minutes  (controlled by the critical orifice size).   A photograph
of the ice bath vessel  (190 mm  diameter by  100 mm high)  containing

-------

Figure 1.  Flue gas isokinetic nozzle and filter assembly.

-------
                          FLUE DUCT WALL
NOZZLE AND FILTER
ASSEMBLY (NFA)
   FLUE GAS
    FLOW
SAMPLING
  PORT     HEATED
        PROBE ASSEMBLY
                                  TEFLON FITTINGS

                                 /           /FILTER
                             HEATED
                             TEFLON
                              LINE
                             iAA
                            ICE
                           BATH
                    THERMOMETER
               DRY
           TEST METER
            BUBBLERS

        EVACUATED
         .BOTTLE
                                                         VMIDGET
                                                         IMPINGERS
                                                   JEWELED
                                                   ORIFICE
     TYGON
       LINE

VACUUM GAGE
             QUICK
           CONNECTOR
                                                        PUMP
                                       to^
           Figure 2.  Schematic diagram of Brookhaven Method 6 apparatus.

-------
Figure 3.  Midget bubbler filter assembly.

-------
the five midget contactors is shown in Figure 4.  The inlet to the
first bubbler (on the left) is connected to the probe via a heated
Teflon line and a Teflon fitting (Beckman union shown).  Oxygen
and carbon monoxide content of the flue gas was determined chroma-
tographically from the contents of the gas bottle sampler (6).
LABORATORY EXPERIMENTS
Determination and Treatment of IPA Oxidant

     Earlier tests at Brookhaven (3) and elsewhere (4) indicated
that an oxidant, in varying amounts depending on the batch, was
present in the IPA, as received.  Spot tests with 2% KI solution,
which released the yellow color of I2 in the presence of a strong
oxidant, confirmed these observations.  A treatment procedure in-
volving copper powder or wire was found to completely reduce the
oxidant.  By acidifying a batch of IPA with HC1 to a pH of 4.0 to
4.5 (the treatment time at a pH of 5.8 was more than 20 hours) and
exposing the solution to pre-cleaned copper powder or copper wire
(cleaned in concentrated HC1 and rinsed in IPA) with gentle stirring
for about 16 hours, no residual-oxidant could be detected as shown
by the tests in Table 1.
                    Table 1.  IPA Oxidant Tests
                 Flush
Flush
Total Dissolved S02, meq,
Test
1
2
3
4
5
6

IPA
Solution
Ul
Cu PI
Cu Wl
U2
U3
Cu Wl

Time,
Min.
16
16
16
16
16
8
(24
Vol
/
8.7
9.2
11.9
11.9
12.1
6.3
days
Bubbler
Direct
0
0
0
0
0

later)
.0124
.0064
.0072
.0072
.0048


0
0
0
0
0
0
<0
1
Evap.
.0080
.0000
.0000
.0022
.0000
.0000
.0002
Bubbler
Direct
0
0
0
0
0


.0204
.0116
.0100
.0076
.0032


0
0
0
0
0
0
<0
2
Evap.
.0116
.0000
.0000
.0028
.0000
.0000
.0002
     That series of tests was performed with 15 mf of 80% IPA sol
ution in each bubbler sparged with 1400 ppm of S02 in simulated

-------






Figure 4.  Midget contactors i                bath.

-------
flue gas for 20 to 30 minutes using the BM6 apparatus followed by
a flushing period for the times and volumes indicated using S02-
free air.  Half the diluted solutions from each bubbler were
titrated directly with 0.02 N NaOH and the other half after evap-
oration to "dryness" to expel the last traces of dissolved 862
("dryness" in the presence of saturated CaCl2 solution in a vacuum
oven).   It was apparent in every test that evaporation markedly
reduced the level of detected 862, indicating that 9 f to 12 f of
flushing air was not sufficient to expel all dissolved SOa-  In
Test 1, the S02 remaining after evaporation can be attributed to
that which was oxidized by the oxidant to H2S04.  S02 was always
higher  in the second bubbler compared to the first in the direct   k
titration because the second bubbler was flushed with air containing
some S02 that was stripped from the first.  However, if a fixed
amount  of oxidant (e.g., H2O2) were present in the IPA, the expected
residual 862 (i.e., H2S04) in each bubbler after evaporation should
have been identical.  That it was not in Test 1 indicated that
the oxidation must have continued to some degree during the evap-
oration when more S02 was present in the second bubbler.  Thus the
oxidant was at least not entirely attributable to H202.

     Tests 2 and 3 showed that the copper treatment was successful
in removing the last traces of oxidant, as evidenced by the lack of
any acid in the evaporated samples.  Tests 4 and 5 with two different
batches of untreated IPA showed the variability of the oxidant
level;  for U3 the oxidant was apparently not present.  The final
test with copper wire treated IPA (preferred over the copper powder)
included an evaluation of 24 days exposure time of the S02-con-
taining solution prior to evaporation.  There was essentially no
H2S04 found and hence no arti/act formation from possible trace
amounts of Cu ions.  Several of the evaporated solutions which
showed no trace of acid were re-analyzed by ion chromatography
yielding negligible (<0.0006 meq) sulfate.

     To determine if H2S04 could be recovered equally well from
treated as from untreated IPA solution, several tests were per-
formed  in which 0.0394 meq of H2S04 were added to both versions of
solution.  The recovery of H2S04 as acid was identical in both
treated and untreated IPA, averaging 97 _+ 1% from evaporated
solutions.  For several 0.01 meq H2S04 aliquots, recovery from
evaporated solutions ranged typically from 80% to 90%.

     Although we have demonstrated a successful method for treating
IPA to  remove unwanted oxidants, it is preferable to use IPA that
is initially free of oxidant.  A procedure of testing for oxidant
by using the 1400 ppm S02 simulated flue gas standard and checking
for residual acid after evaporation was adopted as standard prac-
tice prior to and after field utilization of the BM6.
                                11

-------
SO2 Stripping Tests of IPA

     The field procedure for sampling  the  flue gas consisted of
collecting a sample for ~40 minutes at about 0.75 //min, followed
by an ambient air purge for about 15 to 20 minutes in order to
strip away any dissolved SO2 prior to  acid determination in the
midget bubblers.  If any dissolved SO2 remained in the solution,
a positive error in the acid determination could result.

     Utilizing the BM6 apparatus, copper wire treated IPA (15 m/
of 80% solution) was used in two bubblers  which were sparged with
15 /  of 1400 ppm SO2 (equivalent to 1.71 meq of S02) to fully
saturate the solutions.  In one series of  experiments, the solu-
tions in the bubbler were flushed with air at 0.78 //min for periods
of from 2 to 32 minutes (1.5 to 25 / ).  Half of each solution was
immediately analyzed for residual SO2.  The balance, when brought
to "dryness," showed no presence of acid,  indicating that only
dissolved SO2 had been measured in the non-evaporated solutions.
Assuming that 30 liters of flue gas would  normally be sampled
during field utilization of the method, the residual dissolved
S02 was converted to an apparent H2SO4 concentration in the flue
gas as shown in Figure 5.   With a midget bubbler, even after strip-
ping with 12 t  to 25 t of  air,  residual dissolved S02 could cause
errors from 1 ppm to 3 ppm equivalent sulfuric acid.

     Saturated solutions of 80% IPA at 0C in contact with 140G
ppm SO2 flue gas were found to contain 0.103 meq of SO2 per 15 mf
of solution (equivalent to 40 ppm H2S04 in 30./  of flue gas);  That
was approximately two-thirds the reported solubility in water (Chem-
istry and Physics Handbook).  Considering the bubbler to be a well-
stirred vessel  with the purge air containing SO2  in equilibrium
with the remaining dissolved SO2 and that Henry's Law prevailed,
the apparent H2SO4  concentration in the flue gas as a function of
time was derived to be

                               FH
                                    t
              IH2S04] = 40 e  24'5u                            [1]

where         [H2SO4] = flue gas H2SO4 concentration, vol. ppm

                 F    = flow of purge air, //min

                 H    = Henry's Law constant, 0.41 a tin //mole at 0C

                 t    = purging time, min

                 v    = volume of IPA solution,  liters
                               12

-------
   100

   70



   40
   20 J,
E
O.
O.
O
>
O
CO
 OJ
X

CO
<
O

LU
 10

  7
LU
Q:
2
Q.
 1.0

0.7



0.4




0.2
    O.I
                             MIDGET  BUBBLER
                                  =0.37
\ THEORETICAL

 \    =100

 \

 I

  \

.  \     I
FRITTED
   BUBBLER

 =0.57
                                I
      0
       8      12      16

         FLUSH VOLUME, LITERS
                                              24
             28
 Figure 5.  Residual dissolved S02 as apparent  flue gas

          H2S04 versus  the volume of flushing air used.
                            13

-------
For the conditions used in these tests, the theoretical residual S02
as H2SO4 was given by

               [H2S04] = 40 e ~  1'12 V                           [2]

where               V = volume  of purge air, liters

which is plotted in Figure 5 as the dashed line.

     Since the measured values  for the midget bubbler curve deviated
immediately and quite significantly from the theoretical dilution
line, it was concluded that the midget bubbler was not an ideal
well-stirred vessel.  Inspection of the bubbler in action supported
that conclusion.  From a series of stripping tests using a midget
impinger and then a flat-fritted bubbler, a different curve was gen-
erated for each device as shown in the plot.  Using mathematically
derived slopes from least mean  square empirical correlations of
the data, it was determined that the midget bubbler gave the poorest
approach (e - 0.37) to a well-stirred vessel (e = 1.00) with the
flat-fritted bubbler performing the best of the three (e = 0.57).
Statistically there was no difference in the residual dissolved SO2
in the presence of added acid (0.01 meq)i.e., equivalent to 4 ppm
H2SO4 in 30 t of flue gas.


FLUE GAS SAMPLING RESULTS

     From July 1977 to March 1978, a number of flue gas sampling
measurements using the Brookhaven Method 6 and the Brookhaven con-
trolled condensation system (CCS) were performed at the Long Island
Lighting Company (LILCO) power  stations at Northport (Units 2 and
3) and Island Park, New York (Barrett Unit 1).  The Northport units
generally used No. 6 oil containing 2.4% S with significant amounts
of vanadium (300 ppm to 400 ppm V).  Barrett Unit 1 was a low
sulfur oil-fired boiler (0.3% S) having much less vanadium (5 ppm
to 30 ppm V).

     Details of the results of  sampling at these units have been
reported elsewhere (7)(9).  In  this section, the measurement of
particulate metal sulfates at isokinetic and non-isokinetic sam-
pling velocities, as well as the determination of H2S04 at high
(>10 ppm) and  low (<1 ppm) levels with the IPA approach compared
with the more  reliable CCS, will be presented.
                                14

-------
BM6 Field UtilizationParticulate Metal Sulfates

     When sampling into the direction of flue gas flow, sub-iso-
kinetic runs at Barrett Unit 1 gave particulate metal sulfates
1.32 +_ 0.08 times the isokinetic value when sampling at about
10% of isokinetic velocity.  At Northport Unit 3 at the electro-
static precipitator (ESP) outlet, for the same approximate 10%
of isokinetic velocity, the particulate metal sulfates were 4.4
+1.1 times the isokinetic value.  The ratio of measured to iso-
kinetic metal sulfates was plotted versus the ratio of nozzle
velocity to flue gas velocity in Figure 6.  Since sub-isokinetic
sampling tends to bias the particulate collected with an excess
of large particulate, it is apparent that most of the sulfate-
bearing particulates at Barrett were quite small even though the
unit had no particulate controls.  Surprisingly, the substantial
increase in sulfate with sub-isokinetic sampling at the ESP out-
let of Northport Unit 3 indicated that a major fraction of the
sulfate particles were larger in size (probably greater than 1
jitm to 3 /urn} , even though at the ESP outlet.

     Northport Unit 2, essentially the same as Unit 3 except
that Unit 2 had no ESP, gave results not significantly different
from those  obtained at Barrett.  The presence of a significant
portion of  larger particulates at Northport Unit 3 may have
been due to the  low furnace oxygen levels  (0.0% to 0.2%) nor-
mally used  at  that unit compared to  the others  (typically  1.0%
to 1.5% 02)

     The recommended  Modified EPA Method 6 probe tip consisted  of
an 11 mm inside  diameter  glass probe end,  with  sampling occurring
normal  to  the  direction of  flue  gas  flow  (1).   The straight nozzle
and  filter  assembly devised at Brookhaven  to  duplicate the
extremely  low  sub-isokinetic conditions gave  the results shown  at
the  left in Figure 6.  Two  of the runs gave particulate metal sul-
fate measurements approximately  50%  greater than the isokinetic
value wnen  utilized at Northport Unit 2.

     When  sampling nearly  isokinetically  (within + 20%) by
either  BM6  or  CCS, into  the direction of  flue gas  flow, the mea-
sured particulate metal  sulfates were within  about + 20% of the
isokinetic  value.  The EPA  straight  probe  method on  the average
gave higher (about 30%)  results  than the  isokinetic  value, but
a statistically  significant number of runs was  not made.   Com-
parative measurements between the straight nozzle  and  the  iso-
kinetic  approach  at Northport Unit 3, because of the signifi-
cant velocity  dependence  at that unit, would  be of  interest.
                                15

-------
UJ

<
U_
_l
=>
CD

o
h-
LU
   3.6



   3.2



   2.8
  2 4
0)  ^'^
LJ

<
=>
CO

Q
UJ
o:
=>
en
<
UJ
   2.0



   1.6



   1.2



   0.8
NORTHPORT
UNIT 2
-(STRAIGHT)
                                    NORTHPORT
                                    UNIT 3
NORTHPORT
     UNIT 2
                     BARRETT
      0.01               O.I                 1.0

              NOZZLE VELOCITY/FLUE GAS  VELOCITY
                                                             10.0
      Figure 6. Effect of sub-isokinetlc flue gas sampling on
              measured particulate metal sulfates.

-------
BM6 Field UtilizationSulfuric Acid (High Emissions)

     As discussed earlier, the sulfuric acid from the sampled
flue gas was expected to be recovered from the various sections
of the Brookhaven Method 6 apparatus including the nozzle filter,
the probe, the first midget bubbler, and the bubbler filter;
the second midget bubbler was not expected to have measurable
amounts of H2S04  because of the absolute filtering capabilities
of the bubbler filter.

     The results of seven sampling runs at Northport Unit 2,
at high H2S04  emissions conditions because of high furnace
oxygen levels (1.2% to 2.7% O2), are shown in Table 2.  The
H2S04 collected in the different portions of the sampling ap-
paratus was determined primarily by titration with 0.02 N
NaOH with a few comparisons by ion chromatography sulfate de-
termination (the values in parentheses).

     Neglecting the previously established problems of residual
dissolved SO2  and oxidant in the IPA (the IPA used in these
measurements was shown to be oxidant-free both before and after
the field experiments), the flue gas H2SO4 was determined from the
sum of the milliequivalents found in the nozzle filter, the
probe, the direct (nonevaporated) solution of midget bubbler 1,
and the bubbler filter.   For example, for run 76, a flue gas
H2S04 direct value of  31.1 ppm  (26.5 ppm by  ion chromatography)
was determined.  However, it was apparent that the significant
difference between the direct determination  of midget bubbler 1
by titration versus  that  by ion chromatography was indicative
of a  significant amount of a volatile acid in addition to H2S04.
Indeed,  after evaporation of midget bubbler  1, the H2S04 de-
termined  by both methods  was nearly identical (the samples  of
each  bubbler were always  split equally  so that half could be
measured  directly and  half after evaporation)--indicating that
the dissolved S02 and  any other volatile acid were removed
by the evaporation.

      That there were  substantial amounts of  residual dissolved
S02 ,  even after the  recommended stripping volume of 15 $ of  air,
was evident from the  significant amounts of  S02 milliequivalents
found  in  the directly measured midget bubbler 2.  Even after
evaporating midget bubbler 2, a not-insignificant amount of
acid  (sulfate) still  remained.  Since it was s.own conclusively
that  the  IPA prior to  use contained no  oxidant, that the bub-
bler  filter would allow no H2S04 aerosol to  penetrate to the
second bubbler, and  that  evaporation was a quantitative method
for removing all dissolved S02, it  was  also  concluded that  the
                                17

-------
oo
                                              Table  2.   Brookhaven  Method 6 Field Results
                                                     (Northport  Unit 2 - High H0SO )
Acid Concentration,
Run
No.
PK-75
76
77
78
79
80
81
Noz/.le
Kilter
0.0032
0.0008
0.001(5
0.0032
0.0036
0.0028
0.0048
Probe
0
0
0
0
u
0
0
.0186
.0404
.0588
.0526
.0714
.0702
.0768
mi Hi equivalent sa
Midget Bubbler 1
Direct Evap.
0.0344
0.0364
(0.0224)
0.0400
(0.0283)
0.0584
0.0328
0.0372
0.0300
(0.0169)
O.0132
0.0192
(0.0205)
0.0200
(0.0235)
0.0184
o.oieo
0.0060
0.0092
(0.0146)
Bubbler
Filter
0.0124
0.0088
(0.0099)
0.0074
(0.0080)
0.0098
0.0090
0.0036
0.0044
(0.0064)
Midget Bubbler 2
Direct
0.0228
0.0220
(0.0211)
0.0240
(0.0163)
0.0268
0.020.4
0.0288
0.0240
(0.0157)
Evap.
0.0008
0.0028
(0.0055)
0.0040
(0.0069)
0.0036
0.0040
0.0012
0.0024
(0.0061)
Flue Gas H2S04 ' Vol. ppm
BM6
Direct13
29.7
31 .1
(26.5)
40.3
(36.2)
53.3
48 .0
52.8
49.2
(44.4)
Corr.c CCSd
20
23
(23
31
(31
34
39
37
39
(40
.2
.9 26.6
.8)
^ ^ 	
'.8)
.5
.5 
.8
.3
.9)
            Mi 11 ifxjui valents determined by  titration  with 0.02  N  NaOH  or  by  ion chromatography (in parentheses).
            Direct BM6 results are based on  the  nozzle,  probe,  direct  midget  bubbler 1,  and bubbler filter acid (sulfate).
           "Corrected BM6 results are based  on the  nozzle,  probe,  difference  of Evap.  1  minus Evap. 2, and the bubbler filter
            ae Ld (sulfate) .
           'CCS ari? the controlled condensation  system  results.

-------
flue gas contained a volatile constituent which dissolved in the
IPA solution and which caused the oxidation of dissolved S02
either during the sampling or, more likely, during the evapora-
tion step.   Thus, these flue gas sampling results using the
Brookhaven  procedure and apparatus have uncovered another problem
associated  with the utilization of IPA solution in the presence
of flue gas.

     Assuming that the oxidized S02 after evaporation of the
residual dissolved S02 was the same in both bubblers, the amount
of H2S04 that truly originated in the flue gas could be estimated
by subtracting the contents of evaporated bubbler 2 from evaporated
bubbler 1.   The results, when added to the acid from the nozzle
filter, probe, and bubbler filter, were reported as corrected
flue gas H2S04 .  For run 76, the corrected H2SO4 determination
was 23.9 ppm, in close agreement with the CCS value sampled about
two hours earlier.  (Unfortunately, only one port was available
for sampling at this temporary ducting location.)

     On the average, the directly determined H2S04 flue gas con-
centrations were either about 35% or 11% higher than the cor-
rected levels depending on the method of determination.  One
could conclude that either with or without the correction, this
version of the IPA procedure gave results comparable with the CCS
at these high acid levels.


BM6 Field UtilizationSulfuric Acid (Low Emissions)

     When the BM6 approach was used at two low acid emission
sourcesNorthport Unit 3 and Barrett Unit 1the latter of which
had H2SO4 levels less than 0.2 ppm, the direct determination of
H2S04 was obviously erroneous as shown in Table 3.

     In these runs, multiple ports were available enabling the
controlled condensation system to be used side by side with the
BM6.  The IPA used in all these runs with the exception of run 71
was shown to contain oxidant comparable to the amounts found in
Ul and U2 of Table 1; that for run 71 was oxidant free.

     Before  considering the results obtained from the bubblers,
a case can be made for the efficacy of the controlled condensa-
tion approach for reliably sampling H2SO4 at very low levels
(less than 0.2 ppm).  There was no trace of acid on the bubbler
filters, indicating that the H2S04 was in all cases removed prior
to reaching  the  filter.  Since the average on the bubbler filters
in Table 2 was 0.0080 meq, it would intuitively be expected that
the H2S04 in the flue gas was on  the order of about 50-fold
lessi.e.,  less than 1 ppm.  That, of course, was shown to be

                                  19

-------
                                 Table 3.  Brookhaven Method 6 Field Results
                                 (Northport Unit 3 and Barrett Unit 1 - Low
Runb
No.
PR-44-B
46-B
48-B
51-B
55-0
57-0
59-0
63-1
71-1
Acid Concentration, milliequivalents*
Nozzle
Filter
0.0002
0.0004
0 . 0004
0.0002
0.0010
0.0006
0.0006
0.0006
0 . 0003
Probe0
0.0000
(390"F)
0.0000
(475F)
0.0004
(365F)
0.0000
(350"F)
0.0060
(480F)
0.0000
(430F)
0.0000
(405F)
0.0000
(370F)
0.0022
(300F)
Midget Bubbler 1
Direct Gvap.
0.0150
0.0160
0.0170
0.0190
(0.0175)
0.0364
0.0744
0.0320
0.0340
0.0156
0.0050
0.0080
(0.0124)
0.0080
(0.0121)
0.0076
(0.0132)
0.0156
0.0620
0.0172
0.0200
0.0064
Bubbler
Filter
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
Flue Gas HgSO. , Vol. ppm
Midget Bubbler ?, BM6
Direct
0.0240
(0.0373)
0.0280
0.0310
0.0260
(0.0259)
0.0172
0.0784
0.0208
0.0240
0.0124
Evap. Direct
0.0133 2.41
0.0180 2.79
(0.0267)
0.0200 3.19
(0.0279)
0.0166 3.79
(0.0243)
0.0156 10.0
0.0780 12.5
0.0188 6.9
0.0180 5.8
0.0060 6.0
Corr.e CCST
0.03 0.03
0.07 0.15
0.15 0.12
0.04 0.05
1.6 4.8
0.10 l.'JO
0.13 1.41
0.10 0.08
0.82 0.29
 Milliequivalents determined by titration with 0.02 N NaOH or by ion chromatography (in parentheses).
 B after the run number represents sampling performed at Barrett; 0, the electrostatic precipitator  (ESP)
 outlet at Northport; I, the ESP inlet at Northport.
cProbe temperature in parentheses.
^Direct BM6 results are based on the nozzle, probe, direct midget bubbler 1, and bubbler filter acid (sulfate).
Corrected BM6 results are based on the nozzle and probe acid content only; midget bubbler contents were not used.
 CCS are the controlled condensation system results.

-------
the case as confirmed by the CCS values, which, at Barrett (runs
with B) , varied from 0.03 ppm to 0.15 ppm H2S04.  Taking just the
acid recovered in the nozzle filter and the probe, the corrected
H2S04 levels were computed as shown in Table 2 under the BM6
heading labeled correctedi.e., the bubbler contents were dis-
regarded.  The agreement with the CCS values was remarkably good,
indicating that at very low acid levels, the controlled condensa-
tion approach was quantitatively reliable.

     Run 44 had just a slight amount of acid on the nozzle fil-
ter (0.0002 meq) and none in the probe (probe temperature was
390F); yet that amount of acid was precisely the total amount in
the sampled flue gas, as indicated by the comparison between the
corrected BM6 level and that from the CCS.  For run 46, the probe
temperature was much hotter (475F), and no H2S04 was retained in
the probe.  The corrected BM6 value of 0.07 ppm was less than
that by the CCS (0.15 ppm), indicating that some H2S04 did pass
through the probe into the bubbler system.  That this was the
case was substantiated by run 48 in which the probe tempera-
ture was lower  (365F) .  In that case an equivalent amount of
acid was found  in the probe and the nozzle filter, giving a
corrected BM6 H2S04 value of 0.15 ppm in good agreement with the
CCS value  (0.12 ppm).

     At the Northport Unit 3 ESP inlet  (runs 63-1 and 71-1),
the H2S04 was also quite low.   For run 63, just the acid found
on the  nozzle filter  (none in the probe) accounted for the total
flue gas  H2S04.   Significant acid was found in the probe of run
71 which, when  combined with the nozzle  filter, gave a corrected
BM6 H2SO4 of 0.82 ppmsignificantly higher than  the CCS H2S04
of 0.29 ppm.  The exact reason  for the discrepancy was not
evident;

     The  remaining runs performed at the ESP outlet of Unit 3
(Northport) occurred during somewhat higher acid  emissions (1
ppm to  5  ppm).  Because of the  high probe  temperature for those
three  runs  (55, 57, and 59), a  major portion of the flue gas
acid passed through the probe into the bubbler system, as evi-
denced  by  the low values for the corrected BM6 H2SO4 compared
to the  CCS acid.  Unfortunately, the problems previously de-
scribed concomitant with the utilization of IPA bubblers were
significant enough to preclude  the corrective procedure of the
subtractive technique  that was  used in  the high acid cases.
Thus,  there was no way to adequately determine that portion of
the acid  in those three runs that entered  the midget bubblers.
                                21

-------
     Run 71 represents  the best attempt to utilize correctly the
subtractive procedure.  First, it should be indicated that the
CCS gave an H2SO4 concentration of 0.34 ppm at noon and then 0.29
ppm at  1430the same time the BM6 run 71 was performed.  Since
all of  the acid in  the  flue gas was removed by the nozzle filter
and probe, the difference between evaporated bubbler 1 and eva-
porated bubbler 2 should have been near zero.  That this was the
case confirms the philosophy of the approach.  The amount of acid
found in each evaporated solution of run 71 could only have been
formed  from the oxidation of S02 by an unknown volatile flue gas
oxidant simultaneously  collected by the IPA.  Thus, even if the
subtractive procedure did work, the error associated with the
small difference between two large numbers could be appreciable.
Using the total amount  of acid found in midget bubbler 1, which
for run 71 could only be comprised of dissolved S02 and flue gas
oxidized S02, an acid level of 6.0 ppm H2SO4 would have been
recordeda substantial error.

     For almost all of  the runs in Table 3, the H2SO4 values,
determined by including the directly measured acid in midget
bubbler 1 with the  probe and nozzle filter washings, were from
one to  two orders of magnitude higher than the correct values.
Thus, the utilization of an IPA midget bubbler approach without
any regard for residual dissolved SO2, oxidant in the IPA, or
oxidant in the flue gas, can lead to gross positive errors in
H2SO4 determination.


CONCLUSIONS

     Laboratory and field experiments conclusively showed that
several problems involving the utilization of the IPA methodology
for the collection  of flue gas sulfuric acid could be minimized
when proper precautions were taken during field sampling at high
acid emission sources (>10 ppm H2SO4).  Under those conditions,
the effects of residual dissolved S02, oxidant in the IPA reagent
(a successful method for neutralizing that oxidant was devised),
and a volatile oxidant  in the flue gas could be reduced to prob-
ably less than 10%  positive errors by using the Brookhaven Method
6 (BM6) with the two midget bubbler evaporative subtractive
correction.  Using  the  presently recommended Modified EPA Method
6 (1),  with a single midget bubbler, positive errors at high acid
sources of as much  as 10% to 30% were demonstrated.

     At low acid sources (H2SO4 <1 ppm), the direct Modified
EPA Method 6 could  result in demonstrated 10- to 100-fold posi-
tive biased errors  for  sulfuric acid.  Even the subtractive
approach was not reliable because of the need to determine a
                                22

-------
small difference between two significantly larger numbers and
because the exact amount of residual dissolved S02 was variable
and generally larger in the second midget bubbler.

     If only limited qualitative measurements of the H2S04
emissions at fossil-fueled combustion sources were desired, the
IPA methodology might be adequate.  For example, if the actual
H2S04 level was 20 ppm, measured values of 22 ppm to 24 ppm might
be expected by the IPA methodology.  At 1 ppm H2S04> actual measure-
ments might give values of 2 ppm to 4 ppm H2S04.  And at 0.1 ppm
H2SO4 levels, measurements with the IPA methodology would still
give H2S04 values of 2 ppm to 4 ppm.

     Normally, that performance might be adequate from an emis-
sions inventory standpoint, since only the largest sources would
be the most significant.  However, from a controls standpoint,
the actual behavior of sources even at low H2S04 concentration
might be very important.  If the absolute emission of sulfuric
acid were directly dependent on:   (1) the sulfur content of the
fuel, (2) the vanadium content of  the fuel,  and  (3) the furnace
oxygen level, significant reductions in each of  those three
variables could result in a magnitude of H2SO4 reduction not
measurable by the IPA methodology.  For example  (9), at North-
port Unit 2 during one set of measurements,  the  fuel S content
was 2.1%, the vanadium was 390 ppm, and the  average furnace 02
was 2.0%; the corresponding average H2S04 was 32 ppm.  At Bar-
rett Unit 1,  the  fuel  S content was 0.3%  (7-fold reduction),  the
vanadium  averaged about 15 ppm  (26-fold reduction), and the fur-
nace oxygen  about 1%  (2-fold reduction).  The total reduction  in
H2S04 might  be  expected to be 7 x  26 x 2 or  about 360-fold,
corresponding to  an expected H2S04  level of  32/360 or 0.09 ppm.
The actual  average acid emissions  at Barrett were 0.10 +  0.07
ppm, in excellent agreement  (9).   These low  levels could  only
have been measured by  the controlled condensation system.  And
yet, from the viewpoint of understanding  the factors affecting
those emissions,  the  results were  very meaningful.

     Since  there  is a  better alternative  to  the  IPA methodology,
the recommendation is  made here that the  reference method  for
H2S04 determination be  replaced by a miniaturized version  of  the
Brookhaven  CCS  (3).   Conceptually,  the same  miniature  nozzle  and
filter assembly  designed  for the  BM6 would be used.  But  the  flue
gas, on leaving  the probe  (maintained electrically  at  about
350F), and  heated Teflou  line  (maintained electrically  at  140F),
would pass  directly  into  a version of  the midget bubbler  filter
(also maintained  electrically  at  140F)  for  removal  of  the  last
traces of H2S04  aerosol.   A  research study aimed at  designing
                                23

-------
and validating the midget controlled condensation system should
be implemented.

     Errors associated with the sub-isokinetic sampling of par-
ticulate metal sulfates while sampling normal to the direc-
tion of flue gas flow should be quantified at several other com-
bustion sources.  Tentatively, the right angle approach appears
to produce positive biased errors of about 30%.
ACKNOWLEDGMENTS

     We would like to thank Bob Gergley and Bob Wilson for their
help in performing the field experiments, as well as Lance
Warren and Fred Glaser at Northport and Ted Kempf and Ken
Abrams at Barrett for helping with arrangements and supplying
field data, Harold Cowherd and Fred Lipfert of LILCO for special
arrangements and B. T. Hagewood of LILCO for fuel analyses.
A special appreciation goes to the BNL Analytical Group for the
sulfate, carbon, and elemental analyses and to Irv Meyer of
the BNL glass shop for fabrication of components.  Several dis-
cussions with Jim Homolya and John Nader of EPA were very help-
ful.
                                24

-------
REFERENCES


1.   Cheney,  J. L., W. T. Winberry,  and J. B. Homolya.  Evaluation
     of a Method for Primary Sulfate Emissions from Combustion
     Sources.  J. Environ. Sci. Health, A12 (10): 549-566, 1977.

2.   U.S. Environmental Protection Agency.  Standards of Perfor-
     mance for Stationary Sources.  Federal Register 41(111),
     June 1976.  23083-5.

3.   Dietz, R. N., and R. F. Wieser.  Sulfate Emissions from Fossil
     Fueled Combustion Sources.  Brookhaven National Laboratory,
     Progress Report No. 5, September 1977.

4.   Maddalone, R. F., S. F. Newton, R. G. Rhudy, and R. M. Statnick.
     Laboratory and Field Evaluation of the Controlled Conden-
     sation System (Goksoyr/Ross) for S03 Measurements in Flue
     Gas Streams.  70th  Annual Meeting of the Air Pollution Con-
     trol Assoc., Toronto, June 1977.

5.   Homolya,  J.  U.S. Environmental Protection  Agency.  Personnel
     Communication, August 1977.

6.   Dietz, R. N.  Gas Chromatographic Determination  of Nitric
     Oxide on  Treated Molecular Sieve.  Anal. Chem.,  40:1576-8,
     1968.

7.   Dietz,  R. N., and R.  W.  Garber.   Power Plant Flue Gas and  Plume
     Sampling  Studies.   Brookhaven  National Laboratory, Progress
     Report  No.  1, November  1977.

8.   Dietz,  R. N., and R.  F.  Wieser.   Sulfate Emissions from Fossil
     Fueled  Combustion Sources.   Brookhaven National  Laboratory,
     Progress  Report  No.  6,  March 1978.

9.   Dietz,  R. N., R. F.  Wieser,  and L. Newman.  Operating Para-
     meters  Affecting Sulfate  Emissions from an  Oil-Fired Power
     Unit.   Proceedings  of Workshop on Measurement  Technology and
     Characterization of Primary  Sulfur Oxides Emission from Com-
     bustion  Sources, Southern Pines,  North Carolina, April  1978.
                                25

-------
Measurements of Sulfur Trioxide at Tennessee
Valley Authority Coal-Fired Power Plants Using
the Condenser Method
Elizabeth M. Bailey
H. A. Ruddock
Tennessee Valley Authority
     ABSTRACT

     Previous  and  recent measurements of sulfur  trioxide (S03)
     in the flue gas at several coal-fired generating plants
     in the TVA system will be presented.   Collection of sul-
     fur trioxide  was conducted using the condenser method
     of Lisle  and  Sensenbaugh.  Investigations were conducted
     on the effects of various plant operating parameters on
     the SOa content of the flue gas.  The SO3 content measur-
     ed varied from approximately 1 ppm to over  30 ppm de-
     pending on plant operating parameters.  Some of the sul-
     fur dioxide measurements were determined at one plant
     using impingers containing dilute solutions of H202.

     Tests were conducted to determine the efficiency of S03
     collection using the condenser method.  The optimum oper-
     ating parameters that should be used in sample collec-
     tion were also investigated, and study results will be
     presented.


INTRODUCTION

     The sulfur trioxide (SO3) content of power  plant flue gas is
of interest for both operational and environmental reasons.  Sulfur
trioxide combines  with water to form sulfuric acid (H2S04), which
causes problems in coal-fired power plant operation due to its
corrosive nature.  Sulfuric acid is believed to  affect human health
adversely (1)(2) and to damage both forest and lake ecosystems (3)
(4).  Most sulfuric acid aerosol in ambient air  is formed by oxi-
dation of sulfur dioxide (SO2) emitted from stationary sources.
However, flue  gas  from coal-fired power plants contains some pri-

                               27

-------
mary sulfuric acid aerosol, which is emitted directly from the
power plants into the atmosphere.

     The Tennessee Valley Authority (TVA) conducted three studies
to characterize the S03/H2S04 in flue gases.  The first study was
designed to examine the effects of furnace type and two plant op-
eration conditions, excess oxygen content and temperature of the
flue gas, on the SO3/H2S04 content of flue gas.  Measurements
were taken at four power plant units of different furnace types
over ranges of excess oxygen content and at one plant over a range
of flue gas temperature during this study.  In the second study,
the emission levels of S03/H2S04 and S02 were characterized at one
power plant by sampling flue gas at the base of the stack, and in
the third study, the efficiency of the S03/H2S04 collection system
was evaluated.


EFFECT OF FURNACE TYPE AND PLANT OPERATING CONDITIONS ON THE SULFUR
TRIOXIDE CONTENT OF FLUE GAS

     The TVA Division of Power Production conducted measurements at
four coal-fired power plant units to determine the effect of excess
oxygen content, furnace type, and temperature on the S03/H2S04 con-
tent of flue gas.


Methodology

     The condenser method, in which the gas is cooled to a tempera-
ture between the acid dewpoint and the water dewpoint to allow
collection of the S03/H2S04 in a condenser, was used to obtain samples
The sampling apparatus, identical to that of Lisle and Sensenbaugh
(5), is shown in Figure 1.  A glass-lined probe, containing a glass
wool plug for removing particulate matter, was used to draw flue
gas from the ducts.  The temperature of the probe was held at
290C, and the temperature of the condenser was held between 60C
and 90C.  Sulfur dioxide was collected in two gas-washing bottles,
each containing a solution of 1.5% hydrogen peroxide (H202).  The
gas sample was then passed through a drying tower containing silica
gel, and volume was measured with a wet test meter.

     Samples for studying the effects of excess oxygen were collec-
ted at the air preheater inlet, whereas those for studying the
effects of temperature were collected at the air preheater outlet.
Temperature of flue gas at the air preheater inlets was typically
315C to 370C, except at Allen 1 where it was 400C.  Because
of  this high temperature, slow flow rates of 2 f /min were used to
maximize the residence time of the flue gas in the condenser.
Isokinetic sampling was not possible at these flow rates.

                                28

-------
ro
co
      HEATED  PROBE
         290C
                                                                SULFUR
                                                                DIOXIDE
                                                                   LECTOR
SULFUR TRIOXIDE  COLLECTOR
          HEAT TAPE AND INSULATION
                                                                                       SILICA GEL  COLUMN
                                                      HYDROGEN PEROXIDE IMPINGERS'
                  O   Q
/



O
DRY
/-* A r
\


                                              METER
                                                                    PUMP
                  Figure  1.   Apparatus used to investigate effects of  plant
                              operating parameters on the 803 content of  flue  gases.

-------
     The condenser was rinsed with a solution of 5% isopropanol in
water after each sample was collected.  The rinse solution was
titrated with 0.02 N sodium hydroxide (NaOH), and bromphenol blue
was used as the indicator.  To analyze for SO2, the H202 solutions
were boiled to destroy excess H202,  and then titrated with 0.1 N
NaOH.  Bromphenol blue was used as the indicator.
Results and Discussion

     Samples were collected at four TVA power plant units to deter-
mine the effect of excess oxygen,  varying from 2.7% to 5.9% on
SO3/H2S04 content of flue gas.  The units investigated and their
furnace configurations were Shawnee Unit 3a 175-MW front-wall
unit located in Paducah, Kentucky;  Widows Creek Units 7 and 8
tangential units of 575 and 550 MW, respectively, located in
Stevenson, Alabama; and Allen Unit 1a 330-MW cyclone unit located
in Memphis, Tennessee.  Results of these tests, shown in Table 1
and Figure 2, demonstrated the dependence of the [S03/(SO2 + S03)]
fraction on the excess oxygen content of the flue gas.  A simple
linear regression analysis was applied to the data from each unit
to obtain the fitted equations:


     Shawnee Unit 3 [S03/(S02 + SO3)] (%) =
          0.12(% excess oxygen) -  0.09                (R  = 0.76)

     Widows Creek Unit 8 [SO3/(SO2 + S03)] (%) =
          0.35(% excess oxygen) -  0.50                (R  = 0.98)

     Widows Creek Unit 7 [S03/(SO2 + S03)] (%) =
          0.263(% excess oxygen)                      (R  = 0.84)

     Allen Unit 1 [S03/(S02 + S03)]  (%) =
          0.27(% excess oxygen) -  1.44                (R  = 0.83)


     Results of these tests also indicated that cyclone furnaces
may emit greater percentages of S03/H2SO4 than do front-wall and
tangential furnaces.  For all levels of excess oxygen tested, the
values of the [S03/(SC>3 + S02)] ratio, expressed as percent,
clustered near or below 1% for all units except Allen Unit 1 (a
cyclone furnace), which showed ratios greater than 2%.

     The effect of temperature on  the S03/H2S04 content of flue
gas was determined in a series of  tests conducted at TVA's Cumber-
land Unit 2, a horizontally opposed, 1275-MW unit located in
Cumberland City, Tennessee.  At least five tests were conducted at
                               30

-------
   3.0-
 O

'Z 2.0-

 o

Qi
o
if)
+*
o
oo
o
CO
   i.o-
                                                     Allen
      Widows Creek  7
      Widows Creek 8
             -

       Shawnee  3
     2.0
                       3 0
                                         I

                                        4.0
 I

5.0
                                   Excess  O 2 in flue  gas
      Figure 2.   Variation in  conversion of  S02  to S03 with

                  excess oxygen  content of flue  gas.
                                                                            6.0
                                        31

-------
each temperature range.  Results of these series (Table 2) indicated
that the SO3/H2SO4 content depends on the flue gas temperature and
decreases sharply when  the temperature is lowered.  This effect
may be a result of condensation; it is possible that the SO3/H2S04
content in a coal-fired unit  is highest at the boiler, and as the
gas progresses through  the unit and cools off, a fraction of the
SO3/H2SC>4 condenses.  The greatest amount of condensation would
probably occur in the air preheater, where there is a substantial
temperature drop.
       Table 1.   Sulfur Trioxide Content at TVA Power Plant
                    vs. Excess Oxygen in Flue Gas
Plant Unit
Allen 1




Shawnee 3








Widows 7
Creek




Widows 8
Creek


Capacity Excess 02
(MW) Furnace Design (%)
330 Cyclone 2.8
2.8
3.0
3.3
4.1
175 Front Wall 2.7
2.8
3.2
3.5
3.8
4.5
4.5
4.8
5.9
575 Tangential 3.2
3.5
4.0
4.5
4.5
4.5
550 Tangential 3.0
3.3
3.3
4.1
[S03/(S02 + S03)]
(%)
2.3
2.2
2.3
2.3
2.6
0.19
0.18
0.38
0.35
0.40
0.56
0.34
0.50
0.58
0.80
0.95
1.1
1.3
1.2
1.1
0.60
0.70
0.57
1.5
                               32

-------
    Table 2.   Sulfur Trioxide Content vs.  Flue Gas Temperature
               at Air Preheater Outlet of  Cumberland
                        Steam Plant Unit 2


Experiment   Flue Gas Temperature Range   S03 Concentration Range
  Series     	(C)	(PPm)	

    I                152-154                      2-5

    2                162-167                      6-7

    3                168-173                      9-10
MEASUREMENT OF THE EMISSION OF SULFUR TRIOXIDE AND SULFUR DIOXIDE
FROM A TVA COAL-FIRED POWER PLANT

     The TVA Division of Environmental Planning has conducted two
plume chemistry studies at Cumberland Units 1 and 2, both horizon-
tally opposed 1275-MW units located in Cumberland City, Tennessee,
to investigate the oxidation rate of SO2 in the plume.  Data on
the [S03/(S02 + S03)] fraction in flue gas were needed  to deter-
mine the fraction of sulfur dioxide that was already oxidized when
the gas was emitted to the atmosphere.  To obtain these data, flue
gas samples, drawn from ports  located about 4 m from the base of
the stack, were analyzed for S03/H2S04 and S02.


Methodology

     A modified Lisle and Sensenbaugh apparatus  (Figure 3)  was used
to collect samples.  The probe used in the  first  field  study was
Teflon-lined and contained a Pyrex wool plug;  the portion of the
probe between  the duct wall and  condenser  was  heated.   The  condenser
 (Figure 4) was positioned in a waterbath  held  at  60C  to 70 C.
Sulfur dioxide was collected in  two midget  impingers,  each  con-
taining a  solution of 5% H202; these  impingers were followed by
an  impinger  containing Drierite. Volune  was measured  by monitor-
 ing  a calibrated vacuum  gauge  positioned  in  front of  the pump.
The  samples  were not collected isokinetically;  typical  flow rates
were about 1.4  1/min.  After  sample  collection,  the probe  and  con-
denser were  rinsed with  80%  isopropanol.   The  p^be rinse  solution
was  filtered,  and both solutions were titrated with 0.005  barium
 perchlorate  [Ba(C104)2]  in  80% isopropanol;  thorin  was  used as  the
 indicator.   To analyze  for  S02,  the  H202  solutions  were boiled  to
 destroy  the  excess H202  and then titrated with 0.2  N NaOH.   Brom-
 phenol  blue  was  used  as  the indicator.
                                 33

-------
                                                     5% H2O2SOLUTION
VACUUM

GAUGE
                                              2  4        A                    /
                                            CONDENSER   /  FMPTY   DRIERITE     /        /
                                                /        / \EMPTY    /  VALVE ^     	/
                 HEATING TAPE
STACK
                                                                                      PUMP
* II 1 /
GLASS WOOL PLUG 	
TEFLON-LINED PROBE 	 
IMPINGERS
DU

CT


                                                                                 ICE

                                                                                 BATH
        Figure 3.  Apparatus used in first study of S02 and SOg emissions

                  from Cumberland Steam Plant.

-------
Figure 4.   Condenser used in studies of S02 and SO., emissions
           from Cumberland Steam Plant.
                              35

-------
     In the second  field study, mass loading data were also col-
lected.  For this reason a Research Appliance Company (RAC) iso-
kinetic stack sampling apparatus was modified so that it .could be
used for collection of either mass loading or S03/H2S04 arid SO2
samples.  When set  up to obtain SO3/H2S04 and SO2 data, the RAC
apparatus differed  from the Lisle and Sensenbaugh apparatus in
four ways:  (1)  The condenser was positioned in a heated compart-
ment instead of  in  a waterbath, (2} modified Greenberg-Smith impingers
containing 100 ml of 3% H2O2 solution were used instead of the
midget impingers, (3) a dry test meter was used to measure volume,
and (4) the samples were collected at flow rates ranging from 4.5
to 20.Of /min.   The analytical procedure was the same as for the
first study.


Results and Discussion

     The results of these two studies are presented in Table 3.
In the first study,  the average values for the [S03/{S02- + S03) ]
fractions (percent)  were 0.26 for Unit 1 and 0.22 for B-nit 2.  The
corresponding average excess oxygen contents were 5*9% and 41%-
In the second study, the [SC>3/(S02 + S03)] fractions (percent) were
0.19 for Unit 1  and 0.59 for Unit 2:  the excess oxygen contents
were 5.5% and 6.7%,  These data support trends which- were previ-
ously discussed.  The values of the [S03/(SO2 + SO3)] fraction at
Cumberland, like those for Shawnee Unit 3 and Widows Creek Units
7 and 8, are low compared with the fraction found for the cyclone
fumac* at Allen Unit 1.  For both units, the [SO3/(SO2 + SO3)]
fraction increased  with increasing excess oxygen content, similar
to the behavior  observed at Shawnee Unit 3,  Widows Creek Units
7 and 8, and Allen  Unit 1.
      .Table 3.  Sulfur Trioxide ajid Sulfur Dioxide Emissions
                 Measured at Cumberland Steam Plant
'
'*

Capacity Furnace Field
Unit
1

2

Excess
02
(MW) Design Study (%)
1275 Horizontally
opposed
1275 Horizontally
opposed
1
2
1
2
5.9
5.5
4.1
6.7

SO 3
(ppm)
5.4
7.0
7.5
18.0
i
S02
(ppm)
2790
3470
2720
3020
S03
S02 + SO 3
> f Cf \ 
V. * )
0.26
0.19
0.22
0.59
                                 36

-------
STACK APPARATUS EFFICIENCY TESTS

     Two experiments were conducted to determine the efficiency of
the modified RAC stack sampling apparatus for collecting S03/H2S04.


Methodology

     In the first experiment, as part of the field study procedure,
the probe and Pyrex wool plug were rinsed separately from the con-
denser; the probe and plug rinse solution was titrated- separately
from the condenser rinse to determine the fraction of S03/H2SO4
that remained in the probe.  In the second experiment, the effi-
ciency of the condenser for collecting S03/H2SO4 was determined.
Two condensers were connected in series in the modified RAC appar-
atus and the temperature was maintained between 60C and 70C
for both condensers.  Flue gas was then sampled from the base of
the stack at TVA's Colbert Unit 1, a 200-MW front-wall unit located
near Pride, Alabama.  Both condensers were rinsed after sample col-
lection, and the rinse solutions were analyzed for SO3/H2S04 using
the Ba(ClO4)2 titration procedure.

     The results of the first experiment (Table 4) showed that on
the average 68% of the S03/H2S04 was found in the probe for flow
rates ranging from 0.0057 to 0.020' m3/min.  Results of the second
experiment  (Table 4) indicate that on the average the first con-
denser was 83% efficient for flow rates ranging from 0.0051 to
0.0014 m3/min.  Combining the efficiencies of the probe and first
condenser gives a total system efficiency for S03/H2S04 collection
of 95%.
CONCLUSION

     The results of these studies suggest  that  the  SO3/H2S04 con-
tent of power plant flue gas depends on  the  excess  oxygen  content
of the flue gas, furnace design, and temperature of the  flue gas.
Cyclone furnace units may produce higher percentages of  S03/H2SO4
in their flue gases than do tangential and front-wall furnaces.
However, this should be verified by sampling at other cyclone units,
The S03/H2S04 content of flue gas was found  to  increase  with in-
creasing excess oxygen content  of the flue gas  and  to decrease
with decreasing flue gas temperature.  Because  the  temperature of
the flue gas decreases as the gas passes through a  unit, it is
possible that SO3/H2SO4 condenses inside the unit,  particularly
in the air preheater, where a significant  temperature drop occurs.
                                37

-------
      Table 4.   Efficiency  Tests for S03  Collection System
Determination of Fraction of SO3 Found in the Probe
and in the Condenser
Sample
No.
1
2
3
4
5
6
7
8
Average
Fraction in
Probe
0.63
0.66
0.73
0.71
0.68
0.65
0.67
0.70
0.68*
Determination of Fraction of
Sample
No.
1
2
3
4
5
6
Average
Fraction in
1st Condenser
0.82
0.88
0.91
0.76
0.82
0.78
0.83*
Fraction in
Condenser
0.37
0.34
0.27
0.29
0.32
0.35
0.33
0.30
0.32*
SOa Found in the
Fraction in
2nd Condenser
0.18
1.12
0.09
0.24
0.18
0.22
0.17*
Flow Rate
(m3/min)
0.011
0.0082
0.014
0.020
0.020
0.0057
0.020
0.0068

Condenser
Flow Rate
(m3/min)
0.0051
0.0054
0.0085
0.010
0.013
0.014

     Therefore,  total  collection  efficiency of the system is
0.68 + 0.83(0.32)  =  0.95.
                              38

-------
     Sampling of emitted flue gases at one power plant showed that
the emission rate for S03/H2S04 was about 0.3% of that for S02.
The efficiency of the method of sampling for S03/H2S04 was found
to be satisfactory.


REFERENCES

1.   Office of Air Quality Planning and Standards.  Position Paper
     on Regulation of Atmospheric Sulfates.  EPA-450/2-75-007,
     U.S. Environmental Protection Agency, Research Triangle Park,
     North Carolina,  1975.  87 pp.

2.   Science Applications, Inc.  Effects of Sulfur Oxides on the
     Lung:  An Analytic Base.  EPRI 205 (SAI-75-566-LA),  Electric
     Power Research Institute, Palo Alto,  California, 1975.   206 pp,

3.   Hutchinson, T. C., and L. M. Whitby.   The Effects of Acid
     Rainfall and Heavy Metal Particulates on a Boreal Forest Eco-
     system Near the Sudbury Smelting Region of Canada.  Water,
     Air and Soil Pollution, 7:421-438, 1977.

4.   Braekke, F. H.  Impact of Acid Precipitation on Forest  arid
     Fresh-water Ecosystems in Norway.  SNSF Fagrapport FR6.  Aas-
     NLH, Norway, 1976.  Ill pp.

5.   Lisle, E. S., and J. D. Sensenbaugh.   The Determination of
     Sulfur Trioxide and Acid Dew Point in Flue Gas.  Combustion,
     36(7):12-16, 1965.
                               39

-------
Measurement of SO3/H2SO4 Concentration in
Kraft Recovery Furnace Stack Gas Using
Controlled Condensation
Ashok K. Jain
R. 0. Blosser
Howard S. Oglesby
National Council for Air and Stream Improvement
     ABSTRACT

     Tests were  conducted to determine the effects of sample
     flow rate,  condenser temperature, and frit porosity upon
     the efficiency of controlled condensation equipment to
     measure parts per million levels of S03/H2S04 in a gas
     stream containing up to 30% moisture.  Further  laboratory
     studies were conducted using particulate from several
     kraft recovery furnaces to study the effect  of  particu-
     late upon the passage of HgSCU vapors through a quartz
     filter loaded with particulate.  The results showed that
     H2S04 losses in  the filter ranged from 10% to 60% of
     the inlet concentration and depended upon particulate
     characteristics.


 INTRODUCTION

     The recovery of  chemicals from spent kraft cooking liquor (black
 liquor) involves combustion of concentrated black liquor, which
 consists of wood components, inorganic chemicals, and sulfur com-
 pounds in kraft recovery furnaces.  The presumed  similarity of sulfur-
 containing fossil fuels and black liquor has raised  the question of
 the possible presence of sulfur trioxide (S03) and sulfuric acid
 (H2S04) in kraft recovery furnace flue gas, although liquor is
 burned in a manner  to minimize S02 generation.

     This paper describes the results of laboratory  studies carried
 out to date to develop a technique to measure S03/H2S04 levels in
 kraft recovery furnace stack gas using controlled condensation.
                               41

-------
DESCRIPTION OF EXPERIMENTAL APPARATUS AND METHODS
Sulfuric Acid Evaporator

     A sulfuric acid evaporator, as shown in Figure 1, was used in
the laboratory to  generate test gases containing desired concentra-
tions of moisture  and H2S04.  The evaporator was a quartz tube
packed with glass  beads and wrapped with a heating element and
insulation.  A coarse quartz frit was installed inside the evapora-
tion tube to provide uniform concentrations of H2S04.  Sulfuric
acid was added to  the evaporator with a low flow rate Masterflex
pump manufactured  by Cole-Parmer Instrument Co.  The temperature
of the evaporator  was maintained at 650F.

     In those tests in which the performance of particulate filter
was evaluated, an  additional quartz conditioning tube was added to
the evaporator as  shown in Figure 1.  The conditioning tube was
wrapped with a heating element and insulation and allowed the
sample stream to be split into two fractions.  One fraction went to
the filter holder  (sample train), and the other went to impingers
(reference train)  for determining sulfuric acid concentration in
the sample stream.


Controlled Condensation Coil

     The S03 condenser is shown in Figure 2.  The inside diameter of
the glass tubing for condensing sulfuric acid was 5 mm, and the frit
diameter was 45 mm.  Three different condensers were evaluated
during the study.  The first condenser had approximately three feet
of glass tubing and a coarse frit (Ace Glass Type B); whereas the
second and third condensers had approximately six feet of glass
tubing and Ace Glass Type C and D frits, respectively.  During the
tests the condenser was filled with water, and its temperature was
maintained at 175-185F with a heating mantle.


Filter Holder Design and Particulate Loading

     Filter holders typically used in EPA Method 5 sampling trains
were used initially.  However, these filter holders are not de-
signed for use above 400F.  For higher temperatures a quartz filter
holder shown in Figure 3 was used.  The details of the design are
specified elsewhere (1).
                                42

-------
 \ Air
             H2S04 +  H20
                                                             Temperature
                                                        to Impingers
Figure 1.  H2S04 evaporation assembly.

-------
  SULFUR  TRIOXIDE CONDENSER
              SINTCRED GLASS FILTER-
    GAS SAMPLE IN
         GLASS JACKET
                             WATER
Figure 2.  Sulfur trioxide condenser.
                   44

-------
                 18/9
                 BALL
                                       SPRING
                                       ATTACHMENT
                                       HOOKS
TISSUE QUARTZ
FILTER
                                 18/9
                                SOCKET
en
               THERMOCOUPLE
                  / WELL
                                     STANDARD
                                     TAPER QUARTZ
                                     40/50
 SEAL
 EXTENT1ON
 TO STD.
 TAPER JOINT
EXTRA COARSE
QUARTZ FRIT
            Figure 3.  Quartz filter holder.

-------
     The filter medium used in all tests was Tissuequartz 2500 QAO,
manufactured by Pallflex Products Corporation, Putnam, Conn.  In
those tests in which the effect of recovery furnace particulate
upon H2S04 loss was to be determined, the filter was loaded with
measured amounts of particulate by slowly adding particulate
directly to the filter while drawing air through the filter at a
high rate.


      Sampling Train
     The sampling train for determining sulfuric acid concentration
in the gas stream consisted of two Greenberg Smith impingers
containing isopropyl alcohol, an impinger containing water, a
column of silica gel, a leakproof vacuum pump, a rotameter, and
a calibrated dry gas meter.


Method of Sulfate Analysis

     The samples were analyzed for their sulfate content by
titrating them with barium perchlorate and using Thorin indicator  (2),
Interference of sodium ions in Thorin titration was also investi-
gated.


Determination of SOa Condenser Efficiency

     Figure 4 shows the schematic of the sampling system used for
determining the efficiencies of H2SO4 capture in S03 condensers.
Tests were conducted at different flow rates through the condensers
and with various H2S04  concentrations in the gas stream.  Each test
lasted approximately 30 minutes.  At the end of the test the
condenser was washed and analyzed for sulfuric acid retained in the
condenser.  The impinger solution was also analyzed to determine the
amount of H2S04 passing through the condenser.


Determination of H2S04  Losses in Filter Holders

     Initial tests to determine the retention of H2S04 vapors in
filter holders were conducted at 350-400F, and filter holders used
in EPA Method 5 sampling trains were used.  The schematic of the
sampling system was the same as used for condenser evaluation, except
that the filter holder was installed in the sampling train in place
of the SO3 condenser.
                               46

-------
     The tests with quartz filter holders used the schematic shown
in Figure 5.   Tests were conducted to determine H2S04 loss in the
filter holder when (1)  there was no particulate filter holder in
the sample train,  (2)  there was quartz filter holder in the sample
train, and (3) the quartz filter holder was loaded with particulate
samples from  different  recovery furnaces.  The amount of particulate
on the filter paper was equivalent to an average particulate concen-
tration of approximately 0.2 gm/1000 liters dry gas.


RESULTS AND DISCUSSION
Method of Sulfate Analysis

     Thorin titration was found satisfactory in our studies and gave
sharp end points at low concentrations.  However, since sodium
constitutes a large fraction of recovery furnace particulates, tests
were conducted to determine sodium ion interference with Thorin
titration.  The results are recorded in Table 1 and show that sodium
ion increased the volume of titrant required and caused a positive
interference in sulfate analysis.


          Table 1.  Sodium Interference with Thorin Titration

                        Vol 0.01 N                  Vol 0.022 N
No.	NaNC-3 (ml)	Ba(ClO4)2 Titrant (ml)
1
2
3
4
5
0
10
20
30
40
9.10
9.15
9.30
9.45
9.65
Efficiency of S03 Condensation Coil


     Effect of Frit Porosity and Coil Length on Condenser Efficiency-
The results of tests aimed at determining  the  effect of  frit
porosity upon the efficiency of the SO3 condenser are  recorded in
Table 2.  During these tests the temperature of the condenser was
maintained between 175-185F.  The low efficiency of  H2S04 capture
with the Type B frit may partially be due  to smaller length of the
condensing coil.  The Type D frit gave very high capture efficiency
                                47

-------
                H2S04  +  Water
 Air
           EVAP.
CONDENSER
Impingers
 Figure 4. Schematic for determining condenser efficiency.
                H2S04 +  Water
 Air
           EVAP.
  FILTER
                   Impingers
Figure 5.  Schematic for determining filter losses,
Impingers
                            48

-------
but did not allow sample flow rates higher than 2 lit/min.  The
Type C frit was found to be most acceptable from the point of view
of sample flow rate and efficiency of H2SO4 capture.


      Table 2.  Effect of Frit Porosity Upon Condenser Efficiency


                                  Length of
                 Max. Pore      Condensing Coil  Flow Rate  % H2S04
No.  Frit Type* Dia. Range (MM)      (ft)        (lit/min)  Capture
1
2
3
B
D
C
70-100
10-20
25-50
3
6
6
2
2
2-8
20-40
99
90-99
     *Ace Glass
     Effect of Flow Rate on Condenser EfficiencyThe results of
tests aimed at determining the effect of flow rate on the efficiency
of S03 condenser are summarized in Table 3.  The results show that
the efficiency of H2S04 capture varied from 90.0% to 99.4% during
these tests, and the average efficiency of H2SO4 capture was slightly
higher during high flow rate tests.
      Table 3.  Effect of Flow Rate on Efficiency of S03 Condenser
Flow Rate
(lit/min)
3.5
9.5
No. of
Runs
7
13

Max .
96.8
99.4
% H2S04 Capture
Min.
90.0
91.1

Avg.
93.3
95.4
     Effect of H2SO4 Concentration on Condenser EfficiencyTests
were conducted in the concentration range  of  1 ppm  to 50 ppm to
determine if there was any  effect of H2S04 concentration on its
efficiency of capture.  The results are  summarized  in Table 4.  All
tests were made at flow rates  of approximately 9  lit/min.  The
                                49

-------
results show that the efficiencies of capture were generally higher
than 95%, except when 50 ppm H2S04 was present in the gas stream.
However, only two runs were made at high H2SO4 concentration.

     When a "t" test was performed to test the hypothesis that the me
efficiencies of capture were different for 1 ppm and 50 ppm H2SO4
samples, it was concluded that there was insufficient evidence to
indicate that the two efficiencies of capture were different at the
0.05 level of significance.


     Table 4.  Effect of H2S04 Concentration on Capture Efficiency
H2S04 Cone.
(ppm)
1
5
20
50
No. of
Runs
6
4
1
2

Max.
99.2
99.4
95.9
91.7
% H2S04 Capture
Min.
92.0
94.7
95.9
91.1

Avg.
95.0
97.8
95.9
91.4
Loss of H2S04 in FilterJiolders

     EPA Method 5 Type Filter HoldersTests with EPA Method 5
filter holders were conducted by maintaining the filter holders in
the temperature range of 390 to 440F.  The results of six tests
showed that on an average 45% of the H2S04  present in the gas stream
was retained in the filter holder.  These results indicated the
need to maintain the filter holder at higher temperatures, which
was not possible with this type filter holders.


     Quartz Filter HolderThe results of tests to determine the
effect of the quartz filter assembly upon the passage of H2SO4 are
recorded in Table 5.  A comparison of the tests with and without
the filter support assembly indicates that there was no significant
loss in the filter assembly,  and most of the variability was in the
sampling and analytical technique itself.
                                50

-------
    Table 5.   Effect of Quartz Filter Assembly Upon Sample Recovery
No. of
Runs
9
12
Ref. Train H2S04 Filter Assembly
Cone. Range (ppm) Present
5.8-18.2
5.6-18.0
No
Yes
Max . b
11.4
14.9
o
% Error
Min.
1.7
1.8
Avg.
0.0
1.8
   Error =     .-  Sample)  x 100
 Max. Error = Max.  Abs.  (% Error)
CMin. Error = Min.  Abs.  (% Error)

dAvg. Error = Abs.  ( 2 (% Error)/No.  Runs)


Filter Holder H2S04 Losses in the Presence of Particulate

     The data in Table 6 show a summary of the tests to determine
H2S04 losses in the filter holder when the filter was loaded with
particulate samples from different recovery furnaces.  During the
tests the amount of particulate on the filter was approximately
0.2 gm/1000 lit dry gas.  The results showed large variations in
H2S04 loss in the particulate holder with average losses varying
from 7% to 55%.  The particulate samples were also analyzed
for their alkalinity to determine if the filter loss could be
correlated to alkalinity.  However,  no such correlation was
observed in Table 6.


    Table 6.  Filter Holder Losses for Different Particulate Samples
Source
A
B
C
D
E
F
G
Part. Alkalinity
gm CaCOs/kg
32.4
9.4
45.0
39.0
0
8.4
6.3
No. of
Runs
7
1
3
12
3
3
1
/
Max .
13
27
62
G2
38
15
19
ID H2S04
Min.
2
27
45
29
23
3
19
Loss
Avg.
7
27
55
48
32
8
19
                                51

-------
SUMMARY

1.   Thorin titration was found acceptable for sulfate analysis.
     However,  sodium ions caused a positive interference.

2.   The efficiency of H2SO4  capture in S03 condensers was found
     to be dependent upon the frit porosity and coil length.

3.   The flow rate and concentration of H2S04  did not significantly
     affect the efficiency of H2SO4 capture in the condenser with
     Ace Glass Type C frit.

4.   The quartz filter assembly did not cause  any loss of  H2SO4 when
     it was maintained at 600-650F.
5.   In the presence of particulate, HjSOA 'losses in the filter
     holder varied from 7% to 55% for different particulate samples
     and could not be correlated with particulate alkalinity.


ACKNOWLEDGMENT

     The authors acknowledge the contribution of Mr. John Ruppers-
berger of the Environmental Protection Agency.  Work is being
partially supported by USEPA Industrial Environmental Research
Laboratory under Contract R-804644-01-0.


REFERENCES

1.   Process Measurement Procedures Sulfuric Acid Emissions.  Pre-
     pared for: Industrial Environmental Research Laboratory, U.S.
     Environmental Protection Agency, Research Triangle Park,
     North Carolina, Contract No. 68-02-2165, February 1977.

2.   Method 8 - Determination of Sulfuric Acid Mist and Sulfur
     Dioxide Emissions from Stationary Sources.  Federal Register
     36, December 23, 1971.  247.
                                52

-------
Characterization of Combustion Source Sulfate
Emissions with a Selective Condensation
Sampling System
James L Cheney
James B. Homolya
U.S. Environmental Protection Agency
     ABSTRACT

     Present studies of sulfate emissions  include an exam-
     ination of both the formation of  sulfate during the
     combustion of sulfur-containing  fuels and the long-
     range  impact of sulfates on the  atmosphere.  Like any
     attempt to inventory source emissions, confirmation of
     the  sulfate emissions relies heavily  on the measurement
     methodology used.  Our work has  emphasized not only
     the  measurement of total sulfate  emissions, but also
     the  characterization of these sulfates.  Using this
     approach, a reliable impact can  be  determined by pre-
     dicting the mass emission rates  of  individual sulfate
     components.

     A substantial number of current  source sulfate measure-
     ments  involve the separation of  unassociated H2S04 from
     particulate-related sulfate and  the subsequent recovery
     and  quantitative determination of the separate components.
     At present, most of these efforts involve high temperature
     gas-solid separation by filtration, followed by collec-
     tion of the acid in a temperature-controlled Goksoyr-Ross
     coil.  We have performed controlled sampling experiments
     using  the selective condensation  method to determine its
     reliability in characterizing sulfate emissions.

     The  design of the manual sampling equipment and experi-
     ments  will be described, including  studies of collection
     efficiency, separation, and isolation of the sulfate
     components.  The collected data  will  be presented, and
                               53

-------
     in-depth examination of the data collected as sampling
     variables were changed will be discussed, along with
     suggestions for possible future efforts using this
     sampling approach.


INTRODUCTION

     A considerable amount of effort is currently in progress to
establish the correlation between sulfates emitted from combustion
sources and those measured in the atmosphere.  In addition to
efforts (1) to relate sulfate emissions to local and long-term
atmospheric impact, further efforts involve sulfate emission measure-
ments for evaluating the effectiveness of current control technology.
During the course of the past few years,  a need has developed (2)
for a sulfate characterization method.  The particular emphasis in
sulfate characterization studies, both with respect to atmospheric
effect as well as the emission control studies, has been to determine
sulfuric acid (H2S04) as a separate entity from total sulfate.  For
the high temperature combustion source, the H2SO4 which is unasso-
ciated with particulate matter usually behaves (3) as a gaseous
emission, in contrast to the inorganic sulfate or particle-absorbed
acid.

     Ideally, then, a sulfate measurement method is desired that
will simultaneously provide the concentration of both particulate
sulfate and the concentration of H2S04.  For the purpose of clari-
fication, particulate sulfate is defined here as all metal sulfates
as well as H2S04 absorbed on the particles.  For such a method,
emphasis has been placed most recently on the Shell (4) selec-
tive condensation approach.  The method involves the extractive
sampling of the flue gas, followed by high temperature separation
of the particulate fraction, with a subsequent collection of H2SO4
in a temperature-controlled Goksoyr-Ross condensation coil.  Problems
have been encountered with this approach, however, which include the
usual problem of using bulky glassware for field sampling, high
pressure drops across the backup aerosol collection frit in the
Goksoyr-Ross coil, and an inadequate gas-particle (2) separation
during sampling.

     Prior evidence indicates that an apparent equilibrium between
H2S04 in the gas phase and particulate matter can exist in flue
gases.  The equilibrium between the gaseous acid component and
the particles is dependent on, among many variables, contact time,
concentration of acid and particles, and temperature.  The
difficulty of characterizing the sulfates by this method is in
sampling so as not to disturb the gas-solid equilibrium.
                                54

-------
     Furthermore,  in order to provide an accurate measure of
particulate concentration, isokinetic sampling will be necessary.
As stack gas velocities vary with fuel, excess 02, load, and other
parameters, different sampling velocities would be necessary.

     As there appeared to be minimal available data which pertained
to the effects of  sampling rates, acid collection efficiency, and
possible alternatives to the selective condensation approach, a
study was undertaken to establish such data.  This work.presents
the effects of sampling rates on the gas-particulate sulfate dis-
tribution in a selective condensation sampling system.  Also
addressed in this  study are the collection efficiency of the
Goksoyr-Ross condensation coil and a potential alternative method
for isolating gaseous H2SO4.


EXPERIMENTAL

     All of the sulfate measurements were performed on a 100,000
lb./ hr. oil-fired steam boiler which operates solely as a heat
source for a local college campus.  Throughout the tests, an
excess oxygen of 4% to 5% was utilized to burn a nominal 2.2%
sulfur and 350 ppm (w/w) vanadium residual fuel oil.  The sampling
was conducted just subsequent to the last boiler pass but prior to
the preheater and  stack.

     The manual selective condensation sampling system used during
the measurements is depicted in Figure 1.  A standard 10.2 cm (4
inch) port cap with a centered 2.54 cm (1 inch) Swagelok* fitting
holds a .9 M (3 foot) long, 2.54 cm (1 inch) O.D. heated quartz
probe.  Following  the probe is a high-temperature, all-quartz
filter holder.  Both the probe and filter holder are heated and
temperature controlled with separate variable voltage controls
located in the sampling module.  Also located in this module are
four thermocouple  readouts which provide temperature readings for
the flue gas, probe, filter, and condenser.

     The filters used for the sulfate measurements were fabricated
from high-purity quartz and were shown (5) to be 99.9% effective
for collecting particle sizes down to 0.3 /um.  During the study
the filter was maintained between 265C and 288C during sampling.

     Following the high temperature filter is the Goksoyr-Ross acid
condensation coil  which rtilizes a medium porosity frit.  The purpose
of the frit is for the collection of any acid aerosol which has not
been collected by  impaction in the preceding temperature-controlled
      Mention of brand names does not imply EPA endorsement.
                                55

-------
en
en
                       INSULATION
          PLUG HOLDER
                                                             HEATING MANTLE
                                 OOBDOOeo
                                     CONDENSATION COIL
T  T
              SAMPLING MODULE
                                                                PUMPING METER BOX
                          Figure  1.  Schematic drawing of manual acid condensation system.

-------
spiral coils.   During some of the measurements, a specially con-
structed glass wool plug holder (also shown in Figure 1) was used
either as a backup or substitute acid collector to the condenser.
As with the condenser,  the plug holder was temperature controlled
by circulating water from the water bath in the sampling module.

     Following the temperature-controlled acid collection devices
is an ice bath with isopropanol (IPA) and peroxide (3% H202)
impingers for collecting any acid penetrating the previous devices
as well as the sulfur dioxide (S02).

     After the impingers is a gas meter-gas pump box.  Sampling
rates were determined by timing the gas meter revolutions with the
second-hand of a stopwatch.

     Each sample run consisted of separate fractions collected from
the probe, filter, condenser, backup filter plug, IPA, and H202.
Analysis of each fraction was accomplished by a Barium-Thorin
titration (6).  It was necessary to pass the sample fractions
recovered from the probe wash and filter through a mixed bed resin
to avoid cation interference to the titrations.


RESULTS

     Three basic series of measurements were performed:

     1.   a series of measurements  to establish sampling rate
          effects on sulfate distribution  in the sampling  train

     2.   a series of sulfate measurements to  evaluate  the  collection
          efficiency of the Goksoyr-Ross coil

     3.   a series of measurements  to establish if the  condenser
          approach to acid collection could be substituted  with  a
          more simple approach.

     Sampling was conducted  over a  sampling rate interval of 2 to
 16 1/min.  The results  of  the  sulfates  collected, in terms  of
 total  sulfate and percent  occurrence of total  sulfate in three
 fractions, are presented  in  Table 1.

     A second series of tests  were  performed  to establish  the col-
 lection efficiency of the Goksoyr-Ross  condenser.  The  results  of
 these  measurements are  presented  in Table  2.   The sulfate  concen-
 trations are  presented  as ppm  on  a  vol/vol basis.
                                 57

-------
            Table 1.  Sulfate Distribution and Sampling Rate
   Sample
     Rate
         (1)
T-SO.
     (2)
%-P
   (3)
%-F
   (3)
% "Acid"
16
12
8.
7.
6.
5.
4.
3.
2.
(1)
(2)
(3)


3
5
5
1
7
8
3










In 1/min
36
38
35
31
27
41
37
32
50

.3
.6
.8
.6
.2
.9
.8
.8
.1

Total sulfate collected,

Percent
of
filter (F),
(4)

Percent
of
total sulfate
respectively
total sulfate
10.7
13.9
12.3
13.6
15.8
18 .3
36.0
19.1
23.9





















expressed
collected
, in
ppm
passing
9
7
13
16
17
10
12
30
36

as
in
.9
.8
.9
.8
.8
.6
.2
.0
.9

ppm
79
78
73
69
66
71
51
50
39

(vol/vol)
.3
.4
.8
.6
.4
.1
.8
.9
.2


probe (P) and
(vol/vol)
probe
and
filter

           Table 2.   Total  Sulfate  With and Without Plug
Sample
1
2
3
4
5
6
> PW
4.4
4.3
6.9
8.4
8.9
4.2
Filt
5.0
5.3
5.4
3.5
4.1
6.9
Cond
25.6
21.2
26.5
26.8
21.7
21.2
Plug
0.4
0.7
0.7
1.6
2.7
2.8
S02
824
745
944
805
768
717
*-T(c)
4.12
3.98
4.21
4.72
4.49
4.37
%-T(w;
4.16
4.16
4.28
4.90
4.81
4.73
PW = probe wash;  Filt  =  filter;  Cond = Goksoyr-Ross condenser;
plug = glass wool plug;  and  S02  =  H202 catch.   %-T(c)  and
%-T(wp) = Total S04/Total  S04  A S02) x 100% for condenser and
for condenser and wool plug,  respectively,  using ppm (v/v) .
                               58

-------
     The sampling arrangement used to obtain the data in Table 2
consisted of a Goksoyr-Ross condensation condenser followed by the
glass wool plug as a backup acid collector.  For samples involving
glass wool,  untreated,  reagent grade Pyrex wool which showed no
sulfate blank was used.  Both the condenser and glass plug were
maintained at 60C during sampling.  A slight increase in acid
collection is apparent  when including the plug catch as the
average S02  conversion  factor (%-T) increased from 4.37% to 4.74%.

     The final series of sulfate measurements involved the
substitution of the Goksoyr-Ross condensation condenser with
the temperature controlled glass plug holder.  During these
measurements, the glass plug was maintained at 60C during
sampling.  The results  of several measurements utilizing the
two sampling train arrangements are presented in Table 3.


    Table 3. Acid Collection with Condenser and Glass Wool Plug
           Condenser
                                Glass Plug
Sulfate
       (1)
SO-
   (1)
1  m
01
   (2)
Sulfate
       (1)
SO,
   (1)
,(2)
35.5
31.6
42.2
41.5
38.8
35.6
824
745
944
805
768
717
4.17
4.07
4.28
4.90
4.81
4.73
31.5
37.8
58.5
33.7
30.7
31.3
707
896
858
777
649
728
4.27
4.05
6.38
4.16
4.52
4.12
Averages:
 Sulfate
                      (1)
         SO,
                      (1)
                         ,(2)
Condensation
Glass Plug
37.6 + 3.4
37.4 + 7.4
801 +
769 +
57
58
4.49
4.58
+ 0.32
+ 0.43
     (1)  In ppm (v/v)
     (2)  %-T = Total S04/(Total S04 + S02) x 100%
     Within experimental error,  there  appears  to  be  no  difference
in sulfate collection.  The  average  standard deviation  of +_0.32 to
+0.43 represents a 7.13% to  9.39% deviation for condenser and glass
plug sulfate conversion, respectively.
                                59

-------
DISCUSSION

     A very significant effect on sulfate measurements (in terms
of differentiating gaseous H2SO4 from particulate related sulfate)
occurs as the sampling rate varies.  For combustion source
measurements of sulfates, the flue gas stream velocity can vary
two or threefold from one source to another, and this effect on
acid-particle separation presents a special problem.  For example,
for a flue gas at 300F (150C) and 10% moisture, a variation  from
25 to 75 ft/sec. (7.6 to 22.8 m/sec.) in velocity will result  in
isokinetic sampling rates from 5 to 15 1/min. when using a 3/16
inch nozzle.  From the data in this study, the resulting fraction
of the total sulfate collected as acid beyond the filter would
be from 61% to 75%.  While isokinetic sampling using this
procedure would result in a correct total sulfate value, the
ability of the procedure to serve as a characterization method
appears quite limited.

     Apparently little acid penetrated the coil and frit during
the backup plug collections during this study.  A slight increase
in calculated S02 conversion to sulfate does occur, and, as
the data are presented in the chronological order of collection,
the possibility of some phenomena involving acid retention and acid
carryover from the frit may be occurring.  This, however, is highly
speculative with the limited amount of data presented here.

     This study has also shown that in the sampling procedure
presented an equivalent acid collection component can be used
to replace the awkward condensation condenser.  While the
temperature of the glass plug holder must be maintained above the
H20 dewpoint and below the acid dewpoint, requiring the waterbath
or some type of heat source, it offers a simpler method of acid
collection than the condenser.  For sources with high negative
pressures, the additional pressure drop in the sampling system
(12 in. Hg for the medium frit) makes high sampling rates
difficult to achieve.  During the study a different glass plug
was used in the plug holder for each sample collected.  To be a
convenient method,  the plug should be able to collect acid on a
repeatable basis.  Whether this is possible has yet to be
determined.

     In conclusion, while the selective condensation approach
appears to be the best sulfate measurement method available at
present, it must be used with caution, particularly with respect
to interpretation of results.  Since the problem involves the
filtration process, placing the filter on the sample inlet end of
the probe (in the stack) would eliminate acid-particulate asso-
ciation in the probe; however, the same problem of acid retention
                                60

-------
at the filter interface would still remain.   Use of the condensa-
tion procedure for sulfate measurements should,  therefore,  be
performed by a well-defined procedure if comparisons of results
either from separate sources or by separate  sampling personnel
are to be meaningful.


REFERENCES

1.   Homolya,  J.  B., and J. L.  Cheney.  A Study  of the Local Ambient
     Air Impact of Primary Sulfate Emissions from the Combustion
     of Residual  Oil.   Presented at the 175th Annual ACS,  March
     1978.

2.   Cheney, J. L.,  J.  B. Homolya, and H. M. Barnes.  Measurement
     and Identification of Primary Sulfates  Emitted from Combustion
     Sources.   70th Annual AICHE,  New York,  Nov. 1977.

3.   Cheney, J. L.,  C.  R. Fortune, J. B. Homolya, and H. M. Barnes.
     The Application of an Acid Dewpoint Meter for the Measurement
     of S03/H2S04 Emissions.  Proceedings of the Fourth Annual
     Conference on Energy and the Environment, Cincinnati,  Ohio,
     1976.

4.   Goksoyr,  H., and K. Ross.   Shell Report No. M-211,
     Thornton Research Center,  1962.

5.   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, Nov. 1973.

6.   Cheney, J. L., W.  T. Winberry, and J. B. Homolya.  Evaluation
     of a Method for Measuring Primary Sulfate Emissions from
     Combustion Sources.  J. Environ. Sci. Health, A12(10):549,
     1977.
                                61

-------
A Specific Method for the Determination of
Sulfuric Acid Emissions from Combustion
Sources
Paul Urone
University of Florida

Robert A. Lucas
Employers Insurance of Wausau



     ABSTRACT

     The sampling and analysis of sulfur trioxide and  sulfuric
     acid  in emission sources continue to be a source  of dif-
     ficulty.  Various standardized and proposed alternate
     testing methods are under study, ranging from simple acid-
     base  and colorimetric techniques to advanced instrumented
     methods.  Most of the difficulties in sampling and ana-
     lysis are related to the Jiigh concurrent concentrations
     of sulfur dioxide, other acids, and particulate sulfates.

     A  colorimetric reagent was investigated for the specific
     determination of sulfuric acid emissions.  The method
     utilizes a  sulfone dye precursor, which reacts with the
     sulfuric acid or sulfur trioxide to form the sulfone dye.
     At a  given  pH, the color of the dye is specific for the
     sulfonation reaction and independent of other acids or
     sulfate compounds present.  Three dye precursors, phenolph-
     thalein, thymolphthalein, and acid orange A have  been
     studied; each reacts specifically with sulfuric acid
     aerosols and  strong  (X5M) sulfuric acid solutions.
     Filters impregnated with the dye precursors develop the
     specific colors, but the colors fade with time.  Addition-
     al studies  are being undertaken to prevent reversal of
     the reaction  and to determine its sensitivity limits.
                               63

-------
INTRODUCTION

     The sampling and analysis of sulfur trioxide and sulfuric acid
in emission sources continue to be a source of difficulty.  At
present, a number of methods are used to determine sulfuric acid
mist in either emission sources or ambient air.  These methods can
generally be divided into two basic operations:  (1) the sampling or
collection of the aerosol; and (2) the analysis or measurement of the
collected aerosol (1).

     The major methods of collection of sulfuric acid aerosols are
absorption into isopropyl alcohol or alkaline solutions, controlled
and uncontrolled condensation, impaction, and filtration.

     The major methods of measurement and their chemical basis for
the analysis of H2SO4 at present are:

     a)   Sodium Hydroxide Titration (1).  This is a simple and
          direct method; however, it is not specific and de-
          pends upon the assumption that the only acid present
          is H2S04 and that no basic substances are present to
          interfere.

     b)   Barium Ion Titration (2).  This is the method specified
          by EPA's Method 8.  The reaction is specific for any
          soluble sulfate.  Thorin is used as the indicator.

     c)   Chloranilate Method (1).  This method depends on the
          release of the colored chloranilate ion from the in-
          soluble barium chloranilate salt.  Sulfate reacts
          with the barium chloranilate to form insoluble barium
          sulfate and release the chloranilate ion to color the
          solution in a quantitative manner.

     d)   Turbidimetric Method (1).  In this method barium ion
          is used to precipitate the sulfate as a uniformly dis-
          tributed suspension.  The turbidity of the solution
          is measured spectrophotometrically.

     e)   Flame Photoluminescent Detector (1).  This method relies
          on the principle that all sulfur compounds fluoresce
          in hydrogen-rich flames at a wavelength that is re-
          latively specific for sulfur.  The fluorescent light
          is measured by means of a photomultiplier tube.

     f)   Vanadium Method (3).  The sulfuric acid is determined by
          ammonium vanadate aqueous solution which forms a yellow
                                64

-------
          color.   The intensity of this color is a linear func-
          tion of sulfuric acid concentration.

     g)    Bromphenol Blue Method (4).   This is a simple method and
          depends on a color change due to sulfuric acid in an
          aqueous solution of isopropanol and bromphenol blue.
          It is a measure of total acidity and assumes no basic
          substances are present to interfere.

     h)    Coulometric Method (5).   This is a simple method and non-
          specific and provides a measure of the effective hydro-
          gen ion concentration only.

     The major difficulties encountered by the presently available
methods are due generally to the sampling methods with their inherent
interferences from co-pollutants and sampling efficiency.  The various
analytical methods have been shown to be reliable enough, considering
the limitations of the sampling methods.


Formation of H2S04 Aerosol

     During combustion of a fossil fuel, a small fraction (l%-5%) of
the sulfur is oxidized to S03.  This can result in a yield of 5 ppm
to 50 ppm in the flue gas as S03, depending upon such factors as
sulfur content, excess air, and combustion temperature (6).  Since
most of the sulfur in power plant flue gases appears as S02, reactions
may occur on filter and collection surfaces as well as combustion aero-
sols.  Aerosol formation may also be initiated by homogeneous gas
phase reactions, with subsequent clustering reactions leading to the
formation of new particles  (7).

     New sulfuric acid particle formation may be viewed as being a
three-step process (8)(9):

     1.   Oxidation of S02:
          S02(g) + 1/2 02(g) c--t S03(g)
          The gas phase oxidation of S02 to S03 is assumed to be
          the rate determining step in the production of H2S04 .

     2.   Reaction of S03 with water to form H2SO4:
          S03(g) + H20(g) + M--H2S04(g)
          At temperatures below 400F, essentially all the S03
          present is converted to H2S04 at equilibrium.  In con-
          trast to the formation of SOa, the formation of H2S04
          occurs rapidly in the thermodynamically feasible tem-
          perature range.
                                65

-------
     3.   Clustering of H2S04 and H20 molecules to form prenuc-
          leation embryos, followed by the heteromolecular nucleation
          process:
          H2S04 + X(H2O) -*- (H2S04)(H20)X (cluster)

          (H2SO4 )(H2 O)x + H2O particles -*- larger particle

     There are several conditions that regulate the two possible
reaction channels for the conversion of SO3 to H2SO4.  One involves
a direct gas phase reaction with H2O and the other involves a sur-
face reaction as S03 is scavenged by a pre-existing aerosol particle.
In general, in an environment with higher relative humidity, less
total surface area of suspended particles, lower temperature, and
a higher production rate of S02 molecules, more secondary new sul-
furic acid aerosols will be generated.

Sulfonation

     A specific analytical technique for the analysis of sulfuric
acid is possible due to the sulfonating ability of sulfuric acid
with aromatic and heterocyclic compounds.  By definition sulfonation
is the process of replacement of a hydrogen atom in an organic com-
pound with the sulfonic group (S03H+).  The replaced hydrogen atom
and the hydroxyl ion (OH~) from sulfuric acid form a water molecule
(10), (Figure 1).

     There is no clear cut guideline to follow when considering re-
action temperatures, concentrations, percent product recovery, and
reaction rates when considering sulfonation of species not generally
dealt with in industrial applications.  However, sulfonation of
rings which are substituted with electron withdrawing groups (-CL ,
-Br~, -S03H, -CO2H, -NO2) proceeds with difficulty since the ring
is_already electron poor.  Electron donating groups (-NH2, -OH, -OR,
_O~) attached to the ring aid sulfonation and are ortho and para di-
recting.

     Kinetic and mechanistic studies of the sulfonation of aromatic
compounds have led to the conclusion that it is an S2  reaction with
monomeric SQ3 as the effective reacting species, not only with S03
itself, but also when sulfuric acid or oleum is used.  It has been pro-
posed that in sulfuric acid an acid solvate of S03 is the active
species.  This is considered possible, but less likely than free S03
(10).

     Aromatic compounds can be sulfonated with concentrated H2S04 ,
but as the concentration of the water increases during reaction,
the rate of sulfonation steadily decreases, the reaction rate being
                                66

-------
       SULFONATION REACTION
  2H2S04*=* H30+ + HS04 + S03(Equilibrium)
2. o  + S03  >    m          (Slow)
          *^
               _

3.  (*)    +HS04    (o)  +H2S04(Fost)
     S03             S03H


4.  (o]  + H,0+ *=* fo] + hUO (Equilibrium)
   \x"^             \X^   ^-




     DESULFONATION REACTION
     SO^H
Figure 1.  Sulfonation reactions.
                    67

-------
inversely proportional to the square of the water concentration.  The
reaction ceases when the acid concentration reaches a level charac-
teristic of each compound.  Sulfonation differs from most other types
of electrophillic substitution in that the process is readily re-
versible under mild conditions (11).  Desulfonation of some aroma-
tic compounds proceeds rapidly and in good yield by simple dilution
in aqueous medium.  Therefore, to carry sulfonation reaction to
completion, the removal of water formed from the reaction is es-
sential.  However, the hydrolysis of sulfonated aromatics will not
occur if the acid strength is kept high.

     Filter Material and PretreatmentThe type of filter used to
collect the sulfuric acid aerosol was the Gelman Type A/E glass fi-
ber filter.  Glass fiber filters combine the desirable features of
high retention efficiency with high flow rates.  They have been
shown to have the tear resistance and wet strength necessary to enable
them to withstand being submersed into the precursor solution fol-
lowed by drying and filtration (12).

     These  filters are devoid of any organic binders and have a
minimum retention efficiency of 99.7% for 0.3 /urn particles as measured
by the dioctyl phthalate penetration test.  Various researchers have
found that  glass fiber filters can react with atmospheric sulfur di-
oxide and sulfuric acid catalytically on the glass surface and thus
seriously interfere with the determination of sulfuric acid aerosols
(13^15) .  Barton and McAdie developed a method of treatment of the
glass fiber filters by soaking in 20% sulfuric acid followed by
gentle boiling for 10 minutes, rinsing with distilled water, 80% iso-
propanol, and finally acetone.  The filters are then dried in a desic-
cator (16).  This method of filter treatment was selected for this
study.

     Materials and MethodsAcid orange A, phenolphthalein, and
thymolphthalein (Figure 2) were selected as the indicator precur-
sors.  To prepare a 0.4% solution of both compounds, 0.4 g of each
was dissolved in a slightly alkaline solution of sodium hydroxide
in a 100 ml volumetric flask.  Because both compounds are acid base
indicators, a color change occurred in the basic solution.  To bring
the pH to a neutral range, concentrated HC1 was applied drop-wise
until a clear solution resulted.  The solution was then diluted
with distilled water to the desired final volume.

     The filters were treated by submersing them into the precursor
indicator solutions and allowing them to soak for five minutes.  The
filters were then removed and allowed to dry in a desiccator.
                               68

-------
              ACID  ORANGE  A
           '3N
        H0S- N = N
               PHENOLPHTHALEiN

                      OH
                      ^^.
                      o
               THYMOLPHTHALEIN
                                  '10
                     c-
                     6
Figure 2.  Dye substrates selected for sulfonation reactions.
                        69

-------
     To test the precursor indicators for color change,  two
methods were used to apply the sulfuric acid.  The  first method
was simply  the addition of various concentrations of H2S04 direct-
ly onto the treated filters with an eye dropper.  The second me-
thod consisted of a compressed air supply filtered  through a sili-
ca gel bed  followed by flow measurement in a rotameter.  After
the rotameter the air was used to generate a sulfuric acid aerosol
by the use  of a DeVilbiss model atomizer.  The aerosol generated
was then equilibrated in a 4 f glass chamber.  The  air flow was
separated at the top of the glass chamber with excess air going to
a hood for  removal and the remaining fraction going to a filter hol-
der followed by a critical orifice, silica gel trap, and air pump
(Figure 3).  The air flow was held at 3.8 P/minute  through the
rotameter and 0.9 f/minute through the filter holder.

     An absorbance curve was run on a Bausch and Lomb Spectronic
20 for the  reaction products from the addition of 20 ml  of concen-
trated H2SO4 to 20 ml of phenolphthalein and thymolphthalein solutions
in an ice bath.  The absorbance curves obtained are shown in Figures
4 and 5.

     The next procedure was to see if the sulfonated products could
be easily removed from the filter material.  The various extracting
solutions used were:  distilled water, 6 N hydrochloric  acid, methanol,
20% nitric  acid, triethylene glycol, benzene, and hexane.

     Based  on previous work of Joseph Griffiths (17),. acid orange
A was selected as a possible precursor dye.  Published results (18)
indicate that this dye in a solution of concentrated H2SO4 would
sulfonate easily and result in a very noticeable color change.  Acid
orange A was prepared in the lab by relatively simple chemical pro-
cedures.  Its structure, along with the structure of phenolphtha-
lein and thymolphthalein, appears in Figure 2.

     A 0.15% solution of acid orange A was prepared by dilution of
0.15 g acid orange A in 50 ml of distilled water and 50  ml of me-
thanol.   The same procedure of filter treatment and testing was used
as for the  indicator precursors.

     Three absorbance curves were run on the acid orange A:  one on
the solution itself without any reaction, the second on  the reaction
product of 20 ml of concentrated H2S04  added slowly to 20 ml of acid
orange A in an ice bath, and the third on the solution produced
from rinsing a filter that collected sulfuric acid  aerosol from the
atomizer,  with 6 N hydrochloric acid.  The results  of these data
can be seen graphically in Figure 6.
                                70

-------
          AEROSOL  GENERATION and  COLLECTION
Air
Pump
J L_

~i r~

Silica
Gel




LJ Critical
Orifice
Filter
Holder
r i

L J

Excess
to
Hood
J


r
/ 4 Liter Pyre-
Bottle
Roto
"X_
^r^
TK
S'
jter G

~l
lica
Jel (
U
A
1 	 '
3ompres
ir Supp

sed
y
                                          Atomizer
Figure 3.  Apparatus for the generation and collection of sulfuric acid aerosol.

-------
   400
                    Phenolsulfonphthoiein
500      600      700
  Wavelength (nm)
Figure 4.  Phenolsulfonphthalein absorbance curve.
                    72

-------
.8

 7


 .6

 .5
8
I  -4
_t- T
<  .3


    .2
                        Thy molsu I phonphtha lei n
                                J	I
    400       500       600       700
                 Wavelength (nm)


 Figure 5.  Thymolsulfonphthalein absorbance curve.
                     73

-------
   -6r
   .5
    4
(U

*  .3
o
in
-Q
    .2
    400
       A  Acid  Orange A
       9  Acid  Orange A reaction
          in solution
          Acid  Orange A reaction
          on the filter
500      600       700
 Wavelength  (nm)
     Figure 6.  Acid orange A absorbance curves,
                       74

-------
RESULTS AND DISCUSSION

     Phenolphthalein and thymolphthalein reacted with 5 M H2S04 and
stronger solutions to produce noticeable color developments on the
white filters.  The phenolphthalein filters turned from yellow to
orange as the concentration applied increased.  Thymolphthalein
filters turned various shades of pink as the concentration was
varied.  The products from both indicators reacted in solution
showed strong absorbance peaks, phenolphthalein at 500 nm and
thymolphthalein at 550 nm.

     The removal of the colored products by the various extracting
solutions resulted in a reversal of the sulfonation reaction, and a
clear solution was produced in all cases.  This reversal could be
expected in distilled water due to sulfonation reactions being
reversible in excess water; however, the several alcohols and
organic solutions were not expected to give such results.  After
observing this fact, filters treated with sulfuric acid were al-
lowed to stand exposed to room air for several minutes after treat-
ment.  The color on the filters gradually faded until after about
one hour no color development remained.  This implies that the
reaction is not only readily reversible in aqueous solution but
also when exposed to room air.  Heating at 100, 150, and 200C
failed to fix the product on the treated filters.  In fact, instead
of fixing the product, it aided the reversal  rate.

     Acid orange A changed in color from a light orange to magenta
red for sulfuric acid concentrations greater  than 5 M.  The re-
action on the filter from the application of  6 M H2S04 from an eye
dropper was observed in one hour at room temperature.

     The extraction of the colored product from the filter material
resulted in a reversal of the sulfonation reaction of acid orange A
in all cases except one.  With 6 N (normal) hydrochloric acid as
the extracting solution, the red color was retained, whereas all
the other extracting solutions turned the product back into an
orange solution.  This indicates that hydrolysis of the sulfonated
acid orange A did not occur in the presence of low acidity.

     The absorbance curves in Figure 6 show the various absorbance
peaks for acid orange A before and after sulfonation.  Acid orange
A absorbs strongly at 490 nm compared to acid orange A reacted
with H2S04  in solution absorbing strongly at  515 nm.  When acid
orange A was reacted with H2S04 on the filter and then rinsed off
with 6 N hydrochloric acid, the absorption peak occurred at 500 nm.
                                75

-------
The absorbance of the solution extracted from the filter shows an
intermediate absorbance peak due to the partial reacted and unreac-
ted filter areas.

     Filters were then placed in the aerosol generating system as
shown in Figure 3.  Sulfuric acid was collected until most of the
filter surface became reacted.  After removing the filter, the
color developed faded slowly back to the original color of the
precursor dye.  This was unexpected, due to the fact that the
coated filter which had reacted with H2S04 from an eye dropper had
remained relatively stable for one hour or more.  The aerosol gene-
rated from the atomizer reacted only on the surface of the filter
and showed no signs of penetration due to the original orange color
appearing on the backside of the filter immediately after removal.
The filter in which the sulfuric acid was applied with an eye drop-
per soaked clear through the filter.  This soaking prevented the  :
immediate reversal by the room air.  Heating at various tempera-
tures again did not result in dye fixation.

     At present, measures are being taken to prevent reversal of
the sulfonation products from these three dye and indicator precur-
sors.  Sulfonated dyes are used in industry and have been shown to
be stable under mild conditions (19).  Whether or not any of these
dyes can be sulfonated with H2S04 under experimental conditions in
air sampling is still a question.  If the determination of the sul-
furic acid aerosols is to be accomplished by spectrophotometric
analysis of sulfonated dyes or indicators, they must be stable and
react in a quantitative manner.

ACKNOWLEDGMENT

     This work was supported by grants from the Environmental Pro-
tection Agency and the Smelter Environmental Research Association.
                                76

-------
REFERENCES


1.   Urone,  P.  Source and Ambient Air Analysis of S03  and H2S04:
     State of the Art Report.  Health Lab.  Sci., 11:246,  July 1974.

2.   Barton, S. C.,  and H. G. McAdie.  An Automated Instrument
     for the Specific Determination of Ambient H2S04  Aerosol.
     Presented before the Division of Water,  Air and  Waste
     Chemistry, ACS, New York, 1972.

3.   Baviha, C. J.,  and L. S. Shinkararenka.   Determination of
     Sulfuric Acid in Air by a Vanadate Method.  Neftepererab
     Neftehhim, 9:40-41, 1971.

4.   Barrett, W. J., H. C. Miller, J. E. Smith, and C.  H. Guin.
     Development of a Portable Device to Collect Sulfuric Acid
     Aerosol.  Interim Report from EPA, February 1977.

5.   Scaringelli, F. P., and K. A. Rehme.  Determination of Atmos-
     pheric Concentrations of Sulfuric Acid Aerosol by Spectropho-
     tometry, Coulometry and Flame Photometry. Anal.  Chem., 41:707-
     713, June 1969.

6.   Crumley, P. H., and A.  W. Fletcher.  Fuel Combustion Charac-
     teristics.   Inst.  Fuel, 29:322, 1956.

7.   Robinson, E.,  and  R. C. Robbins.  Sources Abundance and Fate of
     Gaseous  Atmospheric Pollutants.  Stanford Research  Institute
     Project  PR  6755, February 1968.

8.   Kiang, C. S.,  and  D. Stauffer.  Chemical Nucleation Theory for
     Various  Humidities and  Pollutants.  Faraday  Sym., 7:26,
     University  Press,  Aberdeen, N.D., 1973.

9.   Middleton,  P., and C. S. Kiang.  A Kinetic Aerosol Model for
     the  Formation  of New Sulfuric  Acid Particles. Submitted for
     publication, 1976.
                                77

-------
10.  Gilbert,  E.  E.   Sulfonation  and  Related  Reactions.   Inter-
     science Publishers,  New York,  1965.

11.  Butler, G.  B.,  and K.  D. Berlin.  Fundamentals  of Organic
     Chemistry Theory and Application.   Ronald Press Company, New
     York, 1972.

12.  Gelman Instrument Company Catalog,  1975.

13.  Scaringelli, F. P.,  and K. A.  Rehme.   Anal.  Chem.,  41:707,
     June 1969.

14.  Byers, R. L.,  and J. W. Davis.  Sulfur Dioxide  Adsorption  and
     Desorption on Various Filter Media.   Air Poll.  Control  Assoc.,
     20:236.

15.  Pate, J. B., B. E. Ammons, G.  A. Swanson, and J. P. Lodge,
     Jr.  Nitrite Interference in Spectrophotometric Determination
     of Atmospheric Sulfur Dioxide. Anal.  Chem.,  37:942, 1965.

16.  Barton, S.  C., and H. G. McAdie.  Preparation of Glass
     Fiber Filters for Sulfuric Acid Aerosol Collection.  Env.
     Sci. Tech., 4:769-770, September 1970.

17.  Griffiths,  J. F.  Colorimetric Determination of Sulfuric
     Acid Aerosol.  Paper presented to the Graduate  Council  at
     the University of Florida, October 1975.

18.  Green, A. G.  The Analysis of Dyestuffs.  3rd ed.,  Griffin and
     Company, London, 1920.

19.  Lubs, H. A., ed.  The Chemistry of Synthetic Dyes and Pigments,
     Reinhold Publishers, New York, 1955.
                                78

-------
Measurements of Sulfuric Acid Vapor by
Infrared Spectroscopy
Roosevelt Rollins
U.S. Environmental Protection Agency
      ABSTRACT

      Recent  investigations were directed toward the development
      of  an  in  situ method of measuring gaseous sulfuric acid
      emissions in flue gases.  This presentation summarizes the
      results of two  EPA-sponsored studies conducted to deter-
      mine the  feasibility of measuring sulfuric acid vapor us-
      ing infrared spectrophotometric techniques.  These studies
      primarily involved  the generation and analysis of absorp-
      tion spectral data  in the 7-12 microns region.

      Two regions, 8.2 microns  (wave number 1220) and 11.4 mi-
      crons  (wave number  880), were identified as being most
      promising for spectroscopic monitoring purposes.  Spec-
      tral measurements did not reveal any fine absorption line
      structure for the sulfuric acid vapor at atmospheric pres-
      sure conditions although  it was observed that sulfuric  acid
      vapor  absorbs rather strongly in both regions.  Absorption
      co-efficients of 6.4/atm-cro and 6.8/atm-cm were obtained
      at  wave numbers 1222 and  880, respectively.  The maximum
      percentage change in absorption strength occurred on the
      side of the peak at wave  number 1222.

      As  a result of  these spectral studies, it  is believed that
      gaseous sulfuric acid emissions can  be monitored spectro-
      scopically.  A  high-resolution system would most likely be
      required  because of the  lack of fine absorption line struc-
      ture for  gaseous sulfuric acid.  For a tunable diode laser
      system, it is  estimated  that a 2 ppm-m sensitivity would
      be  possible.
                                79

-------
INTRODUCTION

     Recent studies (1) have shown the need for a dependable and-
reliable method for measuring the concentration of sulfuric acid
vapor in flue gases.  All of the present analytical methods require
some type of sample collection.  As a result, measurement errors are
likely to occur because of the probable change in the chemical or
physical composition of the gas sample during the sampling.  The
problem of measuring sulfuric acid emissions is further complicated
by the highly reactive properties of this vapor and the many sub-
stances in the effluent with which it may react.

     Because of the inherent problems of maintaining sample integ-
rity in extractive methods, recent activities have been directed
toward the development of a non-extractive, instrumental method for
measuring sulfuric acid vapor emissions.  An in situ method, employ-
ing optical techniques, would not disturb the sample and, thus, would
provide an indication of the true characteristic of the gas stream.

     Infrared absorption spectroscopy is an effective optical tech-
nique for detecting a variety of gases.  Several investigators (2)(3)
(4) have published low-resolution, qualitative spectra which indicate
that sulfuric acid vapor absorbs infrared radiation in bands centered
at 2.8, 6.9, 8.2, and 11.3 microns.  Since water vapor is known
also to absorb very strongly in the first two bands, only the last
two bands offer promises for stack monitoring purposes.

     This paper summarizes the efforts and experimental results of
two EPA-sponsored studies involving spectral measurements on labora-
tory-generated sulfuric acid vapor.  These studies were conducted
for the purpose of determining the feasibility of measuring sul-
furic acid vapor in flue gases by infrared spectroscopy.  The
initial investigation was performed under EPA Contract 68-02-1774
by Burch and co-workers at Aeronutronic Ford Corporation.  Under
EPA Contract 68-02-2482, further studies were conducted by R. Eng at
Laser Analytics, Inc.


DISCUSSION OF RESULTS

     The spectral measurements made at Aeronutronic Ford (5) resulted
in low-resolution spectra for H2SO4 vapor absorption over the 8-12
microns region.  The measurements were made using a grating spectro-
meter with a spectral resolution that varied from approximately 2
cm~1 at 12 microns to 4 cm"1  at 8 microns.  Figure 1 shows spectral
curves of transmission obtained for three samples of hot H2SO4 vapor.
                                80

-------
                           1200
WAVENUMBER, cm

1000
                                                                  1
00
                                                      WAVELENGTH. ,um

             Figure 1.  Spectral  curves  of  transmittance for three samples  of H2S04.
                        Path  length  =  52 cm.   Vapor temperature ^235C.  The spectral
                        slitwidth varies from approximately 4 cm   at 8 /xm  to 2  cm  at 12

-------
The partial pressures given for H2SO4 are estimates based on ther-
modynamics tables.  Two regions, 8.24 and 11.4 microns, were iden-
tified as having absorption .strengths adequate for monitoring pur-
poses.  At wavelengths of maximum absorption, the absorption
coefficient is approximately 4.0 atm 1 cm 1 .

     Although the absorption by H2SO4 vapor is relatively strong,
the spectral curves do not show any line absorption features.  Alpert
(6) has published high-resolution spectral data which indicate many
individual H2S04 absorption lines in the infrared region from 1312
cm"1 to 1549 cm~1.  Based on his data, it was concluded that there
might be similar individual absorption lines in the 8.2 fi and 11.3/u
bands.  To answer this question, high resolution (<0.1 cm  ) measure-
ments became necessary.

     The high-resolution measurements were performed by Laser
Analytics (7) utilizing a tunable diode laser spectrometer.  Spec-
tral scans were made for the central portions of the H2S04 absorp-
tion bands at 8.2 n (1222 cm"1) and 11.3^(880 cm"1).  Measurements
were made for several H2SO4 vapor samples at both low pressure
(< 1 torr) and atmospheric pressure conditions.  Figure 2 shows
typical recorder tracings for a low pressure H2SO4 absorption near
1223 cm"1.  The traces show some fine structure as well as smooth
absorption.  Except in this spectral region, there was no fine struc-
ture observed anywhere else in the 1222 cm"1 and 880 cm 1  bands.
As expected, the smooth part of the total absorption shows a large
variation with H2S04 partial pressure. _Figure 3 is a summary of the
low pressure scans for the 1210-1240 cm 1 region.  This plot was
obtained by combining a large number of scans like those_shown in
Figure 2.  Similar measurements were made for the 880 cm   region
and the results are shown in Figures 4 and 5.  For the H2SO4 vapor
at low pressure conditions, absorption coefficients are 7.2 and 7.0
atm"1 cm"1 at 1222 cm"1and 880 cm ', respectively.

     Using a flowing-gas cell with dry nitrogen as the carrier gas,
spectral measurements were made for H2S04 vapor at atmospheric pres-
sure.  After identifying prominent absorption locations, scans at
atmospheric pressure were performed for 2 cm   intervals in both
regions.  Figure 6 shows a summary of the atmospheric spectral data
and the  location of the 2 cm"1  interval for the 1222 cm"  region.
Similarly, Figure  7 shows a summary for scans in the 880 cm   region.
For H2SO4 vapor at atmospheric  pressure conditions, absorption^
coefficients of 6.5 and 6.9 atm~1 cm"1 were obtained at 1222 cm   and
880 cm"1, respectively.

     To  determine  the severity  of possible interferences, high
resolution measurements were made for the interferants H20, SO2,
                                82

-------
CO
co
                                                    (FREQUENCY CALIBRATION)
                                                        FREQUENCY, cm
                                                                    1
                                     Figure 2.   H2S04 absorption  near 1223  cm
                                                                                 * 4

-------
o

"-
u.
Ul
o
u
z
o
EC
o
CA
                                               RESERVOIR TEMPERATURE

                                               CELL IODY TEMPERATURE

                                                      0-" Ton
                 170C

                 200C
            1210
1230
                                 1220


                                  FREQUENCY, em'1


.Figure 3.   Low pressure H2S04 absorption (P ~ 0.67 Torr,  T = 170C),
1240
                                       84

-------
        CJ
        z
        
-------
                                               RESERVOIR TEMPERATURE
                                               CELL BODY TEMPERATURE
                                               PH 2804* 0-67 Torr
                                                              170C
                                                              20CC
Figure 5
                            FREQUENCY, cm''
.   Low pressure H2S04absorption (P ~ 0.67 Torr, T
170C),

-------
   2.4
                                  ATMOSPHERIC PRESSURE

                                       L = 64 cm
                                                                       200C
->  1.6
o
V)
CO
   0.8
                                                                       180C 	
          1220
                                            1221
   0.4
                                                                              1222


                                                                              )
   0.3
o
z

00
   0.2
   0.1
                                          LOW PRESSURE


                                           L =  50 cm


                                             = 170C
    1210
Figure 6
                             1220
                                                     1230
                                                                              1240
                       WAVENUMBER, cm !


Location  of the 2 cm    atmospheric scan  region with

respect to the  1210-1240 cm    low  pressure
                                       87

-------
      2.4
      1.6
   u
   z

   CD
   K
   O
   M
   tO
      0.8
                                    ATMOSPHERIC PRESSURE

                                          L = 64cm
                                                                      200C
                 879
                                                 880
                                                                                881
     0.3
   02
   en
   o
   CO
   CM
     0.1
                                   r
LOW PRESSURE


  L= 50 em

  T = 170C
       870
                               880
                                                        890
                                                                                900
                                     WAVENUMBER.cm1
                                   -1
Figure  7.  Location of  the  2  cm  ' atmospheric scan region  with

            respect to  the 870-895 cm   low  pressure scan.
                                         88

-------
and CC>2 .  These absorption measurements were made over the same 2
cm"1  intervals as those used for the H2S04 vapor.  Figures 8 and 9
show the S02 and H20 vapor absorption lines between 1219 and 1222
cm  .  Figures 10 and 11 show the absorption lines for C02 and H2O
from 878 to 881 cm~1.  In the 1222 cm"1  region, the S02 exhibits a
very high line density.  The H20 absorption is much stronger at 1222
cm~1  than at 880 cm"1.  In the 880 cm~1  region, C02 absorption was no1
detectable for a 100 torr sample in a 1.1 meter path.

     A spectroscopic technique was also employed to determine the
partial pressures of H2S04, S03, and H20 above hot azeotropic H2SC>4
solutions.  This was done by measuring the partial pressures of
H20 and 863 directly using known and calibrated absorption strengths.
The total vapor pressure was then measured using a "U" tube mano-
meter.  The partial pressure of H2S04 vapor was calculated as the
difference between the total pressure and the sum of the S03 and H20
partial pressures.  Table 1 shows the results of the partial pressure
measurements.  The numbers in the parentheses are from Luchinski's
published data (8) for the atmospheric azeotrope (98.3% H2S04).
The H2S04 partial pressures determined in this program are about 30%
smaller than those reported in the literature.  Therefore, as a
result, the absorption coefficient for H2SO4 vapor is correspondingly
higher.  The effect of H2S04 dissociation could account for the
discrepancy.


            Table 1.  Partial Pressure vs. Temperature
                          of H2 S04 Azeotrope


       Temp.      	Partial Pressure (Torr)	
       (C)             H20            S03           H2S04


       107        0.024 + 0.002   0.022 + 0.002   0.034 + 0.02
                       (0.01)         (0.005)        (0.04)

       150        0.023 + 0.07    0.21 + 0.02     0.32 + 0.08
                       (0.145)         (0.086)        (0.45)

       200        2.3 + 0.2       2.0 +; 0.2       3.5  0.8
                       (1.7)           (1.3)          (4.4)
                                89

-------
                                                                             L-l.lm

                                                                             T = 200C

                                                                             PS02 "
X
in
oe
o
CM
O
t/t
    0
    1219
1220
                                             WAVENUMBER.cm
1221
1
1222
                                    Figure  8.   S02 absorption.

-------
   1.0
   0.8
J  0.6
to
ec
o
cfl
00

-------
   0.12
   0.08
cc
o
   0.04
O
CJ
     0
     878
                                                                   L = t.1m

                                                                   T = 200C
WEAK C02 LINES (NOT OBSERVED)
879
880
881
                                       WAVENUMBER. cm ^


                                Figure  10.   C02  absorption.
                                           92

-------
   0.20
  0.16
 S  0.12
00
OC
o
c/j
ca


o  0.08
CM
   0.04
     0
     878
                             H20 LINE
                                     L = 1.1m

                                     T = 200C

                                          = IBTorr
  879                         880


         WAVENUMBER.citr1



Figure  11.  H20 absorption.
881
                                            93

-------
CONCLUSIONS

     Laboratory studies have shown that  sulfuric acid  vapor absorbs
quite strongly, but rather broadly, at 880 cm"1 and  1222 cm"1.
The maximum percentage change in absorption strength occurs on  the
high frequency side of the peak at 1222  cm"1.  The absorption in
the 880 cm ' region exhibits a smaller change but is less  subject
to interference.  For H2SO4 vapor at atmospheric conditions,
absorption coefficients of 6.5 and 6.9 a tin"1 cm"1  have  been measured
for the 1222 cm"1  and 8880 cm"1 bands, respectively.

     Even though H2SO4 vapor at atmospheric pressure apparently has
no line absorption structure, it appears that H2S04  can be monitored
spectroscopically by utilizing the broad absorption  features.   The
best choice of spectral regions will depend on possible interfering
gases and the absorption features for H2SO4 vapor.   It seems likely
that a high-resolution system will be required for the monitoring
task.  Several measurement approaches seem feasible, including
differential absorption spectroscopy using diode lasers, derivative
spectroscopy, and Fourier transform interferometry.  For an H2SO4
monitor utilizing diode lasers, a sensitivity of 2 ppm-meters would
probably be possible.
                               94

-------
REFERENCES
1.   Cheney, J. L., J. B. Homolya,  and H. M. Barnes.  Measurement
     and Identification of Primary Sulfates Emitted from Combustion
     Sources.  70th Annual AICHE, New York, November 1977.

2.   Stopperka, K.  The Infrared Spectrum of the Water-free H2S04
     and the Composition of the H20-H2S04 System.  Zeit fur Anorgan
     und Alleg. Chem., 344:263-278, 1963.

3.   Chackalackal, S. M., and F. E. Stafford.  Infrared Spectra of
     the Vapors Above Sulfuric and Deuteriosulfuric Acids.  Journ.
     Amer. Chem. Soc., 88:4, 723-728, February 20, 1966.

4.   Stopperka, K., and F. Kilz.  Die Zusammersetzung der Gasphase
     uber dem flussigen System H2S04 in Abhangigkeit von der Tempera-
     ture.  Z. Anorg. u Allgem. Chem., Band 270, 49-66, 1969.

5.   Burch, D. E., D. E. Gates, and N. Potter.  Infrared Absorption
     by Sulfuric Acid Vapors.  Final Report No. EPA-600/2-76-191,
     July 1976.

6.   Alpert, B. D.  Studies in High Resolution Infrared Spectroscopy.
     Ph.D. Dissertation (Appendix B contains H2S04 data), Ohio State
     University, Columbus, Ohio, 1970.

7.   Eng, R. E., K. W. Nill, and J. R. Butler.  Spectral Measurements
     of Gaseous Sulfuric Acid Using Tunable Diode Lasers.  Final
     Report No. EPA-600/2-78-019, February 1978.

S.   Luchinski, G. P. Zhur, F12 Khem, 30:1207, 1956.
                               95

-------
Chemical Speciation and Concentration
Monitoring of Sulfur Oxides by Laser-Raman
Scattering
Richard K. Chang
Robert E. Benner
Yale University
     ABSTRACT

     By  the laser-Raman scattering  technique, we have
     measured directly, without filter  collection, the con-
     centration of sulfate ions (40 g/m3  level) in
     laboratory-generated aerosol particles of less than
     2 /urn  in diameter.  Our new tabletop design, consisting
     of  an argon laser, multiple-pass cell, concave grating
     spectrograph, and TV camera detector, should give in-
     creased sensitivity.  Details  of such a  laser-Raman
     monitoring system will be discussed,  along with the
     possibility of using Raman shifts  and Raman line widths
     to  distinguish and to monitor  various phases (gas,
     liquid, or solid) and types of sulfur oxides (H2S04,
     SO3,  SO2, HS04> S04 , and metal sulfates) in the presence of
     other atmospheric molecules.  In addition, recent results
     on  the angular distribution of inelastic radiation
     (fluorescence and Raman) from  small monodispersed
     spherical particles will be compared  with elastic
     Lorenz-Mie scattering and fluorescence  from a continuum
     medium.  These data, which are dependent on the
     particle-air index of refraction  interface, are per-
     tinent to the conversion of detected  photon signals
     to  molecular concentrations contained within aerosols.


 INTRODUCTION

     Since the vibrational-rotational  energies are characteristically
 distinct for different types of molecules, chemical speciation
 of a multicomponent mixture can be achieved  using Raman light
                                97

-------
scattering techniques.  In the Raman method, photons with energy
hUj  are incident on the mixture, and a very small fraction of the
light is inelastically scattered, resulting in Raman photons with
energy hus.  The photon energy difference (huj-hus) is precisely
equal to the vibrational-rotational energy of the scattering
molecule.  Raman scattering is describable by a second-order
perturbation process in which the absorption of an incident
photon causes an electron to make a virtual transition to an
intermediate state.  The final state occurs with the emission of
a Raman photon, and the electron is left in the excited vibra-
tional-rotational state [Figure l(a)].  Consequently, by using
monochromatic incident radiation from a laser and by spectrally
analyzing the Raman scattered radiation with a spectrograph or
a spectrometer, the chemical species present in the scattering
volume can be determined from the known vibrational-rotational
energies of different molecules.  In addition, since the inten-
sity of the radiation scattered by a molecular species is
linearly proportional to its concentration, the amount of a
particular constituent can be determined from measured Raman
scattering cross sections.

     An alternative means of probing molecular energy levels
utilizes infrared absorption of photons with photon energy hu,R
which involves a real transition of an electron from its ground
state to the excited vibrational-rotational state [Figure l(b)].
For molecules that do not possess a center of inversion, the
vibrational-rotational state probed by infrared absorption is
identical to that reached by the Raman scattering process.  Both
the infrared absorption and the Raman scattering processes
distinguish different types of molecules by the uniqueness of
their vibrational-rotational energies.  In general, infrared
spectroscopy does not require a laser source, but the use of
lasers in Raman spectroscopy greatly enhances signal/noise ratios,
The vibrational state energies are influenced significantly by
the phase (gas, liquid, or solid) of the sample.  Furthermore,
additional vibrational peaks become observable when the molecular
species changes from the gaseous to the liquid phase and from
the liquid to the solid phase.

     Both the infrared and the Raman processes are affected by
the finite lifetimes of the excited vibrational-rotational
states which result in a broadening of the infrared absorption
and Raman scattering peaks.  In the gaseous phase, the major
contributions to a broadened linewidth are collisions among the
molecules (=s.01 cm"1 at 1 atm) and the Doppler effect (=*.005 cm"1).
For the liquid bhase, the typical collisioned broadened line-
width (1-10 cm"1 ) can be further broadened (5-50 cm 1 ) by proton
                                98

-------
                                    m
             t
       hv.
hv.
                   V
                                     f
                                I
                                                          hv,
                                                            IR
              (a)
                              (b)
Figure 1.   Schematic representation of  the Raman scattering process (a)
           and the infrared absorption  process (b).  Initial,  inter-
           mediate, and final states are  labeled i, m, and f,  respectively.
                                  99

-------
(H3O ) exchange if the pH is low (1).  In the solid phase, the
linewidths generally lie between those of the gaseous and
liquid phases because of the momentum selection rule (K a; 0)
imposed by the periodicity of the crystalline structures.  The
major contribution to the phonon linewidth (0.5-10 cm  ) is due
to anharmonic lattice effects which can cause a phonon to decay
into different vibrational modes.  In principle, the phase of
the molecular scatterer can be deduced not only from the precise
energy of the vibrational state but also from the linewidths
of the infrared and Raman peaks.

     There are three significant differences between infrared
and Raman spectroscopy:  (1)  Since Raman scattering is consider-
ably weaker than infrared absorption, higher power laser sources
are required for the Raman technique as compared with those
needed for infrared absorption.  (2) The Raman approach is better
suited for point-sampling (cylindrical focal volume of 100 jum x
1 cm), while the infrared approach can average over a much longer
path length.  (3) Using one laser source, the Raman scattered
radiation contains chemical speciation and concentration infor-
mation on all the molecules present  in the cylindrical focal
volume, whereas high resolution infrared differential absorption
requires a tunable laser source for each type of molecule.  With
regard to these differences and the difficulties inherent in
using optical techniques to monitor emissions inside a stack or
downstream outside the stack, the infrared and Raman techniques
have complementary attributes.  Which technique is more viable,
particularly in situations in which the simultaneous monitoring
of several emissions is desirable, remains unclear.

      In this paper we shall review what is known about the Raman
effect in sulfur-containing molecules in their gaseous, liquid,
and aerosol forms.  Although much is already known, much important
information is still missing.  It is, therefore, not yet possible
to estimate the sensitivity or the degree of specificity of the
Raman technique when it is applied to stack emission analysis
where gaseous H2SO4f SO3, SO2, and H2S can coexist with liquid
H2S04 and H2S03, as well as with solid metal sulfates/sulfites,
(NH4) SO4, and (NH4)HSO4 in the presence of other atmospheric
gases and aerosols.


GASEOUS PHASE

      The Raman shifts, Au = v{ - vs (in cm" units), and the relative
Raman cross sections of common atmospheric gases are listed in
Table 1.  For these gases, using standard laboratory Raman  systems,
                                100

-------
Molecules
   SO,
             Table 1.   Raman Shifts and Cross Sections
                       for Common Atmospheric Gases
            Raman shifts in cm
Numbers in [ ] denote relative cross sections
1330
1380
1069
1065
652
532
536
Ref,

 a
 b
S02
H2S
N2
02
03
H2
CO
CO2
NO
N20
NH3
H20
CH4
1151
2611
2331
1556
1103
4160
2145
1388
1877
2223
3334
3652
3020
[5.5] 519 [0.11]
[6.6]
[1]
[1.2]
[4.0]
[2.2]
[0.9]
[1.5] 1286 [1.0]
[0.55]
[0.53] 1287 [2.7]
[3.1]
[2.5]
[0.79] 2914 [8.0]
c,d
c
c
c
e
e
c
c
e
e
f
g
e,f
References

a.  Wright, M. L.,  and K. S. Krishman.  EPA-R2-73-219,  1973.
b.  Tang, Sheng-Yuh, and Chris W. Brown.  J. Raman Spectry, 3:387,
    1975.
c.  Fouche, D. G.,  and R. K. Chang.  Appl. Phys. Lett., 18:579, 1971,
d.  Inaba, H.  Topics in Applied Physics, Vol. 14.  Laser Moni-
    toring of the Atmosphere.  E. D. Hinkley, Springer-Verlag,
    Berlin, Heidelberg, 1976. p. 153.
e.  Fouche, D. G.,  and R. K. Chang.  Appl. Phys. Lett,  20:256, 1972.
f.  Murphy, W. F.,  W. Holzer, and H. J. Bernstein.  Appl. Spectry,
    23:211, 1969.
g.  C. M. Penney, L. M. Goldman, and M. Lapp. Nature, 235:110, 1972.
                                101

-------
there is no difficulty in uniquely identifying and measuring trace
molecular concentrations of less than one part-per-million (ppm) .
If the absolute value of the differential cross section for
Raman scattering (do/aS2) is known, the detectable amount of
scattered light intensity can be easily calculated, as follows:
               Ws =  Off/a^aflcTl/Wj,                          [1]

where
               =  the differential Raman cross section in units of
                  cm2/sr.  For N2 gas molecules measured with
                  514.5 nm laser excitation, the value is 4.4 x
                  10~31 cm /sr (2).

            d$l =  the solid angle of the collection optics in
                  units of sr.  For an f/1 lens, the  solid angle
                  is 0.67 sr.

            c  =  the gas concentration in units of number of
                  molecules per cc.

            i\  =  the overall efficiency of the Raman spectrometer,
                  including lens reflection losses, spectrometer
                  throughput, and the quantum efficiency of  the
                  photodetector.  Typical values are  0.5 for lens
                  losses, 0.1 for spectrometer throughput, and
                  0.15 for detector quantum efficiency.  Thus, q
                  is approximately 8 x 10~ .

             / =  the focal length of the incident laser beam in
                  units of cm.  The typical length for f/1 collec-
                  tion optics and an f/8 spectrometer of 2 cm
                  slit height is 0.25 cm.

       W. , Ws  =  the incident and scattered power, respectively,
         1         in units of Watts.  Conversion of Watts into
                  photons per second requires multiplication by
                  6.3 x 10 18 (hvs)~1 , where hus is the scattered
                  photon energy in eV.

     If  the incident photon energy is below the energies of  any
excited  electronic levels of the sample, the Raman scattering
cross section increases proportionally as (hvs)4.  When the
(hus)4 law is obeyed, the inelastic scattering process is known
as the ordinary Raman effect (ORE).  In contrast, the resonance
Raman effect (RRE) , resonance fluorescence (RF) , and  broad
                               102

-------
fluorescence (BG) processes depicted in Figure 2 offer means of
enhancing inelastic light intensity.  For example, the resonant
Raman scattering cross section of I2 molecules is known to be 106
times greater than the ordinary Raman cross section (3).  The
resonant Raman cross sections for O3, OH~,  and SO2 have also been
reported (4)(5)(6).  Rousseau et al. (7) recently made explicit
the distinction between RRE and RF in I2 molecules by observing
the variations in the time decay of inelastic emission as the laser
photon energy was tuned through an absorption peak.  Fouche et al.
(3) further clarified the distinction between RRE and RF in I2
molecules by measuring the pressure quenching coefficient of the
inelastic emission as the photon energy was tuned through resonance,
RF and BF emission have been observed (8) for NO2.  The inelastic
emission cross section was 1C  times greater than the Raman cross
section of N2 at 514.5 nm.

     In spite of the obvious gain in the scattering cross section
that can be achieved with RRE, RF, and BF,  utilization of these
processes sacrifices a principal attribute of the ordinary Raman
techniquethat is, with ORE one monochromatic incident laser
source can provide chemical species and concentration information
on all of the molecular constituents contained within the focal
volume.  With RRE, RF, and BF, different laser wavelengths must
be used to achieve resonance for each type of molecule.  Thus
similar to the infrared absorption approach, a tunable source is
required.  Coherent anti-Stokes Raman spectroscopy (CARS) also
requires specific adjustment of source wavelengths, depending
upon the species under investigation.  While the CARS technique
may be preferable for determining the major component of a gas
mixture, ORE gives greater sensitivity in detecting the minor
components (9).

     Several approaches have been devised to increase the ORE
signal from a gas.  Hill et al. (10) discussed construction of a
multipass light trap employing an ellipsoidal mirror and a flat
spherical mirror assembly which, when used to study Raman
scattered light from atmospheric N2, provided an experimental
gain of 93 relative to the scattering obtained with one beam.  A
simplified configuration consisting of two spherical mirrors
positioned so that their common radii of curvature were coincident
was used by Stafford et al. (11) to obtain an effective W{ 49
times greater than the single pass laser power.   In the latter
system, a third concave mirror was used to redirect Raman
scattered photons that wculd normally be IOF^ back toward the
collection optics.  In a single pass scheme with 340 mW of power
at the 488 nm line of an argon laser and a double spectrometer
of 1 mm slit width and 10 mm slit height, the Raman signal from
atmospheric N2 was found (11) to be 5700 counts per second.
                               103

-------
                    (b)
              0>
^*
0),



V


0
1

-I
1


L

>
M^^B
cu
'

                                        -F*
                                                     m,  E
          m
(jj
                                                    f
                                           Q_HEf
       (ORE)   (RRE)    (RF)       (BF)
Figure 2.  The possible ways in which a laser photon may be scattered
         inelastically by a molecule.  Ordinary Raman effect (ORE) is
         shown in  (a). Resonance Raman effect (RRE) and resonance
         fluorescence (RF) are depicted in (b) and (c). Processes
         (a)-(c) all give rise to discrete energies for the scattered
         emission  (ho>).  Broad fluorescence emission (BF) at (ho>p) is
         shown in  (d) The wavy arrows indicate nonradiative transi-
         tions caused by collisions (8).
                              104

-------
Including the multipass light trap and the third mirror, the
atmospheric N2 signal increased to 5.22 x 105 counts per second,
a gain of approximately 92.  Use of larger mirrors in this con-
figuration can be expected to further increase ORE signal/noise
ratios.

     It should be noted from Table 1 that the cross section of
SOa is unknown and that the Raman shift and cross section of
H2S04  (gaseous) are yet to be determined.


LIQUID STATE

     The Raman effect also can be exploited in investigating the
constitution of acid solutions at varying concentrations (pure
liquid acid to aqueous solutions of acids).  Changes in the
molecular configuration or constitution of the acid caused by
dilution or concentration will cause corresponding changes in
the Raman shifts, linewidths, and intensities.  Figure 3 shows
the distribution of Raman shifts from pure acid H2S04 at varying
concentrations (12).  The shifting of some of the principal
Raman peaks in H2S04 versus concentration (12) is shown in Figure 4.
The absolute cross section of pure acid H2S04 is unknown.

     By dissolving S03 in 100% sulfuric acid, the characteristic
lines of H2S04 gradually disappear and new lines develop.  This
effect has been attributed (12) to the formation of H2S207>
S206, or S309.  The Raman shifts from fuming sulfuric and sulfuric
acids are summarized in Table 2.

  _  Sulfuric acid in the dilute state dissociates into HSO^ and
S04.  The corresponding spectra (11) for  1M H2S04, 1M (NH4)HS04,
and 1M (NH4)2S04 are_shown in Figure 5.   Note that (NH4)HS04 also
dissociates into HS04 and S04 .  The absolute Raman cross section
for SCu in the aqueous state is 6.3 x 1030cm2/sr with 514.5 nm
excitation, which is 14 times the_N2 cross section.  The broad
linewidth associated with the HS04 and S04 peaks is due mainly
to proton transfer (1).  Raman spectra for KHS04 and NaHS04 dis-
solved in water are shown in Figure 6 along with those  for H2S04
and (NH4)HS04.  Once these molecules are  dissociated, their
Raman spectra are insensitive to the cations.  The results of
Stafford et al. (11) established that the vibrational frequency
of SO4 is the same for all solutions because_of its independence
on cations when freely dissociated.  The  HS04 anion has two vi-
brational frequencies near the 981 cm 1 line of S04.  The first
is at 895 cm"1  , which is very near the 910 cm 1 line of an
undissociated H2S04 molecule (13).  However, the H2S04  contribution
                                105

-------
       A i/
      1050
      1040
      1030
       980
       900
       890
       590
       580
       430
       420
       410
I   I    I   I   I   I    I   T
              i    i   i   i   i
                   I	I
              10    30    50    70
                      H?S04
Figure 3. The change in the principal Raman shifts A,, in units of cm
       for various concentrations of H_SO. (13).
                            24
                       106

-------
                                         O       <&       *^i       i^n       ho
                              _          O       O       ui       O       t_n

Common    H.SO,    HSoT    S0~          **       ^       ^       ^       ^
           f-  4        4      4
                     895	

 Figure 4.  The distribution  of Raman shifts (cm  ) from H SO, at varying

            concentrations  (12).

-------
              Table 2.  Raman Shifts from Fuming
                      Sulfuric and Sulfuric Acids*
                        Raman shifts in cm

     Fuming sulfuric acid
                                          -1
Sulfuric acid
(Free S03
68%
245
295
330

484
536

688
737

957
1075

1252
1450
51%
246
297
331

480
535

688
740

960
1075

1250
1450
given in percent)
36%

300
325

480
525
565

735
904
957

1160
1250
1440
23%

305
327
390
427

565

740
915
975

1160
1240
1420
10%

315

388
427

562

735
915
970

1140

1393
100%



395
430

560


917
972

1140

1370
90%




412

566


915

1045
1148


80% 70%




414 424

580 583


907 902

1040 1040



Hibben, James H.  The Raman Effect and Its Chemical Applications,
Reinhold Publishing Corp., New York, 1939. p. 364.
                              108

-------
            CD
            cr
            c/5

            LU
            h-
            <
            s
            <
                 SLIT WIDTH


                   2.8CM-'
                                             NH4HS04
                                            (NHLLSQ
                            4/2-^4
                               1000       900
1100        1000        900       800

       RAMAN SHIFT (CM-')
Figure 5.  Raman spectra of 1M solutions containing  sulfate and bisulfate

          anions.   The  (NH.)2S04 ordinate has been  decreased by a factor


          0.2.  Spectra are restricted to the region about v,, vibration

          ~C 4-Uxt fi-nr. CA  ^TIT^YI -1- QQ1 r*m~l ^^l^
          of the free
      anion at 981 cm 1(11) .
                                  109

-------
                        H2S04
                        KHSO,
                        NaHSCX
                       NH4 HS04
                                    RAMAN
 111111	I I I I I	
1200      1000      800
                   CM"1
                                  600
400
Figure 6.  Raman spectra of 2M H2SO^, KHSO^, NaHSO,, and NH,HSO, solutions (1),

-------
is negligible for 1M H2S04 and is practically nonexistent in
1M NH4HS04.   The second_frequency is_ooncentration dependent
and varies from 1055 cm   to 1030 cm   as the concentration of
HS04  increases_(1)(13) .  These lines are also much broader
than the 981 cm 1 line and overlap to form a complex spectrum.
Unfolding this complex spectrum to accurately determine do/dfi
for HS4   requires making assumptions about the lineshapes.
Zarakhani (14) has attempted this approach with H2S04 and
attained a ratio of S0~ to HG04 integrated strength of approxi-
mately 2/1 for the 981 cm"1 and 1040 cm 1  lines.  Tnis ratio is
compatible with a simple deconvolution of the spectra in Figure
5 using the 1m dissociation constant of HSO^ (1) (15).  The
absolute cross section of HS04 is therefore approximately 1/2
that of 80 = .

     It would be desirable to obtain an integrated Raman intensity
that is linear with total sulfate and bisulfate concentration.
However,  it has been shown that respective concentrations of
804" and HS04 are not linear with the total concentration of_H2S04
(16).  Since the cross sections of the 1040 cm~1 and 981 cm~1
lines are not equal, an integrated Raman  intensity of the two would
not be linear with  total concentration.   This causes a problem
since the two lines must be integrated.   The reason for this is
that the  concentrations of sulfate pollutants in the atmosphere
are sufficiently small that the Raman signal would be too weak
to obtain an accurate  spectrum that would allow an unfolding of
the two overlapping lines.  However, because of this latter point,
the 895 cm~1 line of HS04 must also be included in the integrated
intensity, since it strongly overlaps with the 981 cm"1 line
of SO^.  By taking  the sum of the integrated intensities of these
three lines (1040 cm"1, 981 cm~1 , 895 cm~1), the following
is found:  for one molar concentrations of (NH4)2S04, NH4HS04,
or H2S04, the Raman spectra of which are  shown  in Figure 5, the
sums, normalized to (NH4)2S04, are 1, 0.98, and 1.03, respec-
tively.  These sums are approximately equal within the experimental
accuracy of +10%.   If  one takes the deconvoluted integrated
intensities of Zarakhani et al. (14) for  H2SO4 and also performs
the same sum, for molar concentrations between 0.4 and 5M, the
total sum is approximately linear with total concentration.
For H2SO4, Stafford et al. (11) checked and extended this approxi-
mate linearity from 5M down to 0.01M for  total  integration
without deconvolution.  Over this concentration range, the peak
height ratio of the Raman intensities of  the 981 cm"1 and 1040
cm~1 line reverses, indicating the complexity of the superimposed
spectra.  A 1M solution of NaHS04 was also analyzed with results
confirming those of NH4HS04 and H2S04.  Thus, by using a
spectral slit width that encompasses the  lines  1040, 981, and
                               111

-------
895 cm  ,  the resulting integrated Raman intensity will be approxi-
mately linear with total sulfate and bisulfate concentration in
aqueous solutions over the range 0.01-5M.  The absolute concentra-
tion can then be determined from the integrated Raman intensity and
the absolute differential cross section of SO4 given above.


LIQUID MIST FORM

     Should the sulfur-containing liquid be in mist form, then it
is important to know the size distribution of the droplets.  When
the droplet size is comparable to the wavelength of the infrared
light, elastic Mie scattering and the electric field inside dis-
torted by the droplet boundary can affect the infrared absorption
coefficient.  Similarly, when the droplet size is comparable to
the excitation wavelength in the visible region, the Raman scatter-
ing intensity will be affected by the droplet size, shape, and
refractive index (17)(18)(19).  However, the Raman shifts will not
be influenced by the droplet character of the scatterer.

     The Raman signature of (NH4)2S04 mist (size less than 2 /urn)
flowing past the cylindrical focal volume of the incident laser
has been measured recently (2Q) .  As indicated in Figure 7, eight
parts-per-billion (ppb) of SO4 were detected with a counting period
of 1000 second using the experimental arrangement shown in Figure 8.
The (NH4)2S04 must have been dissociated totally as_the
observed Raman shift coincides with that of free SO4 ions in the
liquid phase (981 cm"1) and not with that in the solid phase (976
cm"1).  Significantly, the Raman_intensity seems to indicate that
the. relative cross section of SO4 is 20 times larger than that of
S04 dissolved in aqueous solution.  This result implies a large
enhancement caused by the droplet size, shape, and index of refrac-
tion.  Further confirmation of this enhancement is desirable.

     The Raman shifts from other common ions are listed in Table
3.  However, the cross sections other than for S04 are not known.


SOLID PHASE

     Sulfates in the solid phase are numerous.  Table 4 summarizes
representative species along with reported Raman shifts of sulfates
in the solid form.  The corresponding shifts (21) (in cm 1) for free
S04 are u, = 981, *>2 = 451,  v3 = 1104, u4 = 613.  Raman cross sec-
tions for bulk sulfate crystals are unknown.   However, Blomer et
al. (22) and Wright et al. (23) have reported some preliminary
results on the absolute cross sections of sulfates in powder form.
                               112

-------
         10
          -2
    or
    CD
    V)
         !0'
    C   10

    ~Z.
    Ld
    CD
    O
    QL
    \-
    LU
          '4
         I0
           5
                1  1  i 1 1 1 1
                             1   1 1 1  M M
10
                                      100
                          IOOO
                                                              10.000
                  SULFATE  CONCENTRATION (ppb)
Figure 7.  Graph of the sulfate Raman signal versus sulfate concentration,
          normalized to the Raman signal of ambient N2 .  The accuracy in
          the sulfate concentration calibration measurement was + 50% (20),
                                  113

-------
         CW  LASER-RAMAN  SPECTROMETER
                              DOUBLE
                          MONOCHROMATOR
                                                   AEROSOL
                                                  GENERATOR
                                                  PHOTON
                                                 COUNTER
                                                           DISPLAY
Figure 8.  Schematic of a laser-Raman scattering experiment used for
         the detection of laboratory generated sulfate aerosols.  The
         Raman signal is enhanced by a multipass optical assembly (20).

-------
         Table 3.   Raman Shifts from Various Ions*
Ions
S0*~ 1104
C10*~ 1102
PO^~ 1082
Nfll" 3134
4
CO*' 1415
N0~ 1390
o
Raman shifts in cm
981 613 451
935 628 462
935 515 363
3033 1685 1397
1063 680
1050 720
Wright, M. L., and K. S. Krishnan.  EPA-R2-73-219, 1973,
                          115

-------
Table 4.  Raman Shifts of Sulfates in the Solid Phase
Cation

L12
LiK
Na2
KH
K2
(NH4)2
Rb2
Cs2
Be2
Cu2
Mg2
Ca2
(Anhydrite)
Ca2
(Gypsum)
Sr2
Ba2
Zn2
Mean S04 frequencies, cm
Vl
1017
997
1010
995
990
980
985
986
978
977
976
976
976
972
1061
990
983
1029
985
1015
1013
1015
1005
999
999
989
989
1051
980
V2
518
440
-
459
460
427
451
454
449
453
450
454
453
442
528
450
448
456
456
415
425
415
456
460
458
451
450
U3
1118
1104
1110
1110
1115
-
1119
1091
1108
1090
1073
1066
1193
1108
1108
1108
1132
1108
1150
1113
1140
1106
1126
1108
\
722
595
-
637
615
593
624
617
621
616
613
600
606
850
620
612
613
638
609
620
620
637
628
633
625
620
Ref .

a
b
b
a
b
b
a
b
a
b
c
a
b
a
a
b
b
a
b
a
b
c
b
a
b
a
b
a
b
                      116

-------
                          Table 4  (continued)
Cation
,
Hg2
A/,
KAf
Tj>2
Pb2
Mn2
Fe2
FeNH4
Ni2
Mean S04 Frequencies, cm '
1027
1010
1000
1014
986
987
981
1000
984
963
978
978
990
978
992
980
462
497
499
499
495
457
480
461
-
445
443
448
-
-
460
1108 621
1050 658
1125 660
612
620
600
-
1105 624
668
1065 608
606
673
-
620
Ref .
a
b
c
a
b
c
b
c
b
d
a
b
c
b
b
b
b
References
a.  Ananthanarayanan, V. Indian J. Pure Appl.  Phys.,  1:58,  1963.

b.  Hibben, J. H.  The Raman Effect and Its Chemical  Applications.
    Reinhold, New York, 1939.

c.  Wright, M. L.,  and K. S. Krishnan.  EPA-RE-7^-219,  1973.

d.  Blomer, F., and H. Moser. Z. Angew. Phys.  27:302, 1969.
                                117

-------
To obtain more precise cross section values, the effect of size,
shape, and index of refraction of sulfate particulates should be
investigated further.

     Particulate matter is present (24) in the atmosphere at a
level of about 100 Mg/m3.  Sulfate particulates represent approxi-
mately 10% (5 Mg/m3-15 Mg/m3).  A quantity of 1 Mg/m3 represents
0.25 ppb of SO" and 0.38 ppb of S02.  Thus, analytical techniques
capable of detecting ambient concentrations of sulfur-containing
molecules require sensitivities to levels of less than 1 ppb.  Par-
ticle sizes range from 0.1 Mm to several microns in diameter.

     Raman scattering techniques are capable of detecting sulfur
compounds in the low ppb range.  For example, as noted above,
Stafford et al. (20) detected 8 ppb of 804 in a mist of ammonium
sulfate flowing past the laser focal volume.  Rosasco et al. (25)
have developed a Raman microprobe technique to study individual
micron size particles positioned on a substrate and have collected
spectra from small sulfate salt particles.  A principal result of
this work is that the Raman shifts from a micron size particle do
not differ from those of the bulk material.  In addition, Rosen
and Novakov (26) have identified (NH4)2SO4 as a major constituent
of filter collected ambient and source-enriched aerosol particu-
lates by observing the Raman spectra from the particulate laden
filter.

     An attribute of the Raman method is its ability to characterize
the chemical constituents of an aerosol as it flows past the laser
focal volume without prior filter collection which might alter the
chemical nature of the species being identified.  However, using
optical techniques to probe micron-size particulates dispersed in
air requires consideration of the spatial distribution of excita-
tion electric field intensity within the particle, as well as the
angular profile of emitted radiation.  This information is essen-
tial if molecular concentrations are to be inferred from measured
Raman signal intensities and bulk scattering cross sections.  The
angular distribution of elastic (Lorenz-Mie) scattering from mono-
dispersed spherical particles having diameters comparable to the
incident light wavelength has undergone extensive investigation both
theoretically and experimentally (27)(28).  Furthermore, Chew et
al. (17) have recently presented a theoretical model for inelas-
tic (Raman and fluorescent) scattering by molecules embedded in
small particles.  Their treatment describes the angular distribu-
tion of radiation from a dipole located at an arbitrary position
within a dielectric sphere of given refractive index after excita-
tion by a plane wave and can be extended easily to a large number
of such dipoles dispersed within the sphere.  Experimental deter-
                                118

-------
minations of the angular distribution of fluorescence intensity from
monodispersed spherical particles having dye molecules dispersed
throughout the sphere have also recently been reported (18)(19).
Figure 9 summarizes results obtained with 0.81 Mm diameter poly-
styrene particles and 0.488 M*n laser excitation  (19).  The value of
m = 1.195 is defined as the ratio of the particle's refractive
index to that of the surrounding medium and provides a measure of
optical index mismatch.  In Figure 9(a), the observed elastic Mie
scattering structure, which can be related to the sphere size, is
shown along with the angular profile of fluorescence measured at a
wavelength of 0.525 nm.  The letters V-V and H-H refer to the inci-
dent and scattered light being polarized perpendicular and parallel
to the observation plane, respectively.  The interference structure
observed in the elastically scattered light is completely absent
in the fluorescence profile.  This observation is consistent with
the fact tha,t Mie scattering requires a coherent sum of polariza-
tions within the particle, while fluorescence entails an incoherent
sum since the phase associated with each emission center is random
in time.  Figure 9(b) represents the difference  in fluorescence
intensity from dye molecules contained within 0.81 Mm diameter
particles relative to dye molecules dispersed in a liquid for the
two polarization cases.  The plotted quantity, a percentage devia-
tion, A , was obtained by first normalizing the particle and liquid
case intensities to be equal at an observation angle of 10.  The
percentage deviation was then defined as the difference betv/een the
normalized particle and liquid fluorescence intensity values divided
by the liquid intensity and multiplied by 100 to obtain a percen-
tage.  The observed deviations were largest for  the backward direc-
tion (near 180) and at 90 for H-H polarization.

     At 168, A for both V-V and H-H polarization approached 0% as
the refractive index ratio, m = np/n/>, between the particle (np =
1.59) and the surrounding fluid was made nearly  equal to one by
adding glycerol (nn = 1.476) to water (na = 1.33).  It can be con-
cluded that A near the backward direction is a result of distor-
tion of the planes of constant phase and electric field amplitude
within the particle caused by the particle-medium index of refrac-
tion mismatch.  This result, when applied to aerosols for which
no = 1, predicts a large signal enhancement in backscattering, and
the conversion of fluorescent LIDAR data (29) to molecular concen-
trations within aerosols must account for the particle-air index
mismatch, as well as the size and shape of the particle.  The
observation by Stafford et al. (20) that the apparent cross section
of SC>4 in liquid mist  fo^m is approximately 20 times greater  than
its bulk value is likely to be a manifestation of this effect.
                                119'

-------
   10
   10
   io-]
  50%

<
Z
i 0%
LU
O
 -50%
          (a)
i    I   1   I    I   I    I   I    I   !   I    I   I    r
                                 d = 0.81/im
 ELASTIC SCATTERING
                                     V
                       FLUORESCENCE
                                                         V-V
i   i    i   i    i   i    i
                                      i   i    i   i
                                      i   i    t
       -  (b)
               = 0.5250.015/im
                                        V-V
             
              i
                                            ../*
                                      (




                            /"  H-H
                30e
      60       90       120
   ANGLE OF OBSERVATION 6
                                                         150*
                                                         180*
                                                   OBS
Figure 9.  (a)  Angular distribution of elastic scattering (top)  and
          fluorescence (bottom) from 0.81 ^ra diameter particles with
          0.488 jum excitation.  (b) Deviation, A,  of 0.81 /urn particle
          fluorescence from dye solution fluorescence.  The inter-
          ference filter used to isolate the fluorescence wavelength,
          A,,  was centered at 0.525 /um and has a half-width of  0.015
          jttm.  The incident wavelength is A. (19).
                                        i
                                 120

-------
     The large 
-------
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
(>
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 130C and have low capacity for particulate loading.
These filters are used for collecting extractive samples where the
temperature can be maintained below 130C 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
125C.  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 100C) 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 (>250C)
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 ISO3

strongly favors the formation of SO3 at  temperatures below about
540C (1000F).  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 204C  (400F), 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 21C  (70F),  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 5C (9F) 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 135C (275F), and condensate composition
at that point would be about 79 wt% H2S04.  If the gas were
cooled to 121C (250F), 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 121C (250F), 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 135C (275F) to
75 vol% at 121C (250F).

     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 17C (30F) to 22C (40F) 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 1C (2F) to 2C (4F) 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 38C (100F) with no sulfur, an increase to
         127C (260F) (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 205C  (400F) and 138C
                                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

-------
(280F) and found that filter temperature significantly affected
the amount of H2SO4 found on the filter.  At 138C (280F),
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 205C (400F), 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 120C (250F) and 150C (300F).  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 [120C (248F), 205C
(401F)], 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 170C (338F).  The  results in Table
3 show at 120C (248F), below the H2S04 dewpoint, most of the
H2S04 was found in the coil and on the test filter.  At 205C
(401F), 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 120C (248F) sampling temperature (or at a temperature
below the H2S04 dewpoint in the stack gas), but only for about
9% at a 205C (401F) 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)
                             120Ca              205C
                           % 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  170C
      Average of two trials


that this contribution of particulate sulfate is greater than
the "dry" particulate sulfate that exists at "pre-collection"
conditionsif 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 205C was less
     than 1 mg/m3, comparable to  sulfate blank analyses for
     these filters.  However, exposures at  205C  with sulfuric
     acid vapor at 10 ppm to 40  ppm and moisture  contents
     up to 25 volume percent (acid moisture dewpoint of 120-
     150C) 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 200C 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 (20C, 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 20C
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 150C (300F) for these sources, to 120C
(250F), 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 - 120C (250F), typical
          collection train temperature, and 205C (400F), 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
120C
X X
X X
X X
X X
Iter Temperature
205C
X X
X X
X X
X X
x - Simultaneous exposure of MSA1106BH and SFJ Thimble filters
B.  Additional Evaluations  (Gas Stream/Filter, Temperature = 205C)

                                  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 - 205C, 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
                 = 205C, 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  = 205C,
     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 205C

     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 500C 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
120C
0.7, 0.9
1.1, 1.4
1.0, 0.9
1.8, 1.4
Gas Stream/Filter
120C
0.6, 0.2
0.6, 0.5
0.3, 0.2
0.5, 0.7
- mg/m3
Temperature
205C
0.8, 0.8
1.7, 1.0
1.3, 1.3
1.7, 2.3
Temperature
205C
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 = 205C, 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 = 205C, 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 200C 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 200C, well above
                                198

-------
             Table 9.  Results of Phase 4 Sulfur Dioxide -
              Sulfuric Acid Mist Challenge Studies (Gas
              Stream/Filter Temperature = 205C, 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 135C.  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 200C (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 60C (140F).  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 220C (428F).  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 (220C [438F], 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 (220C, 428F) .  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 260C (500F).
                                    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 800C          150C
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 400F 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 1200C 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 300F 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 20C 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) vhen 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 350C.  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 350C 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) 350C;  (B) 120C; (C) 80C.
                                260

-------
     (3)  Obtain infrared spectra of the water solubles and
          the water insolubles before and after heating in
          argon at 350C 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 350C.  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 35CC  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  350C  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

-------